Difference between revisions of "John's Linux page"

From ProgClub
Jump to: navigation, search
(34 intermediate revisions by the same user not shown)
Line 13: Line 13:
 
  $ lsb_release
 
  $ lsb_release
  
== Determining which Unix you are running ==
+
== Determining which Linux/Unix you are running ==
  
 
  $ uname
 
  $ uname
 +
 +
Or,
 +
 +
$ uname -mrs
  
 
Or,
 
Or,
Line 70: Line 74:
 
  RAID:      Devices: 1: /dev/md1 2: /dev/md0
 
  RAID:      Devices: 1: /dev/md1 2: /dev/md0
 
  Info:      Processes: 355 Uptime: 11 days Memory: 21198.3/32043.3MB Client: Shell (bash) inxi: 2.3.5
 
  Info:      Processes: 355 Uptime: 11 days Memory: 21198.3/32043.3MB Client: Shell (bash) inxi: 2.3.5
 +
 +
= Power =
 +
 +
== Reporting on PowerShield DEFENDER UPS status ==
 +
 +
To see the status of the [https://powershield.com.au/powersheild_product/defender/ PowerShield DEFENDER] systems on John's LAN:
 +
 +
$ upsc defender
 +
 +
E.g.:
 +
 +
jj5@orac:~$ upsc defender
 +
Init SSL without certificate database
 +
battery.charge: 100
 +
battery.voltage: 27.40
 +
battery.voltage.high: 26.00
 +
battery.voltage.low: 20.80
 +
battery.voltage.nominal: 24.0
 +
device.type: ups
 +
driver.name: blazer_usb
 +
driver.parameter.pollinterval: 2
 +
driver.parameter.port: auto
 +
driver.parameter.synchronous: no
 +
driver.version: 2.7.4
 +
driver.version.internal: 0.12
 +
input.current.nominal: 5.0
 +
input.frequency: 50.1
 +
input.frequency.nominal: 50
 +
input.voltage: 242.6
 +
input.voltage.fault: 242.6
 +
input.voltage.nominal: 240
 +
output.voltage: 242.6
 +
ups.beeper.status: disabled
 +
ups.delay.shutdown: 30
 +
ups.delay.start: 180
 +
ups.load: 14
 +
ups.productid: 5161
 +
ups.status: OL
 +
ups.type: offline / line interactive
 +
ups.vendorid: 0665
 +
 +
== Run commands on PowerShield DEFENDER UPS batteries ==
 +
 +
You can run "instant commands" using the '''upscmd''' command.
 +
 +
We use the 'beeper.toggle' instant command in our Salt Stack config to disable the beeper, see e.g.:
 +
 +
diligence:/srv/salt/conf/app/defender-1200.sls
 +
 +
To see "instant commands" supported by the PowerShield DEFENDER:
 +
 +
$ upscmd -l defender
 +
 +
E.g.:
 +
 +
jj5@orac:~$ upscmd -l defender
 +
Instant commands supported on UPS [defender]:
 +
 +
beeper.toggle - Toggle the UPS beeper
 +
load.off - Turn off the load immediately
 +
load.on - Turn on the load immediately
 +
shutdown.return - Turn off the load and return when power is back
 +
shutdown.stayoff - Turn off the load and remain off
 +
shutdown.stop - Stop a shutdown in progress
 +
test.battery.start - Start a battery test
 +
test.battery.start.deep - Start a deep battery test
 +
test.battery.start.quick - Start a quick battery test
 +
test.battery.stop - Stop the battery test
  
 
= Environment =
 
= Environment =
Line 168: Line 240:
  
 
  $ sudo su -c "svn update" www-data
 
  $ sudo su -c "svn update" www-data
 +
 +
== Reporting user and group info for the current user ==
 +
 +
$ id
  
 
= Memory management =
 
= Memory management =
Line 176: Line 252:
  
 
  $ free -m
 
  $ free -m
 +
 +
== Check for swap thrashing ==
 +
 +
Check your virtual memory status with vmstat:
 +
 +
$ vmstat
  
 
= Video/display management =
 
= Video/display management =
Line 273: Line 355:
 
  # watch zpool iostat -v
 
  # watch zpool iostat -v
 
  # zpool iostat -v 2
 
  # zpool iostat -v 2
  # watch 'zfs list; echo; zpool list'
+
  # watch 'zpool list; echo; zfs list'
 
  # watch zfs get compressratio -o all
 
  # watch zfs get compressratio -o all
 
  # watch cat /proc/spl/kstat/zfs/arcstats
 
  # watch cat /proc/spl/kstat/zfs/arcstats
Line 297: Line 379:
 
  # perf report --stdio
 
  # perf report --stdio
  
= Monitoring disk I/O =
+
You can search for ZFS files like e.g. this:
  
There's an app for that! iotop.
+
root@orac:/# find / -name '*zfs*' -or -name '*zpool*'
  
== Using iotop, top for disks ==
+
You can report history of a zpool:
  
  # iotop -oPa
+
  # zpool history $poolname
  
= File management =
+
You can get a report on the dedup tables:
 +
 
 +
# zpool status -D $poolname
  
== Listing only directories ==
+
Or more detailed dedup table info:
  
  $ ls -l | egrep '^d'
+
  # zdb -DDD $poolname
  
== Listing only files ==
+
Note in the output see [https://unix.stackexchange.com/a/405700 here] for details, basically:
  
  $ ls -l | egrep -v '^d'
+
{|class="wikitable"
 +
! Abbr  !! Description
 +
|-
 +
| LSIZE || logical size (in memory)
 +
|-
 +
| PSIZE  || physical size
 +
|-
 +
| DSIZE  || size on disk
 +
|-
 +
| refcnt || reference count
 +
|}
  
== Listing hidden files ==
+
== Measure data throughput ==
  
$ ls -al .[!.]*
+
Use the 'pv' command from the 'pv' package, e.g.:
  
== Creating a symbolic link ==
+
# cat /dev/sda | pv | cat > /dev/null
  
$ ln -s /path/to/target link-name
+
Or for ZFS:
  
== Creating a hard-link ==
+
# zfs send data/example | pv | cat > /dev/null
  
$ ln /path/to/target file-name
+
== Using Smartctl, Smartd and Hddtemp on Debian ==
  
== Changing the owner of a file ==
+
For notes on using smartctl see [https://www.lisenet.com/2014/using-smartctl-smartd-and-hddtemp-on-debian/ Using Smartctl, Smartd and Hddtemp on Debian].
  
$ chown user:group <files>
+
== Report hard disk usage ==
  
E.g.
+
So you might want to know how much data a process reads or writes to a hard disk. You can monitor process total disk utilisation with the 'iotop' command. Run 'iotop' and then press 'a' for --accumulated.
  
$ chown jj5:staff README
+
== Report hard disk temperatures ==
$ chown root:root *
 
  
To apply recursively into sub-directories use -R,
+
E.g.
  
  $ chown -R root:root /etc/*
+
  # hddtemp /dev/sd[a-e]
  
== Changing file permissions ==
+
= Monitoring disk I/O =
  
{|class="wikitable"
+
There's an app for that! iotop.
|+ Object codes
+
 
! User !! Group !! Other
+
== Using iotop, top for disks ==
|-
+
 
| u    || g     || o
+
# iotop -oPa
|}
+
 
 +
== Monitor disk I/O for performance issues ==
 +
 
 +
# watch iostat
 +
 
 +
Or e.g.
 +
 
 +
# watch iostat -xd /dev/sd[abc]
  
{|class="wikitable"
+
= Monitoring a system =
|+ Permission codes
 
! Read !! Write !! Exectue
 
|-
 
| r    || w    || x
 
|-
 
| 4    || 2    || 1
 
|}
 
  
{|class="wikitable"
+
== Simple ZFS monitoring ==
|+ Numeric codes
 
! 0
 
| None
 
|-
 
! 1
 
| Execute
 
|-
 
! 2
 
| Write
 
|-
 
! 3
 
| Write, Execute
 
|-
 
! 4
 
| Read
 
|-
 
! 5
 
| Read, Execute
 
|-
 
! 6
 
| Read, Write
 
|-
 
! 7
 
| Read, Write, Execute
 
|}
 
  
See [http://catcode.com/teachmod/numeric2.html Numeric Mode in Action].
+
# watch iostat
 +
# iotop
 +
# zpool iostat -v 5
 +
# watch 'hddtemp /dev/sd[a-e]; echo; zpool list; echo; zfs list'
 +
# nethogs
 +
# top
  
$ chmod <user numeric code><group numeric code><other numeric code> <files>
+
= File management =
$ chmod <object codes>+|-<permission codes> <files>
 
  
E.g.
+
== Listing only directories ==
  
  $ chmod 600 my-private-file
+
  $ ls -l | egrep '^d'
$ chmod go-rwx my-private-file
 
$ chmod u+rw my-private-file
 
$ chmod +x my-script
 
  
== Updating config files ==
+
== Listing only files ==
  
If you get given a new config file called new.conf and you want to integrate it with your old config file old.conf then:
+
$ ls -l | egrep -v '^d'
  
$ cp old.conf updated.conf
+
== Listing hidden files ==
$ merge -A updated.conf new.conf old.conf
 
  
Then go through and edit updated.conf resolving all the merge errors, picking and choosing what to update and what to keep. When you're done copy updated.conf to old.conf so it becomes the new config file.
+
$ ls -al .[!.]*
  
The merge program is a part of the RCS package. If you don't have it:
+
== Creating a symbolic link ==
  
  $ sudo apt-get install rcs
+
  $ ln -s /path/to/target link-name
  
== Listing open files ==
+
== Creating a hard-link ==
  
Use lsof to list open files. E.g.:
+
$ ln /path/to/target file-name
  
# lsof
+
== Changing the owner of a file ==
  
See man lsof for options.
+
$ chown user:group <files>
  
== List permissions on a whole directory path ==
+
E.g.
  
E.g.:
+
$ chown jj5:staff README
 +
$ chown root:root *
  
$ namei -om /home/jj5/workspace
+
To apply recursively into sub-directories use -R,
  
Outputs:
+
$ chown -R root:root /etc/*
  
f: /home/jj5/workspace/
+
== Changing file permissions ==
  drwxr-xr-x root root /
 
  drwxr-xr-x root root home
 
  drwxr-xr-x jj5  jj5  jj5
 
  drwxr-xr-x jj5  jj5  workspace
 
  
== Counting non-blank lines in a file ==
+
{|class="wikitable"
 +
|+ Object codes
 +
! User !! Group !! Other
 +
|-
 +
| u    || g    || o
 +
|}
  
E.g.:
+
{|class="wikitable"
 +
|+ Permission codes
 +
! Read !! Write !! Exectue
 +
|-
 +
| r    || w    || x
 +
|-
 +
| 4    || 2    || 1
 +
|}
  
$ cat foo.c | sed '/^\s*$/d' | wc -l
+
{|class="wikitable"
 +
|+ Numeric codes
 +
! 0
 +
| None
 +
|-
 +
! 1
 +
| Execute
 +
|-
 +
! 2
 +
| Write
 +
|-
 +
! 3
 +
| Write, Execute
 +
|-
 +
! 4
 +
| Read
 +
|-
 +
! 5
 +
| Read, Execute
 +
|-
 +
! 6
 +
| Read, Write
 +
|-
 +
! 7
 +
| Read, Write, Execute
 +
|}
  
== Cloning one directory to another with rsync ==
+
See [http://catcode.com/teachmod/numeric2.html Numeric Mode in Action].
  
E.g.:
+
$ chmod <user numeric code><group numeric code><other numeric code> <files>
 +
$ chmod <object codes>+|-<permission codes> <files>
  
rsync --acls --xattrs --stats --human-readable --recursive --del --force --times --links --hard-links --executability --numeric-ids --owner --group --perms --sparse --compress-level=0 /data/source/ hostname:/data/target/
+
E.g.
  
= Symbolic-link management =
+
$ chmod 600 my-private-file
 +
$ chmod go-rwx my-private-file
 +
$ chmod u+rw my-private-file
 +
$ chmod +x my-script
  
== Data used by sym-linked files:
+
== Updating config files ==
  
This will de-reference the sym-links in the current directory and tell you how much data the files pointed to by the sym-links are using:
+
If you get given a new config file called new.conf and you want to integrate it with your old config file old.conf then:
  
  jj5@tact:/data/backup/unity/latest$ du -hD * | sort -h
+
  $ cp old.conf updated.conf
 +
$ merge -A updated.conf new.conf old.conf
  
= File searching =
+
Then go through and edit updated.conf resolving all the merge errors, picking and choosing what to update and what to keep. When you're done copy updated.conf to old.conf so it becomes the new config file.
  
== Finding a file with a particular name ==
+
The merge program is a part of the RCS package. If you don't have it:
  
  $ find -iname "*some-part-of-the-file-name*"
+
  $ sudo apt-get install rcs
  
Will start searching from the current directory, so maybe
+
== Listing open files ==
  
$ cd /
+
Use lsof to list open files. E.g.:
  
first. For a case-sensitive search:
+
# lsof
  
$ find -name "*eXaCT CaSE*"
+
See man lsof for options.
  
== Finding a file with particular content ==
+
== List permissions on a whole directory path ==
  
To search in /etc/ for a file with particular content:
+
E.g.:
  
  $ grep -R "search-string" /etc/*
+
  $ namei -om /home/jj5/workspace
  
To search the current directory for *.cs files containing the word "Up":
+
Outputs:
  
  $ find . -name '*.cs' -exec grep --color=auto -H Up {} \;
+
  f: /home/jj5/workspace/
 +
  drwxr-xr-x root root /
 +
  drwxr-xr-x root root home
 +
  drwxr-xr-x jj5  jj5  jj5
 +
  drwxr-xr-x jj5  jj5  workspace
  
== Finding a list of files with particular content ==
+
== Counting non-blank lines in a file ==
  
E.g. to find all the files with the word 'creativity':
+
E.g.:
  
  $ grep -R creativity . | sed 's/:/ /' | awk '{ print $1 }' | sort | uniq
+
  $ cat foo.c | sed '/^\s*$/d' | wc -l
  
== Using the locate command to find files ==
+
== Cloning one directory to another with rsync ==
  
$ locate part-of-filename
+
E.g.:
  
E.g.
+
rsync --acls --xattrs --stats --human-readable --recursive --del --force --times --links --hard-links --executability --numeric-ids --owner --group --perms --sparse --compress-level=0 /data/source/ hostname:/data/target/
  
$ locate texvc
+
== Counting number of files in current directory and all subdirectories ==
  
== Updating locate command's database ==
+
$ ls -AlhR . | egrep '^-' | wc -l
  
# updatedb
+
== Counting number of directories in current directory and all subdirectories ==
  
= Job control =
+
$ ls -AlhR . | egrep '^d' | wc -l
  
== Stopping a running process ==
+
= Compression =
  
Press Ctrl+Z to stop a running process.
+
== How to use pigz with tar ==
  
== Listing current jobs and their status ==
+
See [https://stackoverflow.com/a/39904353 here]:
  
  $ jobs
+
  $ tar cf - paths-to-archive | pigz --best -p 8 > archive.tgz
  
== Resuming a stopped job in the backgroud ==
+
== Best parallel compression with pigz ==
  
To resume a stopped process in the background
+
$ pigz --best
  
$ bg %1
+
== Best parallel compression with xz ==
  
where '1' is the job number reported by bash when you pressed Ctrl+Z (or ran 'jobs').
+
$ xz -9e -T 0
  
== Resuming a stopped job in the foreground ==
+
== Reporting compression ratios with xz ==
  
To resume a stopped process in the foreground
+
e.g.
  
  $ fg %1
+
  root@love:/data/image/archive# xz -l *
 +
Strms  Blocks  Compressed Uncompressed  Ratio  Check  Filename
 +
    1       3    372.2 MiB    442.3 MiB  0.841  CRC64  1999.txz
 +
    1      29  5,281.3 MiB  5,542.5 MiB  0.953  CRC64  2001.txz
 +
    1      11  1,364.3 MiB  2,084.3 MiB  0.655  CRC64  2002.txz
 +
    1      9    568.5 MiB  1,660.2 MiB  0.342  CRC64  2003.txz
 +
    1    639    66.8 GiB    119.6 GiB  0.558  CRC64  2004.txz
 +
    1    313    12.7 GiB    58.6 GiB  0.217  CRC64  2005.txz
 +
    1    414    35.0 GiB    77.4 GiB  0.452  CRC64  2006.txz
 +
    1    485    44.5 GiB    90.9 GiB  0.490  CRC64  2007.txz
 +
    1  1,690    150.0 GiB    316.8 GiB  0.473  CRC64  2008.txz
 +
    1      3    457.9 MiB    526.0 MiB  0.871  CRC64  2009.txz
 +
    1    168    27.3 GiB    31.4 GiB  0.868  CRC64  2010.txz
 +
    1      4    477.1 MiB    702.8 MiB  0.679  CRC64  2011.txz
 +
-------------------------------------------------------------------------------
 +
    12  3,768    344.6 GiB    705.5 GiB  0.488  CRC64  12 files
  
where '1' is the job number reported by bash when you pressed Ctrl+Z (or ran 'jobs').
+
= Symbolic-link management =
  
== Killing a stopped job ==
+
== Data used by sym-linked files:
  
To kill a job
+
This will de-reference the sym-links in the current directory and tell you how much data the files pointed to by the sym-links are using:
  
  $ kill %1
+
  jj5@tact:/data/backup/unity/latest$ du -hD * | sort -h
  
where '1' is the job number reported by bash when you pressed Ctrl+Z (or ran 'jobs').
+
= File searching =
  
== Periodically run a program and watch its output ==
+
== Finding a file with a particular name ==
  
  $ watch /your/command
+
  $ find -iname "*some-part-of-the-file-name*"
  
= Debian/Ubuntu package management =
+
Will start searching from the current directory, so maybe
  
Also see [https://wiki.debian.org/WhereIsIt Where "is" it?] on the Debian Wiki.
+
$ cd /
  
== configuring debconf ==
+
first. For a case-sensitive search:
  
  # dpkg-reconfigure debconf
+
  $ find -name "*eXaCT CaSE*"
  
Set priority to low to get asked detailed questions.
+
== Finding a file with particular content ==
  
== Showing list of installed packages ==
+
To search in /etc/ for a file with particular content:
  
  # dpkg --get-selections
+
  $ grep -R "search-string" /etc/*
  
== Searching for installed package ==
+
To search the current directory for *.cs files containing the word "Up":
  
  # dpkg --get-selections | grep package-name
+
  $ find . -name '*.cs' -exec grep --color=auto -H Up {} \;
  
or
+
== Finding a list of files with particular content ==
  
# aptitude search package-name
+
E.g. to find all the files with the word 'creativity':
  
== Showing which files are installed as part of a package ==
+
$ grep -R creativity . | sed 's/:/ /' | awk '{ print $1 }' | sort | uniq
  
# dpkg -L package-name
+
== Using the locate command to find files ==
  
== Installing a package ==
+
$ locate part-of-filename
  
# apt-get install package-name
+
E.g.
  
== Uninstalling a package ==
+
$ locate texvc
  
# apt-get remove package-name
+
== Updating locate command's database ==
  
== Showing system architecture ==
+
# updatedb
  
$ dpkg --print-architecture
+
= Job control =
  
== Showing which package a file belongs to ==
+
== Stopping a running process ==
  
$ which echo
+
Press Ctrl+Z to stop a running process.
/bin/echo
 
$ dpkg -S /bin/echo
 
coreutils: /bin/echo
 
$ dpkg -l | grep coreutils
 
ii  coreutils                        6.10-6                  The GNU core utilities
 
  
== Showing package information ==
+
== Listing current jobs and their status ==
  
  $ apt-cache showpkg coreutils
+
  $ jobs
  
Or for even more information:
+
== Resuming a stopped job in the backgroud ==
  
$ apt-cache show coreutils
+
To resume a stopped process in the background
  
== List all installed packages with package version info ==
+
$ bg %1
  
dpkg-query -l
+
where '1' is the job number reported by bash when you pressed Ctrl+Z (or ran 'jobs').
  
== Reporting which version of a package is installed ==
+
== Resuming a stopped job in the foreground ==
  
  $ dpkg -l | grep package-name
+
To resume a stopped process in the foreground
 +
 
 +
  $ fg %1
  
E.g.:
+
where '1' is the job number reported by bash when you pressed Ctrl+Z (or ran 'jobs').
  
root@hope:~/letsencrypt# dpkg -l | grep augeas
+
== Killing a stopped job ==
ii  augeas-lenses                  0.7.0-1ubuntu1                Set of lenses needed by libaugeas0 to parse
 
ii  libaugeas0                      0.7.0-1ubuntu1                The augeas configuration editing library and
 
  
== Comprehensive upgrade ==
+
To kill a job
  
Try the following:
+
$ kill %1
  
# apt-get update
+
where '1' is the job number reported by bash when you pressed Ctrl+Z (or ran 'jobs').
# apt-get dist-upgrade
 
# apt-get autoremove
 
# apt-get remove $(deborphan)
 
# update-flashplugin-nonfree --install
 
  
= Networking =
+
== Periodically run a program and watch its output ==
  
== net-tools vs iproute2 ==
+
$ watch /your/command
  
The older 'net-tools' package has been replaced with 'iproute2' e.g. in [https://www.debian.org/releases/stretch/amd64/release-notes/ch-information.en.html#iproute2 stretch].
+
= Debian/Ubuntu package management =
  
{|class="wikitable"
+
Also see [https://wiki.debian.org/WhereIsIt Where "is" it?] on the Debian Wiki.
! legacy net-tools commands
 
! iproute2 replacement commands
 
|-
 
| arp      || ip n (ip neighbor)
 
|-
 
| ifconfig || ip a (ip addr), ip link, ip -s (ip -stats)
 
|-
 
| iptunnel || ip tunnel
 
|-
 
| iwconfig || iw
 
|-
 
| nameif  || ip link, ifrename
 
|-
 
| netstat  || ss, ip route (for netstat-r), ip -s link (for netstat -i), ip maddr (for netstat-g)
 
|-
 
| route    || ip r (ip route)
 
|}
 
  
== Restart networking ==
+
== configuring debconf ==
  
For servers:
+
# dpkg-reconfigure debconf
  
# service networking restart
+
Set priority to low to get asked detailed questions.
  
For desktops:
+
== Showing list of installed packages ==
  
  # service network-manager restart
+
  # dpkg --get-selections
  
== Pinging with particular packet size ==
+
== Searching for installed package ==
  
  $ ping -M do -s <packet size in bytes> <host>
+
  # dpkg --get-selections | grep package-name
  
E.g.
+
or
  
  $ ping -M do -s 1400 charity.progclub.org
+
  # aptitude search package-name
  
== Setting [http://en.wikipedia.org/wiki/Maximum_segment_size MSS] for a particular IP address on a particular interface ==
+
== Showing which files are installed as part of a package ==
  
  # ip route add <host> dev <interface> advmss <packet size>
+
  # dpkg -L package-name
  
E.g.
+
== Installing a package ==
  
  # ip route add 10.0.0.1 dev eth0 advmss 1400
+
  # apt-get install package-name
  
== Dropping configured MMS for a particular IP address ==
+
== Uninstalling a package ==
  
  # ip route flush <host>
+
  # apt-get remove package-name
  
E.g.
+
== Showing system architecture ==
  
  # ip route flush 10.0.0.1
+
  $ dpkg --print-architecture
  
== Listing open ports and socket information ==
+
== Showing which package a file belongs to ==
  
Including which process is listening on which port.
+
$ which echo
 
+
/bin/echo
  # netstat -tulpn
+
$ dpkg -S /bin/echo
 +
coreutils: /bin/echo
 +
$ dpkg -l | grep coreutils
 +
  ii  coreutils                        6.10-6                  The GNU core utilities
  
Or use the 'ss' command:
+
== Showing package information ==
  
  # ss -s
+
  $ apt-cache showpkg coreutils
# ss -l
 
# ss -pl
 
# ss -o state established '( dport = :smtp or sport = :smtp )'
 
  
== Listing open IPv4 connections ==
+
Or for even more information:
  
  # lsof -Pnl +M -i4
+
  $ apt-cache show coreutils
  
You might need to install the lsof package:
+
== List all installed packages with package version info ==
  
  # apt-get install lsof
+
  dpkg-query -l
  
== Query for DNS MX record ==
+
== Reporting which version of a package is installed ==
  
  $ nslookup
+
  $ dpkg -l | grep package-name
> server 127.0.0.1
 
> set q=mx
 
> mail.blackbrick.com
 
  
== Query for DNS SOA record ==
+
E.g.:
  
  $ dig @ns2.staticmagic.net -t SOA staticmagic.net
+
  root@hope:~/letsencrypt# dpkg -l | grep augeas
 +
ii  augeas-lenses                  0.7.0-1ubuntu1                Set of lenses needed by libaugeas0 to parse
 +
ii  libaugeas0                      0.7.0-1ubuntu1                The augeas configuration editing library and
  
== Using nmap to list open ports on remote host ==
+
== Comprehensive upgrade ==
  
To check the 1,000 most common ports:
+
Try the following:
  
  # nmap server.example.com
+
  # apt-get update
 +
# apt-get dist-upgrade
 +
# apt-get autoremove
 +
# apt-get remove $(deborphan)
 +
# update-flashplugin-nonfree --install
  
Or for a specific port range (e.g. 101 to 102):
+
== Searching all available packages ==
  
  # nmap -p 101-102 server.example.com
+
  $ apt-cache search . | sort -d | less
  
Or for all ports (1 to 65,535):
+
= Networking =
  
# nmap -p- server.example.com
+
== net-tools vs iproute2 ==
  
== Network monitoring ==
+
The older 'net-tools' package has been replaced with 'iproute2' e.g. in [https://www.debian.org/releases/stretch/amd64/release-notes/ch-information.en.html#iproute2 stretch].
  
See [http://www.binarytides.com/linux-commands-monitor-network/ here] for details. Basically:
+
{|class="wikitable"
 
+
! legacy net-tools commands
# Overall bandwidth: nload, bmon, slurm, bwm-ng, cbm, speedometer, netload
+
! iproute2 replacement commands
# Overall bandwidth (batch style output): vnstat, ifstat, dstat, collectl
+
|-
# Bandwidth per socket connection: iftop, iptraf, tcptrack, pktstat, netwatch, trafshow
+
| arp      || ip n (ip neighbor)
# Bandwidth per process: nethogs
+
|-
 +
| ifconfig || ip a (ip addr), ip link, ip -s (ip -stats)
 +
|-
 +
| iptunnel || ip tunnel
 +
|-
 +
| iwconfig || iw
 +
|-
 +
| nameif  || ip link, ifrename
 +
|-
 +
| netstat  || ss, ip route (for netstat-r), ip -s link (for netstat -i), ip maddr (for netstat-g)
 +
|-
 +
| route    || ip r (ip route)
 +
|}
  
== nload ==
+
== Restart networking ==
  
You can watch network traffic in real-time with nload:
+
For servers:
  
  # nload -u M
+
  # service networking restart
  
== Reporting network (NIC) speed ==
+
For desktops:
  
From [https://askubuntu.com/questions/431911/how-can-i-verify-the-speed-of-my-nic-in-ubuntu#431912 here]:
+
# service network-manager restart
  
# dmesg | grep eth0
+
== Pinging with particular packet size ==
# mii-tool -v eth0
 
# ethtool eth0
 
  
Note: use ifconfig to get device name.
+
$ ping -M do -s <packet size in bytes> <host>
 +
 
 +
E.g.
  
== Path MTU discovery ==
+
$ ping -M do -s 1400 charity.progclub.org
  
To do a [https://en.wikipedia.org/wiki/Path_MTU_Discovery Path MTU Discovery], from the iputils-tracepath package:
+
== Setting [http://en.wikipedia.org/wiki/Maximum_segment_size MSS] for a particular IP address on a particular interface ==
  
  # tracepath host.example.com
+
  # ip route add <host> dev <interface> advmss <packet size>
  
== Listing available Ethernet devices ==
+
E.g.
  
To see a list of NICs available on the host:
+
# ip route add 10.0.0.1 dev eth0 advmss 1400
  
$ cat /proc/net/dev
+
== Dropping configured MMS for a particular IP address ==
  
== Links ==
+
# ip route flush <host>
  
* [http://www.cyberciti.biz/faq/linux-unix-open-ports/ HowTo: UNIX / Linux Open TCP / UDP Ports]
+
E.g.
  
= IPTables =
+
# ip route flush 10.0.0.1
  
== Applying firewall rules ==
+
== Listing open ports and socket information ==
  
For configuration info see [http://articles.slicehost.com/2008/4/25/ubuntu-hardy-setup-page-1 this article].
+
Including which process is listening on which port.
  
$ sudo vim /etc/iptables.test.rules
+
  # netstat -tulpn
$ sudo /sbin/iptables -F
 
$ sudo /sbin/iptables-restore < /etc/iptables.test.rules
 
$ sudo iptables -L
 
$ sudo -s
 
  # iptables-save > /etc/iptables.up.rules
 
# exit
 
  
= ufw =
+
Or use the 'ss' command:
  
== Denying hosts with ufw ==
+
# ss -s
 +
# ss -l
 +
# ss -pl
 +
# ss -o state established '( dport = :smtp or sport = :smtp )'
  
See [[Admin_reference#Denying_hosts_with_UFW|denying hosts with ufw]].
+
== Listing open IPv4 connections ==
  
= Bind9 =
+
# lsof -Pnl +M -i4
  
== Viewing Bind9 querylog ==
+
You might need to install the lsof package:
  
  $ sudo rndc querylog
+
  # apt-get install lsof
$ tail -f /var/log/syslog
 
  
= IPSec =
+
== Query for DNS MX record ==
  
== Disabling IPSec ==
+
$ nslookup
 +
> server 127.0.0.1
 +
> set q=mx
 +
> mail.blackbrick.com
  
# setkey -FP
+
== Query for DNS SOA record ==
  
= OpenSSL =
+
$ dig @ns2.staticmagic.net -t SOA staticmagic.net
  
== Debugging IMAPS with OpenSSL ==
+
== Using nmap to list open ports on remote host ==
  
# openssl s_client -connect localhost:993
+
To check the 1,000 most common ports:
> a1 LOGIN username@host password
 
> a2 LOGOUT
 
  
== Debugging HTTPS with OpenSSL ==
+
# nmap server.example.com
  
$ openssl s_client -connect www.example.com:443
+
Or for a specific port range (e.g. 101 to 102):
GET /example.html HTTP/1.1
 
host: www.example.com
 
  
== Links ==
+
# nmap -p 101-102 server.example.com
  
* [http://www.madboa.com/geek/openssl/ OpenSSL Command-Line HOWTO]
+
Or for all ports (1 to 65,535):
  
= Pluggable Authentication Modules (PAM) =
+
# nmap -p- server.example.com
  
== Links ==
+
== Network monitoring ==
  
* [http://www.centos.org/docs/5/html/Deployment_Guide-en-US/ch-pam.html 42.4. Pluggable Authentication Modules (PAM)]
+
See [http://www.binarytides.com/linux-commands-monitor-network/ here] for details. Basically:
  
= SSH =
+
# Overall bandwidth: nload, bmon, slurm, bwm-ng, cbm, speedometer, netload
 
+
# Overall bandwidth (batch style output): vnstat, ifstat, dstat, collectl
== Configuring SSH key login ==
+
# Bandwidth per socket connection: iftop, iptraf, tcptrack, pktstat, netwatch, trafshow
 +
# Bandwidth per process: nethogs
  
On the client machine generate a key-pair (if necessary, check for existing ~/.ssh/id_rsa.pub):
+
== nload ==
  
$ ssh-keygen -t rsa
+
You can watch network traffic in real-time with nload:
  
Copy the public key from the client to the server:
+
# nload -u M
  
$ scp ~/.ssh/id_rsa.pub user@example.org:
+
== Reporting network (NIC) speed ==
  
Configure the authorized keys on the server:
+
From [https://askubuntu.com/questions/431911/how-can-i-verify-the-speed-of-my-nic-in-ubuntu#431912 here]:
  
  $ ssh user@example.org
+
  # dmesg | grep eth0
  $ mkdir ~/.ssh
+
  # mii-tool -v eth0
$ chmod go-w .ssh
+
  # ethtool eth0
$ cat ~/id_rsa.pub >> ~/.ssh/authorized_keys
 
$ chmod 600 ~/.ssh/authorized_keys
 
  $ rm ~/id_rsa.pub
 
  
== Tunneling over SSH ==
+
Note: use ifconfig to get device name.
  
For example, connecting a remote MySQL server to the localhost:
+
== Path MTU discovery ==
  
$ ssh -L 3306:localhost:3306 jselliot@ssh.progsoc.org
+
To do a [https://en.wikipedia.org/wiki/Path_MTU_Discovery Path MTU Discovery], from the iputils-tracepath package:
  
If the machine you want to connect to is not the localhost of the machine you're ssh'ing to,
+
# tracepath host.example.com
 +
 
 +
== Listing available Ethernet devices ==
  
  $ ssh -L 3306:muspell.progsoc.uts.edu.au:3306 ssh.progsoc.uts.edu.au
+
To see a list of NICs available on the host:
  
The -L stanza is localport:remotehost:remoteport where localport is a
+
$ cat /proc/net/dev
port on your machine, forwarded to remoteport on remotehost.
 
  
== Tunneling over SSH with PuTTY ==
+
Also
  
See [http://www.anchor.com.au/hosting/support/MySQL/Connecting_to_mysql_remotely Connecting to the MySQL database remotely (via an SSH Tunnel)]
+
$ ip link
  
* run putty.exe
+
== 59 Linux Networking commands and scripts ==
* Connection -> SSH -> Tunnels
 
** Port forwarding: source port to 3306
 
** destination: 127.0.0.1:3306
 
** check Local
 
** click Add
 
  
== Enabling verbose SSH logging ==
+
See [https://haydenjames.io/linux-networking-commands-scripts/ 59 Linux Networking commands and scripts].
  
To see what's going on with your ssh connections,
+
== Links ==
  
$ ssh -v user@host
+
* [http://www.cyberciti.biz/faq/linux-unix-open-ports/ HowTo: UNIX / Linux Open TCP / UDP Ports]
  
Or
+
= IPTables =
  
$ ssh -vv user@host
+
== Applying firewall rules ==
  
== Unlocking SSH key for session ==
+
For configuration info see [http://articles.slicehost.com/2008/4/25/ubuntu-hardy-setup-page-1 this article].
  
  jj5@orac:~/.config/autostart$ cat ssh-add.desktop
+
  $ sudo vim /etc/iptables.test.rules
  [Desktop Entry]
+
  $ sudo /sbin/iptables -F
  Type=Application
+
  $ sudo /sbin/iptables-restore < /etc/iptables.test.rules
  Name=ssh-add
+
  $ sudo iptables -L
  Comment=Adds my private key to my session.
+
  $ sudo -s
  Exec=/usr/bin/konsole -e 'ssh-add /home/$USER/.ssh/id_rsa'
+
  # iptables-save > /etc/iptables.up.rules
 +
# exit
  
== Links ==
+
= ufw =
  
* [http://blogs.perl.org/users/smylers/2011/08/ssh-productivity-tips.html SSH Can Do That? Productivity Tips for Working with Remote Servers]
+
== Denying hosts with ufw ==
* [http://www.chiark.greenend.org.uk/~sgtatham/putty/download.html PuTTY Download Page]
 
  
= Standard IO =
+
See [[Admin_reference#Denying_hosts_with_UFW|denying hosts with ufw]].
  
== cat EOF ==
+
= Bind9 =
  
$ cat > output <<EOF
+
== Viewing Bind9 querylog ==
> text
 
> EOF
 
  
  $ cat output
+
  $ sudo rndc querylog
  text
+
  $ tail -f /var/log/syslog
  
= Script =
+
= IPSec =
  
== Creating a session log with script ==
+
== Disabling IPSec ==
  
  $ script -t 2> timing
+
  # setkey -FP
  
The session log is in the file 'typescript' and the timing data is in 'timing'.
+
= OpenSSL =
  
== Replaying a scripted session ==
+
== Debugging IMAPS with OpenSSL ==
  
  $ scriptreplay timing
+
  # openssl s_client -connect localhost:993
 +
> a1 LOGIN username@host password
 +
> a2 LOGOUT
  
Uses the default file 'typescript' and the 'timing' file as specified.
+
== Debugging HTTPS with OpenSSL ==
  
= Screen =
+
$ openssl s_client -connect www.example.com:443
 +
GET /example.html HTTP/1.1
 +
host: www.example.com
  
== Creating a new screen or reconnecting to a detached screen ==
+
== Links ==
  
$ screen -R
+
* [http://www.madboa.com/geek/openssl/ OpenSSL Command-Line HOWTO]
  
== Detaching a screen ==
+
= Pluggable Authentication Modules (PAM) =
  
$ screen -D
+
== Links ==
  
== Reconnecting to screen ==
+
* [http://www.centos.org/docs/5/html/Deployment_Guide-en-US/ch-pam.html 42.4. Pluggable Authentication Modules (PAM)]
  
$ screen -D
+
= SSH =
$ screen -R
 
  
I have a script in ~/bin/reconnect like so,
+
== Configuring SSH key login ==
  
#!/bin/bash
+
On the client machine generate a key-pair (if necessary, check for existing ~/.ssh/id_rsa.pub):
screen -D
 
screen -R
 
  
This will detach your last screen, and reconnect it on the current terminal.
+
$ ssh-keygen -t rsa
  
== Scrolling in screen ==
+
Copy the public key from the client to the server:
  
See [https://www.saltycrane.com/blog/2008/01/how-to-scroll-in-gnu-screen/ How to scroll in GNU Screen]. Basically press Ctrl+A ESC then use Page Up and Page Down. Press ESC again to exit copy mode. As usual you can use Ctrl+[ in place of ESC.
+
$ scp ~/.ssh/id_rsa.pub user@example.org:
  
= Vim =
+
Configure the authorized keys on the server:
  
== First, why Vim? ==
+
$ ssh user@example.org
 +
$ mkdir ~/.ssh
 +
$ chmod go-w .ssh
 +
$ cat ~/id_rsa.pub >> ~/.ssh/authorized_keys
 +
$ chmod 600 ~/.ssh/authorized_keys
 +
$ rm ~/id_rsa.pub
  
Read [http://www.viemu.com/a-why-vi-vim.html Why, oh WHY, do those #?@! nutheads use vi?]
+
== Tunneling over SSH ==
  
== Visual modes ==
+
For example, connecting a remote MySQL server to the localhost:
  
Use 'v' for visual mode, 'V' for visual line mode and Ctrl+V for visual block mode.
+
$ ssh -L 3306:localhost:3306 jselliot@ssh.progsoc.org
  
== Configuring spaces instead of tabs ==
+
If the machine you want to connect to is not the localhost of the machine you're ssh'ing to,
  
I use two spaces instead of tabs. To configure, edit your .vimrc file:
+
  $ ssh -L 3306:muspell.progsoc.uts.edu.au:3306 ssh.progsoc.uts.edu.au
  
$ vim ~/.vimrc
+
The -L stanza is localport:remotehost:remoteport where localport is a
 +
port on your machine, forwarded to remoteport on remotehost.
  
and include the following lines:
+
== Tunneling over SSH with PuTTY ==
  
set tabstop=2
+
See [http://www.anchor.com.au/hosting/support/MySQL/Connecting_to_mysql_remotely Connecting to the MySQL database remotely (via an SSH Tunnel)]
set shiftwidth=2
 
set expandtab
 
  
== Configuring syntax highlighting ==
+
* run putty.exe
 +
* Connection -> SSH -> Tunnels
 +
** Port forwarding: source port to 3306
 +
** destination: 127.0.0.1:3306
 +
** check Local
 +
** click Add
  
See [http://www.cyberciti.biz/faq/turn-on-or-off-color-syntax-highlighting-in-vi-or-vim/ here].
+
== Enabling verbose SSH logging ==
  
Use:
+
To see what's going on with your ssh connections,
  
  :syntax on
+
  $ ssh -v user@host
  
to turn on syntax highlighting.
+
Or
  
Use:
+
$ ssh -vv user@host
  
:syntax off
+
== Unlocking SSH key for session ==
  
to turn off syntax highlighting.
+
jj5@orac:~/.config/autostart$ cat ssh-add.desktop
 +
[Desktop Entry]
 +
Type=Application
 +
Name=ssh-add
 +
Comment=Adds my private key to my session.
 +
Exec=/usr/bin/konsole -e 'ssh-add /home/$USER/.ssh/id_rsa'
  
To always use syntax highlighting:
+
== Links ==
  
$ vim ~/.vimrc
+
* [http://blogs.perl.org/users/smylers/2011/08/ssh-productivity-tips.html SSH Can Do That? Productivity Tips for Working with Remote Servers]
 +
* [http://www.chiark.greenend.org.uk/~sgtatham/putty/download.html PuTTY Download Page]
  
and add:
+
= Standard IO =
  
syntax on
+
== cat EOF ==
  
To get a list of supported colour schemes open vim and type:
+
$ cat > output <<EOF
 +
> text
 +
> EOF
  
  :colorscheme[space][Ctrl+D]
+
  $ cat output
 +
text
  
To always use a particular colorscheme edit ~/.vimrc and add (for example):
+
= Script =
  
colorscheme desert
+
== Creating a session log with script ==
  
== Inserting a TAB character when expandtab is on ==
+
$ script -t 2> timing
  
The problem here is that you have configured vim to insert spaces, but for a particular file (e.g. a Makefile) you need to insert a character.
+
The session log is in the file 'typescript' and the timing data is in 'timing'.
  
Press Ctrl+V TAB to insert a literal tab character.
+
== Replaying a scripted session ==
  
Or you can disable tab expansion altogether with:
+
$ scriptreplay timing
  
:set expandtab!
+
Uses the default file 'typescript' and the 'timing' file as specified.
  
== Changing 2 space indent to 4 space indent (e.g. for python files) ==
+
= Screen =
  
:%s/^\s*/&&/g
+
== Creating a new screen or reconnecting to a detached screen ==
  
For more information [https://www.progclub.org/blog/2013/08/10/vim-reformat-a-python-file-to-have-4-space-indentations/ see here].
+
$ screen -R
  
== Recording and replaying a macro ==
+
== Detaching a screen ==
  
To record a macro press 'q' and then a number between 1 and 9. E.g. press "q1". The macro is now recording. When you've finished issuing your commands press 'q' again to finish recording. To replay a macro press '@' followed by the number of the macro. That is, if you pressed "q1" to record the macro, press "@1" to replay the macro. To replay the last macro again press "@@".
+
$ screen -D
  
== Deleting to end of line ==
+
== Reconnecting to screen ==
  
  d$
+
  $ screen -D
 +
$ screen -R
  
== Deleting to beginning of line ==
+
I have a script in ~/bin/reconnect like so,
  
  d^
+
  #!/bin/bash
 +
screen -D
 +
screen -R
  
== Finding text ==
+
This will detach your last screen, and reconnect it on the current terminal.
  
To search forward for "text":
+
== Scrolling in screen ==
  
/text
+
See [https://www.saltycrane.com/blog/2008/01/how-to-scroll-in-gnu-screen/ How to scroll in GNU Screen]. Basically press Ctrl+A ESC then use Page Up and Page Down. Press ESC again to exit copy mode. As usual you can use Ctrl+[ in place of ESC.
  
To search backward for "text":
+
= Vim =
  
?text
+
== First, why Vim? ==
  
To repeat the last search in a forward direction press 'n', or to search again backwards press 'N'.
+
Read [http://www.viemu.com/a-why-vi-vim.html Why, oh WHY, do those #?@! nutheads use vi?]
  
== Finding and replacing text ==
+
== Visual modes ==
  
To replace the first instance of "search" on the current line with "destroy":
+
Use 'v' for visual mode, 'V' for visual line mode and Ctrl+V for visual block mode.
  
:s/search/destroy/
+
== Configuring spaces instead of tabs ==
  
To replace all instances of "search" on the current line with "destroy":
+
I use two spaces instead of tabs. To configure, edit your .vimrc file:
  
  :s/search/destroy/g
+
  $ vim ~/.vimrc
  
To replace all instances of "search" on lines 13 to 37 with "destroy":
+
and include the following lines:
  
  :13,37 s/search/destroy/g
+
  set tabstop=2
 +
set shiftwidth=2
 +
set expandtab
  
To replace all instances of "search" in the entire file with "destroy":
+
== Configuring syntax highlighting ==
  
:%s/search/destroy/g
+
See [http://www.cyberciti.biz/faq/turn-on-or-off-color-syntax-highlighting-in-vi-or-vim/ here].
  
== Changing DOS/Windows line-endings (CRLF) to Unix line-endings ==
+
Use:
  
To set the line-ending to Unix line endings run the command:
+
:syntax on
  
:setlocal ff=unix
+
to turn on syntax highlighting.
  
More information on managing file formats [http://vim.wikia.com/wiki/File_format available here].
+
Use:
  
== Disabling auto-indent etc. to paste from clipboard ==
+
:syntax off
  
To disable smart indenting when you're going to paste in text:
+
to turn off syntax highlighting.
  
:set paste
+
To always use syntax highlighting:
  
To turn it off again:
+
$ vim ~/.vimrc
  
:set nopaste
+
and add:
  
There's more info in this article: [http://vim.wikia.com/wiki/Toggle_auto-indenting_for_code_paste Toggle auto-indenting for code paste]
+
syntax on
  
== Positioning windows ==
+
To get a list of supported colour schemes open vim and type:
  
Use -o for horizontal split, e.g.:
+
:colorscheme[space][Ctrl+D]
  
vim -o a.txt b.txt
+
To always use a particular colorscheme edit ~/.vimrc and add (for example):
  
Use -O for vertical split, e.g.:
+
colorscheme desert
  
vim -o a.txt b.txt
+
== Inserting a TAB character when expandtab is on ==
  
Use ^W to navigate windows then use directional keys h, j, k, l, etc.
+
The problem here is that you have configured vim to insert spaces, but for a particular file (e.g. a Makefile) you need to insert a character.
  
Use ^W and &lt; or &gt; to resize windows.
+
Press Ctrl+V TAB to insert a literal tab character.
  
== To indent a block of text in Vim ==
+
Or you can disable tab expansion altogether with:
  
Use the > command. E.g. to indent five lines:
+
:set expandtab!
  
5 > >
+
== Changing 2 space indent to 4 space indent (e.g. for python files) ==
  
Press . (dot) to keep indenting.
+
:%s/^\s*/&&/g
  
Or inside a block (e.g. curly brace, HTML/XML element, etc.) you can put your cursor in the element on on the curly brace and then:
+
For more information [https://www.progclub.org/blog/2013/08/10/vim-reformat-a-python-file-to-have-4-space-indentations/ see here].
  
> %
+
== Recording and replaying a macro ==
  
See [http://stackoverflow.com/questions/235839/indent-multiple-lines-quickly-in-vi#235841 here] for more.
+
To record a macro press 'q' and then a number between 1 and 9. E.g. press "q1". The macro is now recording. When you've finished issuing your commands press 'q' again to finish recording. To replay a macro press '@' followed by the number of the macro. That is, if you pressed "q1" to record the macro, press "@1" to replay the macro. To replay the last macro again press "@@".
  
== Open a file in a new window/tab ==
+
== Deleting to end of line ==
  
To open a file on the left hand side:
+
d$
  
:vert new filename.ext
+
== Deleting to beginning of line ==
  
Note: ':vnew filename.ext' and ':vsp filename.ext' also work.
+
d^
  
To open a file at the top:
+
== Finding text ==
  
:new filename.ext
+
To search forward for "text":
  
See [http://stackoverflow.com/questions/10760310/how-to-open-a-new-file-in-vim-in-a-new-window#10762678 here] for more.
+
/text
  
== Explore files in Vim ==
+
To search backward for "text":
  
Enter:
+
?text
  
:Explore
+
To repeat the last search in a forward direction press 'n', or to search again backwards press 'N'.
  
== Switch between Vim tabs ==
+
== Finding and replacing text ==
  
Use gt and gT.
+
To replace the first instance of "search" on the current line with "destroy":
  
== Switch between Vim windows ==
+
:s/search/destroy/
  
To toggle between open windows use:
+
To replace all instances of "search" on the current line with "destroy":
  
  Ctrl+W W
+
  :s/search/destroy/g
  
To move in a direction use:
+
To replace all instances of "search" on lines 13 to 37 with "destroy":
  
  Ctrl+W h/j/k/l
+
  :13,37 s/search/destroy/g
 +
 
 +
To replace all instances of "search" in the entire file with "destroy":
  
See [http://superuser.com/questions/280500/how-does-one-switch-between-windows-on-vim#280501 here] for more.
+
:%s/search/destroy/g
  
== Insert block comment in Vim ==
+
== Changing DOS/Windows line-endings (CRLF) to Unix line-endings ==
  
See [https://stackoverflow.com/a/253391/868138 here] for line-commenting.
+
To set the line-ending to Unix line endings run the command:
  
So it's:
+
:setlocal ff=unix
  
# Ctrl+V (Note: not Shift+V!)
+
More information on managing file formats [http://vim.wikia.com/wiki/File_format available here].
# Up/Down to select rows
 
# Shift+I
 
# Enter your text, e.g. '#' or '//'
 
# Ctrl+[ (or 'Esc')
 
  
== Navigate to matching tag ==
+
== Disabling auto-indent etc. to paste from clipboard ==
  
To navigate to the matching beginning or end tag use '%'.
+
To disable smart indenting when you're going to paste in text:
  
You can also use e.g. '[{' to match the previous '{', or e.g. '])' to match the next ')'.
+
:set paste
  
== Auto-format HTML tags ==
+
To turn it off again:
  
Stolen from [https://www.quora.com/How-do-you-auto-format-HTML-in-Vim here].
+
:set nopaste
  
# first join all the lines - ggVGgJ
+
There's more info in this article: [http://vim.wikia.com/wiki/Toggle_auto-indenting_for_code_paste Toggle auto-indenting for code paste]
# Now break tags to new lines - :%s/>\s*</>\r</g
 
# Now set filetype - :set ft=html (you can do this before too)
 
# Now Indent - ggVG=
 
  
== Links ==
+
== Positioning windows ==
  
* [http://www.vim.org/ Vim: the editor]
+
Use -o for horizontal split, e.g.:
* [http://yannesposito.com/Scratch/en/blog/Learn-Vim-Progressively/ Learn Vim Progressively]
 
* [http://michael.peopleofhonoronly.com/vim/ Vim cheat sheet for programmers]
 
* [http://stackoverflow.com/questions/4781070/how-to-insert-tab-character-when-expandtab-option-is-on-in-vim How to insert Tab character when expandtab option is ON in VIM]
 
* [https://www.linux.com/learn/tutorials/8255-vim-tips-the-basics-of-search-and-replace Vim tips: the basics of search and replace]
 
* [http://vim.wikia.com/wiki/File_format File format]
 
* [http://www.viemu.com/a_vi_vim_graphical_cheat_sheet_tutorial.html Graphical vi-vim Cheat Sheet and Tutorial]
 
* [http://www.angelwatt.com/coding/notes/vim-commands.html Vim Commands Cheat Sheet]
 
  
= Write =
+
vim -o a.txt b.txt
  
== Talking to other users on the system ==
+
Use -O for vertical split, e.g.:
  
'''write''' is a unix command for talking to other users on the system. To use '''write''':
+
vim -o a.txt b.txt
  
1. SSH to <username>@<hostname> and login with your username and password.
+
Use ^W to navigate windows then use directional keys h, j, k, l, etc.
  
2. Issue the following command to find out who is logged onto the system:
+
Use ^W and &lt; or &gt; to resize windows.
  
$ who
+
== To indent a block of text in Vim ==
  
3. Issue the following command to talk to a specific user:
+
Use the > command. E.g. to indent five lines:
  
  $ write <username>
+
  5 > >
  
4. Enter the message you'd like to send the user, followed by Ctrl+C to send. Press Ctrl+D to cancel.
+
Press . (dot) to keep indenting.
  
= Date =
+
Or inside a block (e.g. curly brace, HTML/XML element, etc.) you can put your cursor in the element on on the curly brace and then:
  
== Reporting the time on the server ==
+
> %
  
$ date
+
See [http://stackoverflow.com/questions/235839/indent-multiple-lines-quickly-in-vi#235841 here] for more.
  
== Reporting UTC time ==
+
== Open a file in a new window/tab ==
  
$ date --utc
+
To open a file on the left hand side:
  
== Getting the date in yyyy-MM-dd-hhmmss format ==
+
:vert new filename.ext
  
$ date="`date +%F-%H%M%S`"
+
Note: ':vnew filename.ext' and ':vsp filename.ext' also work.
  
== Getting the year in four digits ==
+
To open a file at the top:
  
  $ year="`date +%Y`"
+
  :new filename.ext
  
== Getting the month in two digits ==
+
See [http://stackoverflow.com/questions/10760310/how-to-open-a-new-file-in-vim-in-a-new-window#10762678 here] for more.
  
$ month="`date +%m`"
+
== Explore files in Vim ==
  
== Getting the day of the month in two digits ==
+
Enter:
  
  $ day="`date +%d`"
+
  :Explore
  
== Getting yesterday's date ==
+
== Switch between Vim tabs ==
  
$ date --date='1 day ago' +%Y-%m-%d
+
Use gt and gT.
  
== Converting Unix time (seconds since epoch) ==
+
== Switch between Vim windows ==
  
For timestamp '1501370200':
+
To toggle between open windows use:
  
  $ date -d @1501370200 +%F-%H%M%S
+
  Ctrl+W W
  
== Running timedatectl from systemd ==
+
To move in a direction use:
  
There's a new command bundled with systmed:
+
Ctrl+W h/j/k/l
  
# timedatectl
+
See [http://superuser.com/questions/280500/how-does-one-switch-between-windows-on-vim#280501 here] for more.
  
It reports on (and controls) how the system time is configured.
+
== Insert block comment in Vim ==
  
= MySQL =
+
See [https://stackoverflow.com/a/253391/868138 here] for line-commenting.
  
== Run mysql without authentication/authorisation ==
+
So it's:
  
# service mysql stop
+
# Ctrl+V (Note: not Shift+V!)
# mysqld_safe --skip-grant-tables &
+
# Up/Down to select rows
 
+
# Shift+I
Then you can connect without a password, e.g.:
+
# Enter your text, e.g. '#' or '//'
 +
# Ctrl+[ (or 'Esc')
  
# mysql -u root mysql
+
== Navigate to matching tag ==
  
To stop the unauthenticated service:
+
To navigate to the matching beginning or end tag use '%'.
  
# mysqladmin shutdown
+
You can also use e.g. '[{' to match the previous '{', or e.g. '])' to match the next ')'.
  
Then restart a normal service:
+
== Auto-format HTML tags ==
  
# service mysql start
+
Stolen from [https://www.quora.com/How-do-you-auto-format-HTML-in-Vim here].
  
== Logging all database queries ==
+
# first join all the lines - ggVGgJ
 +
# Now break tags to new lines - :%s/>\s*</>\r</g
 +
# Now set filetype - :set ft=html (you can do this before too)
 +
# Now Indent - ggVG=
  
# vim /etc/mysql/my.cnf
+
== Links ==
  
In the [mysqld] section add:
+
* [http://www.vim.org/ Vim: the editor]
 +
* [http://yannesposito.com/Scratch/en/blog/Learn-Vim-Progressively/ Learn Vim Progressively]
 +
* [http://michael.peopleofhonoronly.com/vim/ Vim cheat sheet for programmers]
 +
* [http://stackoverflow.com/questions/4781070/how-to-insert-tab-character-when-expandtab-option-is-on-in-vim How to insert Tab character when expandtab option is ON in VIM]
 +
* [https://www.linux.com/learn/tutorials/8255-vim-tips-the-basics-of-search-and-replace Vim tips: the basics of search and replace]
 +
* [http://vim.wikia.com/wiki/File_format File format]
 +
* [http://www.viemu.com/a_vi_vim_graphical_cheat_sheet_tutorial.html Graphical vi-vim Cheat Sheet and Tutorial]
 +
* [http://www.angelwatt.com/coding/notes/vim-commands.html Vim Commands Cheat Sheet]
  
log=/tmp/mysql.log
+
= Write =
  
Then:
+
== Talking to other users on the system ==
  
# service mysql restart
+
'''write''' is a unix command for talking to other users on the system. To use '''write''':
  
Watch the log with:
+
1. SSH to <username>@<hostname> and login with your username and password.
  
# tail -f /tmp/mysql.log
+
2. Issue the following command to find out who is logged onto the system:
  
== Dumping a MySQL database ==
+
$ who
  
You can dump the database into a file using:  
+
3. Issue the following command to talk to a specific user:
 
$ mysqldump -h hostname -u user --password=password databasename > filename
 
  
== Loading a MySQL database from a dump file ==
+
$ write <username>
  
You can create a database using:
+
4. Enter the message you'd like to send the user, followed by Ctrl+C to send. Press Ctrl+D to cancel.
  
$ echo create database databasename | mysql -h hostname -u user -p
+
= Date =
  
You can restore a database using:
+
== Reporting the time on the server ==
 
$ mysql -h hostname -u user --password=password databasename < filename
 
  
== Creating a MySQL user ==
+
$ date
  
# mysql -h localhost -u root --password=<password>
+
== Reporting UTC time ==
mysql> create user 'username'@'localhost' identified by '<password>';
 
  
== Granting all MySQL user permissions ==
+
$ date --utc
  
# mysql -h localhost -u root --password=<password>
+
== Getting the date in yyyy-MM-dd-hhmmss format ==
mysql> grant all privileges on dbname.* to user@host;
 
  
== Select domain name from email address ==
+
$ date="`date +%F-%H%M%S`"
  
SELECT SUBSTR( email, INSTR( email, '@' ) + 1 )
+
== Getting the year in four digits ==
  
== Check if MySQL connection is encrypted with TLS/SSL ==
+
$ year="`date +%Y`"
  
Check the SSL version in use:
+
== Getting the month in two digits ==
  
  show status like 'Ssl_version';
+
  $ month="`date +%m`"
  
Or check the cipher in use:
+
== Getting the day of the month in two digits ==
  
  show status like 'Ssl_cipher';
+
  $ day="`date +%d`"
  
= Apache =
+
== Getting yesterday's date ==
  
== Maintaining .htaccess passwords ==
+
$ date --date='1 day ago' +%Y-%m-%d
  
To add or modify the password for a user:
+
== Converting Unix time (seconds since epoch) ==
  
$ htpasswd /etc/apache2/passwd username
+
For timestamp '1501370200':
  
== Configuring PHP session timeout in .htaccess ==
+
$ date -d @1501370200 +%F-%H%M%S
  
For a session timeout of 9 hours:
+
== Running timedatectl from systemd ==
  
php_value session.cookie_lifetime 32400
+
There's a new command bundled with systmed:
php_value session.gc_maxlifetime 32400
 
  
== Disabling PHP magic quotes in .htaccess ==
+
# timedatectl
  
php_flag magic_quotes_gpc Off
+
It reports on (and controls) how the system time is configured.
  
== Requiring HTTP Auth in .htaccess ==
+
= MySQL =
  
AuthType Basic
+
== Run mysql without authentication/authorisation ==
AuthName "Speak Friend And Enter"
 
AuthUserFile /home/jj5/.htpasswd
 
Require valid-user
 
  
== Restarting Apache ==
+
# service mysql stop
 +
# mysqld_safe --skip-grant-tables &
  
The hard way
+
Then you can connect without a password, e.g.:
  
  $ sudo /etc/init.d/apache2 restart
+
  # mysql -u root mysql
  
The graceful way (avoids dropping active connections)
+
To stop the unauthenticated service:
  
  $ sudo apache2ctl graceful
+
  # mysqladmin shutdown
  
== Allowing directory browsing ==
+
Then restart a normal service:
  
To show directory index pages, in the apache config file:
+
# service mysql start
  
<Directory /var/www/data>
+
== Logging all database queries ==
  Options Indexes
 
</Directory>
 
  
= C =
+
# vim /etc/mysql/my.cnf
  
== Locating memset function ==
+
In the [mysqld] section add:
  
The memset function is in &lt;string.h> as described in this article [http://www.java-samples.com/showtutorial.php?tutorialid=591 Using memset(), memcpy(), and memmove() in C]
+
log=/tmp/mysql.log
  
== Links ==
+
Then:
  
* [http://www.ibm.com/developerworks/linux/library/l-memory/ Inside memory management]
+
# service mysql restart
  
= PHP =
+
Watch the log with:
  
== Including a file relative to the including file ==
+
# tail -f /tmp/mysql.log
  
require_once( dirname( __FILE__ ) . '/relative/path/to.php' );
+
== Dumping a MySQL database ==
  
== Enabling error reporting ==
+
You can dump the database into a file using:
 +
 +
$ mysqldump -h hostname -u user --password=password databasename > filename
  
error_reporting( E_ALL | E_STRICT );
+
== Loading a MySQL database from a dump file ==
ini_set( 'display_errors', 'On' );
 
  
== Setting an error handler ==
+
You can create a database using:
  
  set_error_handler( "error_handler", E_ALL | E_STRICT );
+
  $ echo create database databasename | mysql -h hostname -u user -p
  
  function error_handler( $error_code, $error_message, $error_file, $error_line, $error_context ) {
+
You can restore a database using:
  // ...
+
   
  }
+
  $ mysql -h hostname -u user --password=password databasename < filename
  
== Disable HTML content in var_dump ==
+
== Creating a MySQL user ==
  
  ini_set( 'html_errors', 'off' );
+
  # mysql -h localhost -u root --password=<password>
 +
mysql> create user 'username'@'localhost' identified by '<password>';
  
= BASH scripting =
+
== Granting all MySQL user permissions ==
  
For a primer on bash scripting see [http://www.progsoc.org/tfm/tfm03/node37.html TFM: Erotic Fantasy: /bin/sh Programming].
+
# mysql -h localhost -u root --password=<password>
 +
mysql> grant all privileges on dbname.* to user@host;
  
== Telling a script to run in bash ==
+
== Select domain name from email address ==
  
The first line of the file should be:
+
SELECT SUBSTR( email, INSTR( email, '@' ) + 1 )
  
#!/bin/bash
+
== Check if MySQL connection is encrypted with TLS/SSL ==
  
== Checking if a command-line argument was passed in ==
+
Check the SSL version in use:
  
  if [ -n "$1" ]; then
+
  show status like 'Ssl_version';
  echo "Missing parameter 1.";
 
  exit 1;
 
fi
 
  
== Checking if a command-line argument was not passed in ==
+
Or check the cipher in use:
  
  if [ "$1" = "" ]; then
+
  show status like 'Ssl_cipher';
  echo "Missing parameter 1.";
 
  exit 1;
 
fi
 
  
Or:
+
= Apache =
  
if [ -z "$1" ]; then
+
== Reporting loaded Apache modules ==
  echo "Missing parameter 1.";
 
  exit 1;
 
fi
 
  
== Checking command exit status ==
+
# apache2ctl -M
  
cd /my/path
+
== Maintaining .htaccess passwords ==
if [ "$?" -ne "0" ]; then
 
  echo "Cannot change dir.";
 
  exit 1;
 
fi
 
  
== Checking if a file does/doesn't exist ==
+
To add or modify the password for a user:
  
Check if file exists:
+
$ htpasswd /etc/apache2/passwd username
  
if [ -f "/my/file" ]; then
+
== Configuring PHP session timeout in .htaccess ==
  cat /my/file
 
fi
 
  
Check if file doesn't exist:
+
For a session timeout of 9 hours:
  
  if [ ! -f "/my/file" ]; then
+
  php_value session.cookie_lifetime 32400
  touch /my/file
+
  php_value session.gc_maxlifetime 32400
  fi
 
  
== Checking if a directory does/doesn't exist ==
+
== Disabling PHP magic quotes in .htaccess ==
  
Check if directory exists:
+
php_flag magic_quotes_gpc Off
  
if [ -d "/my/dir" ]; then
+
== Requiring HTTP Auth in .htaccess ==
  rmdir /my/dir
 
fi
 
  
Check if directory doesn't exist:
+
AuthType Basic
 +
AuthName "Speak Friend And Enter"
 +
AuthUserFile /home/jj5/.htpasswd
 +
Require valid-user
  
if [ ! -d "/my/dir" ]; then
+
== Restarting Apache ==
  mkdir /my/dir
 
fi
 
  
== Deleting old backups ==
+
The hard way
  
To keep only the latest five backups:
+
$ sudo /etc/init.d/apache2 restart
  
find . -maxdepth 1 -type f -printf '%T@ %p\0' | sort -r -z -n | awk 'BEGIN { RS="\0"; ORS="\0"; FS="" } NR > 5 { sub("^[0-9]*(.[0-9]*)? ", ""); print }' | xargs -0 rm -f
+
The graceful way (avoids dropping active connections)
  
This script stolen from [http://stackoverflow.com/questions/25785/delete-all-but-the-most-recent-x-files-in-bash stackoverflow].
+
$ sudo apache2ctl graceful
  
Requires GNU find for -printf, GNU sort for -z, GNU awk for "\0" and GNU xargs for -0, but handles files with embedded newlines or spaces.
+
== Allowing directory browsing ==
  
== Changing into the script's directory ==
+
To show directory index pages, in the apache config file:
  
  cd "`dirname $0`"
+
  <Directory /var/www/data>
 +
  Options Indexes
 +
</Directory>
  
== Getting the absolute path of a relative path ==
+
= C =
  
readlink -f ./some/path
+
== Locating memset function ==
  
== Creating a temp directory ==
+
The memset function is in &lt;string.h> as described in this article [http://www.java-samples.com/showtutorial.php?tutorialid=591 Using memset(), memcpy(), and memmove() in C]
  
dir=`mktemp -d` && cd $dir
+
== Links ==
  
== Reading secret input from stdin ==
+
* [http://www.ibm.com/developerworks/linux/library/l-memory/ Inside memory management]
  
You can read a secret, such as a password, like this:
+
= PHP =
  
echo -n "Enter passphrase: "
+
== Including a file relative to the including file ==
stty -echo
+
 
  read passphrase;
+
  require_once( dirname( __FILE__ ) . '/relative/path/to.php' );
stty echo
+
 
echo ""
+
== Enabling error reporting ==
  
After running the above the secret will be in the $passphrase environment variable.
+
error_reporting( E_ALL | E_STRICT );
 +
ini_set( 'display_errors', 'On' );
  
== String replacements in bash ==
+
== Setting an error handler ==
  
See the [http://www.tldp.org/LDP/abs/html/string-manipulation.html string manipulation] doco. Basically, to replace first occurrence:
+
set_error_handler( "error_handler", E_ALL | E_STRICT );
  
  result=${var/find/replace}
+
  function error_handler( $error_code, $error_message, $error_file, $error_line, $error_context ) {
 +
  // ...
 +
}
  
To replace all occurrences:
+
== Disable HTML content in var_dump ==
  
  result=${var//find/replace}
+
  ini_set( 'html_errors', 'off' );
  
A practical example, get an ISO date and turn it into a path:
+
== Report PHP modules ==
  
  date="$(date +%Y-%m-%d)"
+
  $ php -m
work_dir=${date//-//}
 
  
== Sending a HEREDOC to a file ==
+
== PHP Security Best Practices For Sys Admins ==
  
cat << EOF > /tmp/yourfilehere
+
See [https://www.cyberciti.biz/tips/php-security-best-practices-tutorial.html Linux 25 PHP Security Best Practices For Sys Admins].
These contents will be written to the file.
 
        This line is indented.
 
EOF
 
  
== Bash case/switch statement ==
+
= BASH scripting =
  
See [http://tldp.org/LDP/Bash-Beginners-Guide/html/sect_07_03.html using case statements], e.g.:
+
For a primer on bash scripting see [http://www.progsoc.org/tfm/tfm03/node37.html TFM: Erotic Fantasy: /bin/sh Programming].
  
case $space in
+
== Telling a script to run in bash ==
[1-6]*)
 
  Message="All is quiet."
 
  ;;
 
[7-8]*)
 
  Message="Start thinking about cleaning out some stuff.  There's a partition that is $space % full."
 
  ;;
 
9[1-8])
 
  Message="Better hurry with that new disk...  One partition is $space % full."
 
  ;;
 
99)
 
  Message="I'm drowning here!  There's a partition at $space %!"
 
  ;;
 
*)
 
  Message="I seem to be running with an nonexistent amount of disk space..."
 
  ;;
 
esac
 
  
== Using dotglob shopt to match dot-files ==
+
The first line of the file should be:
  
To enable dot-file matching in globs, set the dotglob shell option:
+
#!/bin/bash
  
$ shopt -s dotglob
+
== Checking if a command-line argument was passed in ==
  
= Sed =
+
if [ -n "$1" ]; then
 +
  echo "Missing parameter 1.";
 +
  exit 1;
 +
fi
  
== Find and replace with sed ==
+
== Checking if a command-line argument was not passed in ==
  
To update the current file use '-i'. E.g.:
+
if [ "$1" = "" ]; then
 +
  echo "Missing parameter 1.";
 +
  exit 1;
 +
fi
  
sed -i 's/search-text/replace-text/' file
+
Or:
  
= Awk =
+
if [ -z "$1" ]; then
 +
  echo "Missing parameter 1.";
 +
  exit 1;
 +
fi
  
== Listing IP addresses in an Apache web log ==
+
== Checking command exit status ==
  
  awk '/GET \/path\/for\/url/ { print $1 }' /var/log/apache2/access.log | sort | uniq
+
  cd /my/path
 +
if [ "$?" -ne "0" ]; then
 +
  echo "Cannot change dir.";
 +
  exit 1;
 +
fi
  
== Printing space-separated field ==
+
== Checking if a file does/doesn't exist ==
  
echo 'no no yes no' | awk '{print $3}'
+
Check if file exists:
  
== Printing delimited field ==
+
if [ -f "/my/file" ]; then
 +
  cat /my/file
 +
fi
  
echo 'no:no:yes:no' | awk -F ':' '{print $3}'
+
Check if file doesn't exist:
  
= Subversion =
+
if [ ! -f "/my/file" ]; then
 +
  touch /my/file
 +
fi
  
== Setting svn:externals from the command-line ==
+
== Checking if a directory does/doesn't exist ==
  
See [http://beerpla.net/2009/06/20/how-to-properly-set-svn-svnexternals-property-in-svn-command-line/ here].
+
Check if directory exists:
  
To set an svn:externals from the command-line:
+
if [ -d "/my/dir" ]; then
 +
  rmdir /my/dir
 +
fi
  
svn propset svn:externals 'rdfind-php https://www.progclub.org/svn/pcrepo/rdfind.php/branches/0.1' .
+
Check if directory doesn't exist:
svn ci -m 'Adding svn:externals for rdfind-php...'
 
svn up
 
  
Or to use a file:
+
if [ ! -d "/my/dir" ]; then
 +
  mkdir /my/dir
 +
fi
  
svn propset svn:externals -F svn.externals .
+
== Deleting old backups ==
  
== Setting svn:ignore from the command line ==
+
To keep only the latest five backups:
  
See [http://tedone.typepad.com/blog/2010/03/setting-svnignore-from-the-command-line.html here].
+
find . -maxdepth 1 -type f -printf '%T@ %p\0' | sort -r -z -n | awk 'BEGIN { RS="\0"; ORS="\0"; FS="" } NR > 5 { sub("^[0-9]*(.[0-9]*)? ", ""); print }' | xargs -0 rm -f
  
$ svn propset svn:ignore [file|folder] [path]
+
This script stolen from [http://stackoverflow.com/questions/25785/delete-all-but-the-most-recent-x-files-in-bash stackoverflow].
  
Or use a file and apply recursively:
+
Requires GNU find for -printf, GNU sort for -z, GNU awk for "\0" and GNU xargs for -0, but handles files with embedded newlines or spaces.
  
$ svn propset svn:ignore -RF ./svn-ignore-list.txt .
+
== Changing into the script's directory ==
  
= Git =
+
cd "`dirname $0`"
  
== Showing status of working copy ==
+
== Getting the absolute path of a relative path ==
  
  git status
+
  readlink -f ./some/path
  
== Showing repo history ==
+
== Creating a temp directory ==
  
  git log
+
  dir=`mktemp -d` && cd $dir
  
== Showing remote repositories (including 'origin') ==
+
== Reading secret input from stdin ==
  
git remote -v
+
You can read a secret, such as a password, like this:
  
== Handy git aliases ==
+
echo -n "Enter passphrase: "
 +
stty -echo
 +
read passphrase;
 +
stty echo
 +
echo ""
  
Save these to your ~/.gitconfig file.
+
After running the above the secret will be in the $passphrase environment variable.
  
For a nicer view of history than standard 'git log' -- colourful, one-line-per commit, etc:
+
== String replacements in bash ==
  
  graph = !git log --all --graph --color --abbrev-commit --pretty=oneline
+
See the [http://www.tldp.org/LDP/abs/html/string-manipulation.html string manipulation] doco. Basically, to replace first occurrence:
  
To show only the files that have changed, rather than the full line-by-line content:
+
result=${var/find/replace}
  
  dif  = !git diff --name-status
+
To replace all occurrences:
  
= IRC =
+
result=${var//find/replace}
  
== Instructing ChanServ to op an admin ==
+
A practical example, get an ISO date and turn it into a path:
  
  /msg ChanServ op #channel user
+
  date="$(date +%Y-%m-%d)"
 +
work_dir=${date//-//}
  
E.g.
+
== Sending a HEREDOC to a file ==
  
  /msg ChanServ op #gnurc jj5
+
  cat << EOF > /tmp/yourfilehere
 +
These contents will be written to the file.
 +
        This line is indented.
 +
EOF
  
Sub 'op' for 'deop' to remove op privilege.
+
== Bash case/switch statement ==
 +
 
 +
See [http://tldp.org/LDP/Bash-Beginners-Guide/html/sect_07_03.html using case statements], e.g.:
  
= C++ =
+
case $space in
 +
[1-6]*)
 +
  Message="All is quiet."
 +
  ;;
 +
[7-8]*)
 +
  Message="Start thinking about cleaning out some stuff.  There's a partition that is $space % full."
 +
  ;;
 +
9[1-8])
 +
  Message="Better hurry with that new disk...  One partition is $space % full."
 +
  ;;
 +
99)
 +
  Message="I'm drowning here!  There's a partition at $space %!"
 +
  ;;
 +
*)
 +
  Message="I seem to be running with an nonexistent amount of disk space..."
 +
  ;;
 +
esac
  
== C++ books ==
+
== Using dotglob shopt to match dot-files ==
  
=== Books I want ===
+
To enable dot-file matching in globs, set the dotglob shell option:
  
* [http://smile.amazon.com/dp/1785283073 Boost.Asio C++ Network Programming 2ed]
+
$ shopt -s dotglob
* [http://smile.amazon.com/dp/1783986549 Boost.Asio C++ Network Programming Cookbook]
 
  
* [http://www.amazon.com/dp/020170353X Accelerated C++] by Andrew Koening
+
== Stopping a script from running if it previously exited due to error ==
* [http://www.amazon.com/dp/0321334876 Effective C++] by Scott Meyers
 
* [http://www.amazon.com/dp/1491903996 Effective Modern C++] by Scott Meyers
 
* [http://www.amazon.com/dp/020163371X More Effective C++] by Scott Meyers
 
* [http://www.amazon.com/dp/0201749629 Effective STL] by Scott Meyers
 
* [http://www.amazon.com/dp/0201615622 Exceptional C++] by Herb Sutter
 
* [http://www.amazon.com/dp/020170434X More Exceptional C++] by Herb Sutter
 
* [http://www.amazon.com/dp/0201760428 Exceptional C++ Style] by Herb Sutter
 
* [http://www.amazon.com/dp/0321227255 C++ Template Metaprogramming] by David Abrahams
 
* [http://www.amazon.com/dp/059652269X 97 Things Every Software Architect Should Know] by Richard Monson-Haefel
 
* [http://www.amazon.com/dp/9491028022 Introduction to the Boost C++ Libraries; Volume II - Advanced Libraries] by Robert Demming
 
  
=== Books I own ===
+
persistentDataDir=/var/lib/something
 +
alarm() {
 +
  touch $persistentDataDir/alarm
 +
}
 +
trap alarm ERR
 +
[ -f $persistentDataDir/alarm ] && exit 1
  
* [http://www.amazon.com/dp/0321563840 The C++ Programming Language 4ed] by Bjarne Stroustrup
+
== Make sure only one instance of a script is running at a time ==
* [http://smile.amazon.com/dp/9491028022 Introduction to the Boost C++ Libraries; Volume II - Advanced Libraries]
 
* [http://smile.amazon.com/dp/1849514887 Boost C++ Application Development Cookbook]
 
* [http://smile.amazon.com/dp/1782163263 Boost.Asio C++ Network Programming]
 
* [http://www.amazon.com/dp/0321113586 C++ Coding Standards] by Herb Sutter &#x2713;
 
* [http://www.amazon.com/dp/0201704315 Modern C++ Design] by Andrei Alexandrescu &#x2713;
 
* [http://www.amazon.com/dp/0596809484 97 Things Every Programmer Should Know] by Kevlin Henney &#x2713;
 
* [http://www.amazon.com/dp/0321133544 Beyond the C++ Standard Library] by Björn Karlsson &#x2713;
 
* [http://www.amazon.com/dp/9491028014 Introduction to the Boost C++ Libraries; Volume I - Foundations] by Robert Demming &#x2713;
 
* [http://www.amazon.com/dp/0123850037 API Design for C++] by Martin Reddy &#x2713;
 
* [http://www.amazon.com/dp/B00CB23URA Advanced C++ Metaprogramming] by Davide Di Gennaro &#x2713;
 
** Note: the next version of this book is: [http://www.amazon.com/dp/1484210115 Advanced Metaprogramming in Classic C++]
 
* [http://www.amazon.com/dp/1933988770 C++ Concurrency in Action: Practical Multithreading] by Anthony Williams &#x2713;
 
  
=== Books I'm not reading ===
+
ephemeralDataDir=/var/run/something
 +
unlock() {
 +
  rmdir $ephemeralDataDir/lock
 +
}
 +
mkdir $ephemeralDataDir/lock || exit 1;
 +
trap unlock EXIT
  
* [http://www.amazon.com/dp/0321563840 The C++ Programming Language 3ed] by Bjarne Stroustrup &#x2713;
+
= Sed =
** Note: 3ed is obsolete. Buy 4ed (above).
 
  
=== Books I've read ===
+
== Find and replace with sed ==
  
* [http://www.amazon.com/dp/0596004966 C++ Pocket Reference] by Kyle Loudon &#x2713;
+
To update the current file use '-i'. E.g.:
  
== C++ blogs/articles ==
+
sed -i 's/search-text/replace-text/' file
  
* [http://blogs.msdn.com/b/hsutter/ Herb Sutter's MSDN blog]
+
= Awk =
* [http://herbsutter.com/ Herb Sutter's personal blog]
 
* [http://herbsutter.com/gotw/ Herb Sutter's Guru of the Week (GotW)] updated from [http://gotw.ca/gotw/ gotw.ca]
 
  
== C++ performance tips ==
+
== Listing IP addresses in an Apache web log ==
  
* ++c can be faster than c++.
+
awk '/GET \/path\/for\/url/ { print $1 }' /var/log/apache2/access.log | sort | uniq
* use const for everything that you possibly can.
 
* use 'inline' when you need to define a function in a header. Typically only do that if it's small and the increase in code size from inlining is worth the cost to avoid the cost of a function call. For anything except trivially small functions you'll probably need to profile to know if it's worth it.
 
* don't use registers.
 
* const [http://www.gotw.ca/gotw/081.htm rarely affects performance].
 
* debunking a number of [http://www.open-std.org/jtc1/sc22/wg21/docs/TR18015.pdf C++ myths that won't die].
 
* std::sort<> is typically faster than qsort() because it can avoid indirection at runtime.
 
* if you've got parallelisation going on, you may be able to just replace a std::for_each with a parallel equivalent.
 
* read about [http://stackoverflow.com/questions/579887/how-expensive-is-rtti performance cost of RTTI] (Run Time Type Information) and [http://stackoverflow.com/questions/4486609/when-can-compiling-c-without-rtti-cause-problems how to disable it]
 
* don't use dynamic_cast because it is slow (typeid is faster but still relies on RTTI)
 
* prefer unique_ptr to shared_ptr when possible. unique_ptr has less overhead.
 
* [http://sunsite.uakom.sk/sunworldonline/swol-02-1996/swol-02-perf.html Which is better, static or dynamic linking?]
 
* [http://stackoverflow.com/questions/2550281/floating-point-vs-integer-calculations-on-modern-hardware Integer vs Floating-Point performance]
 
  
= systemd =
+
== Printing space-separated field ==
  
[https://en.wikipedia.org/wiki/Systemd systemd] is an init system used in most Linux distributions to bootstrap the user space and manage all processes subsequently.
+
echo 'no no yes no' | awk '{print $3}'
 +
 
 +
== Printing delimited field ==
 +
 
 +
echo 'no:no:yes:no' | awk -F ':' '{print $3}'
 +
 
 +
= Subversion =
  
== Following a service log ==
+
== Setting svn:externals from the command-line ==
  
e.g. for bind9:
+
See [http://beerpla.net/2009/06/20/how-to-properly-set-svn-svnexternals-property-in-svn-command-line/ here].
  
# journalctl -f -u bind9
+
To set an svn:externals from the command-line:
  
or for everything:
+
svn propset svn:externals 'rdfind-php https://www.progclub.org/svn/pcrepo/rdfind.php/branches/0.1' .
 +
svn ci -m 'Adding svn:externals for rdfind-php...'
 +
svn up
  
# journalctl -f
+
Or to use a file:
  
== System status ==
+
svn propset svn:externals -F svn.externals .
  
To see spawned services hierarchy:
+
== Setting svn:ignore from the command line ==
  
# systemctl status
+
See [http://tedone.typepad.com/blog/2010/03/setting-svnignore-from-the-command-line.html here].
  
Or for a specific service e.g.:
+
$ svn propset svn:ignore [file|folder] [path]
  
  # systemctl status networking
+
Or use a file and apply recursively:
 +
 
 +
$ svn propset svn:ignore -RF ./svn-ignore-list.txt .
 +
 
 +
= Git =
 +
 
 +
== Showing status of working copy ==
 +
 
 +
git status
 +
 
 +
== Showing repo history ==
 +
 
 +
git log
 +
 
 +
== Showing remote repositories (including 'origin') ==
 +
 
 +
git remote -v
 +
 
 +
== Handy git aliases ==
 +
 
 +
Save these to your ~/.gitconfig file.
 +
 
 +
For a nicer view of history than standard 'git log' -- colourful, one-line-per commit, etc:
 +
 
 +
  graph = !git log --all --graph --color --abbrev-commit --pretty=oneline
 +
 
 +
To show only the files that have changed, rather than the full line-by-line content:
 +
 
 +
  dif  = !git diff --name-status
 +
 
 +
= IRC =
 +
 
 +
== Instructing ChanServ to op an admin ==
 +
 
 +
/msg ChanServ op #channel user
 +
 
 +
E.g.
 +
 
 +
/msg ChanServ op #gnurc jj5
 +
 
 +
Sub 'op' for 'deop' to remove op privilege.
 +
 
 +
= C++ =
 +
 
 +
== C++ books ==
 +
 
 +
=== Books I want ===
 +
 
 +
* [http://smile.amazon.com/dp/1785283073 Boost.Asio C++ Network Programming 2ed]
 +
* [http://smile.amazon.com/dp/1783986549 Boost.Asio C++ Network Programming Cookbook]
 +
 
 +
* [http://www.amazon.com/dp/020170353X Accelerated C++] by Andrew Koening
 +
* [http://www.amazon.com/dp/0321334876 Effective C++] by Scott Meyers
 +
* [http://www.amazon.com/dp/1491903996 Effective Modern C++] by Scott Meyers
 +
* [http://www.amazon.com/dp/020163371X More Effective C++] by Scott Meyers
 +
* [http://www.amazon.com/dp/0201749629 Effective STL] by Scott Meyers
 +
* [http://www.amazon.com/dp/0201615622 Exceptional C++] by Herb Sutter
 +
* [http://www.amazon.com/dp/020170434X More Exceptional C++] by Herb Sutter
 +
* [http://www.amazon.com/dp/0201760428 Exceptional C++ Style] by Herb Sutter
 +
* [http://www.amazon.com/dp/0321227255 C++ Template Metaprogramming] by David Abrahams
 +
* [http://www.amazon.com/dp/059652269X 97 Things Every Software Architect Should Know] by Richard Monson-Haefel
 +
* [http://www.amazon.com/dp/9491028022 Introduction to the Boost C++ Libraries; Volume II - Advanced Libraries] by Robert Demming
 +
 
 +
=== Books I own ===
 +
 
 +
* [http://www.amazon.com/dp/0321563840 The C++ Programming Language 4ed] by Bjarne Stroustrup
 +
* [http://smile.amazon.com/dp/9491028022 Introduction to the Boost C++ Libraries; Volume II - Advanced Libraries]
 +
* [http://smile.amazon.com/dp/1849514887 Boost C++ Application Development Cookbook]
 +
* [http://smile.amazon.com/dp/1782163263 Boost.Asio C++ Network Programming]
 +
* [http://www.amazon.com/dp/0321113586 C++ Coding Standards] by Herb Sutter &#x2713;
 +
* [http://www.amazon.com/dp/0201704315 Modern C++ Design] by Andrei Alexandrescu &#x2713;
 +
* [http://www.amazon.com/dp/0596809484 97 Things Every Programmer Should Know] by Kevlin Henney &#x2713;
 +
* [http://www.amazon.com/dp/0321133544 Beyond the C++ Standard Library] by Björn Karlsson &#x2713;
 +
* [http://www.amazon.com/dp/9491028014 Introduction to the Boost C++ Libraries; Volume I - Foundations] by Robert Demming &#x2713;
 +
* [http://www.amazon.com/dp/0123850037 API Design for C++] by Martin Reddy &#x2713;
 +
* [http://www.amazon.com/dp/B00CB23URA Advanced C++ Metaprogramming] by Davide Di Gennaro &#x2713;
 +
** Note: the next version of this book is: [http://www.amazon.com/dp/1484210115 Advanced Metaprogramming in Classic C++]
 +
* [http://www.amazon.com/dp/1933988770 C++ Concurrency in Action: Practical Multithreading] by Anthony Williams &#x2713;
 +
 
 +
=== Books I'm not reading ===
 +
 
 +
* [http://www.amazon.com/dp/0321563840 The C++ Programming Language 3ed] by Bjarne Stroustrup &#x2713;
 +
** Note: 3ed is obsolete. Buy 4ed (above).
 +
 
 +
=== Books I've read ===
 +
 
 +
* [http://www.amazon.com/dp/0596004966 C++ Pocket Reference] by Kyle Loudon &#x2713;
 +
 
 +
== C++ blogs/articles ==
 +
 
 +
* [http://blogs.msdn.com/b/hsutter/ Herb Sutter's MSDN blog]
 +
* [http://herbsutter.com/ Herb Sutter's personal blog]
 +
* [http://herbsutter.com/gotw/ Herb Sutter's Guru of the Week (GotW)] updated from [http://gotw.ca/gotw/ gotw.ca]
 +
 
 +
== C++ performance tips ==
 +
 
 +
* ++c can be faster than c++.
 +
* use const for everything that you possibly can.
 +
* use 'inline' when you need to define a function in a header. Typically only do that if it's small and the increase in code size from inlining is worth the cost to avoid the cost of a function call. For anything except trivially small functions you'll probably need to profile to know if it's worth it.
 +
* don't use registers.
 +
* const [http://www.gotw.ca/gotw/081.htm rarely affects performance].
 +
* debunking a number of [http://www.open-std.org/jtc1/sc22/wg21/docs/TR18015.pdf C++ myths that won't die].
 +
* std::sort<> is typically faster than qsort() because it can avoid indirection at runtime.
 +
* if you've got parallelisation going on, you may be able to just replace a std::for_each with a parallel equivalent.
 +
* read about [http://stackoverflow.com/questions/579887/how-expensive-is-rtti performance cost of RTTI] (Run Time Type Information) and [http://stackoverflow.com/questions/4486609/when-can-compiling-c-without-rtti-cause-problems how to disable it]
 +
* don't use dynamic_cast because it is slow (typeid is faster but still relies on RTTI)
 +
* prefer unique_ptr to shared_ptr when possible. unique_ptr has less overhead.
 +
* [http://sunsite.uakom.sk/sunworldonline/swol-02-1996/swol-02-perf.html Which is better, static or dynamic linking?]
 +
* [http://stackoverflow.com/questions/2550281/floating-point-vs-integer-calculations-on-modern-hardware Integer vs Floating-Point performance]
 +
 
 +
= systemd =
 +
 
 +
[https://en.wikipedia.org/wiki/Systemd systemd] is an init system used in most Linux distributions to bootstrap the user space and manage all processes subsequently.
 +
 
 +
== Following a service log ==
 +
 
 +
e.g. for bind9:
 +
 
 +
# journalctl -f -u bind9
 +
 
 +
or for everything:
 +
 
 +
# journalctl -f
 +
 
 +
== System status ==
 +
 
 +
To see spawned services hierarchy:
 +
 
 +
# systemctl status
 +
 
 +
Or for a specific service e.g.:
 +
 
 +
  # systemctl status networking
 +
 
 +
= SaltStack =
 +
 
 +
== Running a command on specified minions ==
 +
 
 +
salt 'host' cmd.run 'update-locale'
 +
 
 +
== Running a command on all minions ==
 +
 
 +
salt '*' cmd.run 'update-locale'
 +
 
 +
== Listing active jobs ==
 +
 
 +
salt-run jobs.active
 +
 
 +
== Listing available grains ==
 +
 
 +
salt 'example' grains.items
 +
 
 +
== Listing available pillar ==
 +
 
 +
salt 'example' pillar.items
 +
 
 +
== Reporting a grain value ==
 +
 
 +
e.g. for the 'mem_total' grain:
 +
 
 +
salt '*' grains.item mem_total
 +
 
 +
= KDE =
 +
 
 +
== Running user login script (X11/XOrg/XWindows) ==
 +
 
 +
A way to run user login scripts which works for KDE Plasma (and apparently other [https://en.wikipedia.org/wiki/X.Org_Server X.Org Server X Window System] environments) is to create a *.desktop file in ~/.config/autostart/. For example I have a ~/.config/autostart/ssh-add.desktop file with the following contents to register my SSH key in the SSH Agent:
 +
 
 +
[Desktop Entry]
 +
Type=Application
 +
Name=ssh-add
 +
Comment=Adds my private key to my session.
 +
Exec=/usr/bin/konsole -e 'ssh-add /home/$USER/.ssh/id_rsa'
  
= SaltStack =
+
= VirtualBox =
  
== Running a command on specified minions ==
+
See [https://askubuntu.com/questions/19430/mount-a-virtualbox-drive-image-vdi/50290#50290 here]:
  
salt 'host' cmd.run 'update-locale'
+
Install qemu if necessary:
  
== Running a command on all minions ==
+
# apt install qemu
  
salt '*' cmd.run 'update-locale'
+
Then you'll need to load the network block device module:
  
== Listing active jobs ==
+
# rmmod nbd
 +
# modprobe nbd max_part=16
  
salt-run jobs.active
+
Attach the .vdi image to one of the nbd you just created:
  
== Listing available grains ==
+
# qemu-nbd -c /dev/nbd0 drive.vdi
  
salt 'example' grains.items
+
Now you will get a /dev/nbd0 block device, along with several /dev/nbd0p* partition device nodes.
  
== Listing available pillar ==
+
# mount /dev/nbd0p1 /mnt
  
salt 'example' pillar.items
+
Once you are done, unmount everything and disconnect the device:
  
== Reporting a grain value ==
+
  # qemu-nbd -d /dev/nbd0
 
 
e.g. for the 'mem_total' grain:
 
 
 
  salt '*' grains.item mem_total
 
 
 
= KDE =
 
 
 
== Running user login script (X11/XOrg/XWindows) ==
 
 
 
A way to run user login scripts which works for KDE Plasma (and apparently other [https://en.wikipedia.org/wiki/X.Org_Server X.Org Server X Window System] environments) is to create a *.desktop file in ~/.config/autostart/. For example I have a ~/.config/autostart/ssh-add.desktop file with the following contents to register my SSH key in the SSH Agent:
 
 
 
[Desktop Entry]
 
Type=Application
 
Name=ssh-add
 
Comment=Adds my private key to my session.
 
Exec=/usr/bin/konsole -e 'ssh-add /home/$USER/.ssh/id_rsa'
 

Revision as of 10:20, 10 October 2019

Hi there, I'm John. I just wanted a page where I could document various Linux things that I bump into. This is that page. Thank you ProgClub. :)

Note: I have some other disorganised notes on UNIX, which include a few tips for MacOS. I also have some tips for OS X.

System

Determining which Debian/Ubuntu release your are running

$ lsb_release -r

Or for more information:

$ lsb_release

Determining which Linux/Unix you are running

$ uname

Or,

$ uname -mrs

Or,

$ uname -a

Configuring system swappiness

Swappiness is a number between 0 and 100 that regulates how much the system uses the swap file. I like setting this value to 0 to keep my apps as responsive as possible. Create a file /etc/sysctl.d/local.conf and add this line:

vm.swappiness = 0

If you want to set the value for the current session only:

echo 0 > /proc/sys/vm/swappiness

Hardware information

For information about the hardware attached to your system, check out:

# lshw

And for CPUs:

# lscpu

And for PCI devices:

# lspci

And for DMI info:

# dmidecode

Or the grand daddy of them all:

# hwinfo

There's also inxi, e.g.:

$ inxi -b

System:    Host: tact Kernel: 4.9.0-4-amd64 x86_64 (64 bit) Desktop: KDE Plasma 5.8.6
           Distro: Debian GNU/Linux 9 (stretch)
Machine:   Device: desktop Mobo: ASUSTeK model: STRIX Z270F GAMING v: Rev 1.xx
           UEFI [Legacy]: American Megatrends v: 0906 date: 03/22/2017
CPU:       Quad core Intel Core i7-7700K (-HT-MCP-) speed/max: 799/4600 MHz
Graphics:  Card: Intel Device 5912
           Display Server: X.Org 1.19.2 drivers: modesetting (unloaded: fbdev,vesa)
           Resolution: 1920x1080@60.00hz, 1920x1080@60.00hz
           GLX Renderer: Mesa DRI Intel Kabylake GT2 GLX Version: 3.0 Mesa 13.0.6
Network:   Card: Intel Ethernet Connection (2) I219-V driver: e1000e
Drives:    HDD Total Size: 13026.6GB (42.0% used)
RAID:      Devices: 1: /dev/md1 2: /dev/md0
Info:      Processes: 355 Uptime: 11 days Memory: 21198.3/32043.3MB Client: Shell (bash) inxi: 2.3.5

Power

Reporting on PowerShield DEFENDER UPS status

To see the status of the PowerShield DEFENDER systems on John's LAN:

$ upsc defender

E.g.:

jj5@orac:~$ upsc defender
Init SSL without certificate database
battery.charge: 100
battery.voltage: 27.40
battery.voltage.high: 26.00
battery.voltage.low: 20.80
battery.voltage.nominal: 24.0
device.type: ups
driver.name: blazer_usb
driver.parameter.pollinterval: 2
driver.parameter.port: auto
driver.parameter.synchronous: no
driver.version: 2.7.4
driver.version.internal: 0.12
input.current.nominal: 5.0
input.frequency: 50.1
input.frequency.nominal: 50
input.voltage: 242.6
input.voltage.fault: 242.6
input.voltage.nominal: 240
output.voltage: 242.6
ups.beeper.status: disabled
ups.delay.shutdown: 30
ups.delay.start: 180
ups.load: 14
ups.productid: 5161
ups.status: OL
ups.type: offline / line interactive
ups.vendorid: 0665

Run commands on PowerShield DEFENDER UPS batteries

You can run "instant commands" using the upscmd command.

We use the 'beeper.toggle' instant command in our Salt Stack config to disable the beeper, see e.g.:

diligence:/srv/salt/conf/app/defender-1200.sls

To see "instant commands" supported by the PowerShield DEFENDER:

$ upscmd -l defender

E.g.:

jj5@orac:~$ upscmd -l defender
Instant commands supported on UPS [defender]:

beeper.toggle - Toggle the UPS beeper
load.off - Turn off the load immediately
load.on - Turn on the load immediately
shutdown.return - Turn off the load and return when power is back
shutdown.stayoff - Turn off the load and remain off
shutdown.stop - Stop a shutdown in progress
test.battery.start - Start a battery test
test.battery.start.deep - Start a deep battery test
test.battery.start.quick - Start a quick battery test
test.battery.stop - Stop the battery test

Environment

Configuring vim as your editor

Sometimes all you need is:

$ export EDITOR=/usr/bin/vim

Which works for svn, for example. Add it to your ~/.profile file to have it set for all login sessions.

Other times you need to run

# update-alternatives --config editor

And then select vim from the list. This is what you do to configure your visudo editor.

Configuring your locale

$ sudo /usr/sbin/locale-gen en_AU.UTF-8
$ sudo /usr/sbin/update-locale LANG=en_AU.UTF-8

User and group management

Adding a user

To add a new user on a linux system:

# useradd username
# passwd username

To have the home directory created from '/etc/skel' use the 'adduser' script instead:

# adduser username

Adding a user to a group

To add an existing user to an existing group:

# gpasswd -a username group

e.g. to add user 'jj5' to the 'sudo' group:

# gpasswd -a jj5 sudo

Alternatively you can use adduser, passing the username and group:

# adduser username group

e.g. to add user 'sclaughl' to the 'staff' group:

# adduser sclaughl staff

Disabling a user account

You can disable a user account with:

# passwd -l user

Note: that's a lower-case L, not a one.

Enabling a disabled user account

To can re-enable a locked user account with:

# passwd -u user

Finding which user you are logged in as

To determine which user you are running as enter the command:

$ whoami

Finding which groups you are a member of

To find which groups you are a member of:

$ groups

or

$ groups username

Where 'username' is the username of the user you are querying, e.g.:

$ groups jj5

Finding who else is logged in to the system

To see who else is logged in,

$ who

Running a command as a particular user

To run "svn update" as the user www-data:

$ sudo su -c "svn update" www-data

Reporting user and group info for the current user

$ id

Memory management

Checking available memory

To report memory statistics in megabytes:

$ free -m

Check for swap thrashing

Check your virtual memory status with vmstat:

$ vmstat

Video/display management

Viewing EDID data for attached monitor

To view EDID data for an attached monitor (requires the edid-decode package):

$ cd /sys/class/drm
$ ls
$ cd card0-HDMI-A-1
$ edid-decode edid

Process management

Using 'top' for dynamic resource usage reporting

To run top:

$ top

See 15 Practical Linux Top Command Examples for some hints on usage.

To see usage for a specific user run e.g.:

$ top -u jj5

To see full command-line press 'c'.

When you're in 'top' you can:

  • press '1' (one) to toggle CPU aggregation
  • press < and > to change the sort column

Changing memory reporting in 'top'

To run top:

$ top

Press 'E' to switch between top memory units (KiB, MiB, GiB, etc.)

Press 'e' to switch between bottom memory units (KiB, MiB, GiB, etc.)

Press 'M' to sort by memory utilisation.

Press 'm' to switch between various display modes.

Disk management

Listing disk drives

# fdisk -l

(That's an L for "list")

Checking available disk space

$ df -h

Getting disk information

# lsblk

And

# cat /proc/partitions

Or the Grand Daddy of them all:

# lshw -class disk

(Requires the lshw package.)

Getting partition UUID and file-system type

# blkid

Checking for SSD vs magnetic disk

# cat /sys/block/sda/queue/rotational

Will be 0 for SSD and 1 for magnetic.

Monitoring a ZFS server

So some commands I run to keep an eye on my new ZFS servers:

# top
# iotop
# nethogs
# watch free -h
# watch slabtop -o
# slabtop
# watch cat /proc/meminfo
# perf top
# watch "df -h | grep -v -e tmpfs -e udev -e by-uuid"
# watch zpool iostat -v
# zpool iostat -v 2
# watch 'zpool list; echo; zfs list'
# watch zfs get compressratio -o all
# watch cat /proc/spl/kstat/zfs/arcstats

If you have a scrub or resilvering in progress you can report on progress with:

# watch zpool status -v

You can poke about in internals, e.g.:

# cat /proc/spl/kstat/zfs/arcstats
root@orac:/sys/module/zfs/parameters# tail *

You can report on property values with e.g.:

# zfs get all data

If you want to get funky:

# cd /tmp
# perf record -ag #(Ctrl+C after ~15 seconds)
# perf report --stdio

You can search for ZFS files like e.g. this:

root@orac:/# find / -name '*zfs*' -or -name '*zpool*'

You can report history of a zpool:

# zpool history $poolname

You can get a report on the dedup tables:

# zpool status -D $poolname

Or more detailed dedup table info:

# zdb -DDD $poolname

Note in the output see here for details, basically:

Abbr Description
LSIZE logical size (in memory)
PSIZE physical size
DSIZE size on disk
refcnt reference count

Measure data throughput

Use the 'pv' command from the 'pv' package, e.g.:

# cat /dev/sda | pv | cat > /dev/null

Or for ZFS:

# zfs send data/example | pv | cat > /dev/null

Using Smartctl, Smartd and Hddtemp on Debian

For notes on using smartctl see Using Smartctl, Smartd and Hddtemp on Debian.

Report hard disk usage

So you might want to know how much data a process reads or writes to a hard disk. You can monitor process total disk utilisation with the 'iotop' command. Run 'iotop' and then press 'a' for --accumulated.

Report hard disk temperatures

E.g.

# hddtemp /dev/sd[a-e]

Monitoring disk I/O

There's an app for that! iotop.

Using iotop, top for disks

# iotop -oPa

Monitor disk I/O for performance issues

# watch iostat

Or e.g.

# watch iostat -xd /dev/sd[abc]

Monitoring a system

Simple ZFS monitoring

# watch iostat
# iotop
# zpool iostat -v 5
# watch 'hddtemp /dev/sd[a-e]; echo; zpool list; echo; zfs list'
# nethogs
# top

File management

Listing only directories

$ ls -l | egrep '^d'

Listing only files

$ ls -l | egrep -v '^d'

Listing hidden files

$ ls -al .[!.]*

Creating a symbolic link

$ ln -s /path/to/target link-name

Creating a hard-link

$ ln /path/to/target file-name

Changing the owner of a file

$ chown user:group <files>

E.g.

$ chown jj5:staff README
$ chown root:root *

To apply recursively into sub-directories use -R,

$ chown -R root:root /etc/*

Changing file permissions

Object codes
User Group Other
u g o
Permission codes
Read Write Exectue
r w x
4 2 1
Numeric codes
0 None
1 Execute
2 Write
3 Write, Execute
4 Read
5 Read, Execute
6 Read, Write
7 Read, Write, Execute

See Numeric Mode in Action.

$ chmod <user numeric code><group numeric code><other numeric code> <files>
$ chmod <object codes>+|-<permission codes> <files>

E.g.

$ chmod 600 my-private-file
$ chmod go-rwx my-private-file
$ chmod u+rw my-private-file
$ chmod +x my-script

Updating config files

If you get given a new config file called new.conf and you want to integrate it with your old config file old.conf then:

$ cp old.conf updated.conf
$ merge -A updated.conf new.conf old.conf

Then go through and edit updated.conf resolving all the merge errors, picking and choosing what to update and what to keep. When you're done copy updated.conf to old.conf so it becomes the new config file.

The merge program is a part of the RCS package. If you don't have it:

$ sudo apt-get install rcs

Listing open files

Use lsof to list open files. E.g.:

# lsof

See man lsof for options.

List permissions on a whole directory path

E.g.:

$ namei -om /home/jj5/workspace

Outputs:

f: /home/jj5/workspace/
 drwxr-xr-x root root /
 drwxr-xr-x root root home
 drwxr-xr-x jj5  jj5  jj5
 drwxr-xr-x jj5  jj5  workspace

Counting non-blank lines in a file

E.g.:

$ cat foo.c | sed '/^\s*$/d' | wc -l

Cloning one directory to another with rsync

E.g.:

rsync --acls --xattrs --stats --human-readable --recursive --del --force --times --links --hard-links --executability --numeric-ids --owner --group --perms --sparse --compress-level=0 /data/source/ hostname:/data/target/

Counting number of files in current directory and all subdirectories

$ ls -AlhR . | egrep '^-' | wc -l

Counting number of directories in current directory and all subdirectories

$ ls -AlhR . | egrep '^d' | wc -l

Compression

How to use pigz with tar

See here:

$ tar cf - paths-to-archive | pigz --best -p 8 > archive.tgz

Best parallel compression with pigz

$ pigz --best

Best parallel compression with xz

$ xz -9e -T 0

Reporting compression ratios with xz

e.g.

root@love:/data/image/archive# xz -l *
Strms  Blocks   Compressed Uncompressed  Ratio  Check   Filename
    1       3    372.2 MiB    442.3 MiB  0.841  CRC64   1999.txz
    1      29  5,281.3 MiB  5,542.5 MiB  0.953  CRC64   2001.txz
    1      11  1,364.3 MiB  2,084.3 MiB  0.655  CRC64   2002.txz
    1       9    568.5 MiB  1,660.2 MiB  0.342  CRC64   2003.txz
    1     639     66.8 GiB    119.6 GiB  0.558  CRC64   2004.txz
    1     313     12.7 GiB     58.6 GiB  0.217  CRC64   2005.txz
    1     414     35.0 GiB     77.4 GiB  0.452  CRC64   2006.txz
    1     485     44.5 GiB     90.9 GiB  0.490  CRC64   2007.txz
    1   1,690    150.0 GiB    316.8 GiB  0.473  CRC64   2008.txz
    1       3    457.9 MiB    526.0 MiB  0.871  CRC64   2009.txz
    1     168     27.3 GiB     31.4 GiB  0.868  CRC64   2010.txz
    1       4    477.1 MiB    702.8 MiB  0.679  CRC64   2011.txz
-------------------------------------------------------------------------------
   12   3,768    344.6 GiB    705.5 GiB  0.488  CRC64   12 files

Symbolic-link management

== Data used by sym-linked files:

This will de-reference the sym-links in the current directory and tell you how much data the files pointed to by the sym-links are using:

jj5@tact:/data/backup/unity/latest$ du -hD * | sort -h

File searching

Finding a file with a particular name

$ find -iname "*some-part-of-the-file-name*"

Will start searching from the current directory, so maybe

$ cd /

first. For a case-sensitive search:

$ find -name "*eXaCT CaSE*"

Finding a file with particular content

To search in /etc/ for a file with particular content:

$ grep -R "search-string" /etc/*

To search the current directory for *.cs files containing the word "Up":

$ find . -name '*.cs' -exec grep --color=auto -H Up {} \;

Finding a list of files with particular content

E.g. to find all the files with the word 'creativity':

$ grep -R creativity . | sed 's/:/ /' | awk '{ print $1 }' | sort | uniq

Using the locate command to find files

$ locate part-of-filename

E.g.

$ locate texvc

Updating locate command's database

# updatedb

Job control

Stopping a running process

Press Ctrl+Z to stop a running process.

Listing current jobs and their status

$ jobs

Resuming a stopped job in the backgroud

To resume a stopped process in the background

$ bg %1

where '1' is the job number reported by bash when you pressed Ctrl+Z (or ran 'jobs').

Resuming a stopped job in the foreground

To resume a stopped process in the foreground

$ fg %1

where '1' is the job number reported by bash when you pressed Ctrl+Z (or ran 'jobs').

Killing a stopped job

To kill a job

$ kill %1

where '1' is the job number reported by bash when you pressed Ctrl+Z (or ran 'jobs').

Periodically run a program and watch its output

$ watch /your/command

Debian/Ubuntu package management

Also see Where "is" it? on the Debian Wiki.

configuring debconf

# dpkg-reconfigure debconf 

Set priority to low to get asked detailed questions.

Showing list of installed packages

# dpkg --get-selections

Searching for installed package

# dpkg --get-selections | grep package-name

or

# aptitude search package-name

Showing which files are installed as part of a package

# dpkg -L package-name

Installing a package

# apt-get install package-name

Uninstalling a package

# apt-get remove package-name

Showing system architecture

$ dpkg --print-architecture

Showing which package a file belongs to

$ which echo
/bin/echo
$ dpkg -S /bin/echo
coreutils: /bin/echo
$ dpkg -l | grep coreutils
ii  coreutils                         6.10-6                   The GNU core utilities

Showing package information

$ apt-cache showpkg coreutils

Or for even more information:

$ apt-cache show coreutils

List all installed packages with package version info

dpkg-query -l

Reporting which version of a package is installed

$ dpkg -l | grep package-name

E.g.:

root@hope:~/letsencrypt# dpkg -l | grep augeas
ii  augeas-lenses                   0.7.0-1ubuntu1                 Set of lenses needed by libaugeas0 to parse 
ii  libaugeas0                      0.7.0-1ubuntu1                 The augeas configuration editing library and

Comprehensive upgrade

Try the following:

# apt-get update
# apt-get dist-upgrade
# apt-get autoremove
# apt-get remove $(deborphan)
# update-flashplugin-nonfree --install

Searching all available packages

$ apt-cache search . | sort -d | less

Networking

net-tools vs iproute2

The older 'net-tools' package has been replaced with 'iproute2' e.g. in stretch.

legacy net-tools commands iproute2 replacement commands
arp ip n (ip neighbor)
ifconfig ip a (ip addr), ip link, ip -s (ip -stats)
iptunnel ip tunnel
iwconfig iw
nameif ip link, ifrename
netstat ss, ip route (for netstat-r), ip -s link (for netstat -i), ip maddr (for netstat-g)
route ip r (ip route)

Restart networking

For servers:

# service networking restart

For desktops:

# service network-manager restart

Pinging with particular packet size

$ ping -M do -s <packet size in bytes> <host>

E.g.

$ ping -M do -s 1400 charity.progclub.org

Setting MSS for a particular IP address on a particular interface

# ip route add <host> dev <interface> advmss <packet size>

E.g.

# ip route add 10.0.0.1 dev eth0 advmss 1400

Dropping configured MMS for a particular IP address

# ip route flush <host>

E.g.

# ip route flush 10.0.0.1

Listing open ports and socket information

Including which process is listening on which port.

# netstat -tulpn

Or use the 'ss' command:

# ss -s
# ss -l
# ss -pl
# ss -o state established '( dport = :smtp or sport = :smtp )'

Listing open IPv4 connections

# lsof -Pnl +M -i4

You might need to install the lsof package:

# apt-get install lsof

Query for DNS MX record

$ nslookup
> server 127.0.0.1
> set q=mx
> mail.blackbrick.com

Query for DNS SOA record

$ dig @ns2.staticmagic.net -t SOA staticmagic.net

Using nmap to list open ports on remote host

To check the 1,000 most common ports:

# nmap server.example.com

Or for a specific port range (e.g. 101 to 102):

# nmap -p 101-102 server.example.com

Or for all ports (1 to 65,535):

# nmap -p- server.example.com

Network monitoring

See here for details. Basically:

  1. Overall bandwidth: nload, bmon, slurm, bwm-ng, cbm, speedometer, netload
  2. Overall bandwidth (batch style output): vnstat, ifstat, dstat, collectl
  3. Bandwidth per socket connection: iftop, iptraf, tcptrack, pktstat, netwatch, trafshow
  4. Bandwidth per process: nethogs

nload

You can watch network traffic in real-time with nload:

# nload -u M

Reporting network (NIC) speed

From here:

# dmesg | grep eth0
# mii-tool -v eth0
# ethtool eth0

Note: use ifconfig to get device name.

Path MTU discovery

To do a Path MTU Discovery, from the iputils-tracepath package:

# tracepath host.example.com

Listing available Ethernet devices

To see a list of NICs available on the host:

$ cat /proc/net/dev

Also

$ ip link

59 Linux Networking commands and scripts

See 59 Linux Networking commands and scripts.

Links

IPTables

Applying firewall rules

For configuration info see this article.

$ sudo vim /etc/iptables.test.rules
$ sudo /sbin/iptables -F
$ sudo /sbin/iptables-restore < /etc/iptables.test.rules
$ sudo iptables -L
$ sudo -s
# iptables-save > /etc/iptables.up.rules
# exit

ufw

Denying hosts with ufw

See denying hosts with ufw.

Bind9

Viewing Bind9 querylog

$ sudo rndc querylog
$ tail -f /var/log/syslog

IPSec

Disabling IPSec

# setkey -FP

OpenSSL

Debugging IMAPS with OpenSSL

# openssl s_client -connect localhost:993
> a1 LOGIN username@host password
> a2 LOGOUT

Debugging HTTPS with OpenSSL

$ openssl s_client -connect www.example.com:443
GET /example.html HTTP/1.1
host: www.example.com

Links

Pluggable Authentication Modules (PAM)

Links

SSH

Configuring SSH key login

On the client machine generate a key-pair (if necessary, check for existing ~/.ssh/id_rsa.pub):

$ ssh-keygen -t rsa

Copy the public key from the client to the server:

$ scp ~/.ssh/id_rsa.pub user@example.org:

Configure the authorized keys on the server:

$ ssh user@example.org
$ mkdir ~/.ssh
$ chmod go-w .ssh
$ cat ~/id_rsa.pub >> ~/.ssh/authorized_keys
$ chmod 600 ~/.ssh/authorized_keys
$ rm ~/id_rsa.pub

Tunneling over SSH

For example, connecting a remote MySQL server to the localhost:

$ ssh -L 3306:localhost:3306 jselliot@ssh.progsoc.org

If the machine you want to connect to is not the localhost of the machine you're ssh'ing to,

 $ ssh -L 3306:muspell.progsoc.uts.edu.au:3306 ssh.progsoc.uts.edu.au

The -L stanza is localport:remotehost:remoteport where localport is a port on your machine, forwarded to remoteport on remotehost.

Tunneling over SSH with PuTTY

See Connecting to the MySQL database remotely (via an SSH Tunnel)

  • run putty.exe
  • Connection -> SSH -> Tunnels
    • Port forwarding: source port to 3306
    • destination: 127.0.0.1:3306
    • check Local
    • click Add

Enabling verbose SSH logging

To see what's going on with your ssh connections,

$ ssh -v user@host

Or

$ ssh -vv user@host

Unlocking SSH key for session

jj5@orac:~/.config/autostart$ cat ssh-add.desktop 
[Desktop Entry]
Type=Application
Name=ssh-add
Comment=Adds my private key to my session.
Exec=/usr/bin/konsole -e 'ssh-add /home/$USER/.ssh/id_rsa'

Links

Standard IO

cat EOF

$ cat > output <<EOF
> text
> EOF
$ cat output
text

Script

Creating a session log with script

$ script -t 2> timing

The session log is in the file 'typescript' and the timing data is in 'timing'.

Replaying a scripted session

$ scriptreplay timing

Uses the default file 'typescript' and the 'timing' file as specified.

Screen

Creating a new screen or reconnecting to a detached screen

$ screen -R

Detaching a screen

$ screen -D

Reconnecting to screen

$ screen -D
$ screen -R

I have a script in ~/bin/reconnect like so,

#!/bin/bash
screen -D
screen -R

This will detach your last screen, and reconnect it on the current terminal.

Scrolling in screen

See How to scroll in GNU Screen. Basically press Ctrl+A ESC then use Page Up and Page Down. Press ESC again to exit copy mode. As usual you can use Ctrl+[ in place of ESC.

Vim

First, why Vim?

Read Why, oh WHY, do those #?@! nutheads use vi?

Visual modes

Use 'v' for visual mode, 'V' for visual line mode and Ctrl+V for visual block mode.

Configuring spaces instead of tabs

I use two spaces instead of tabs. To configure, edit your .vimrc file:

$ vim ~/.vimrc

and include the following lines:

set tabstop=2
set shiftwidth=2
set expandtab

Configuring syntax highlighting

See here.

Use:

:syntax on

to turn on syntax highlighting.

Use:

:syntax off

to turn off syntax highlighting.

To always use syntax highlighting:

$ vim ~/.vimrc

and add:

syntax on

To get a list of supported colour schemes open vim and type:

:colorscheme[space][Ctrl+D]

To always use a particular colorscheme edit ~/.vimrc and add (for example):

colorscheme desert

Inserting a TAB character when expandtab is on

The problem here is that you have configured vim to insert spaces, but for a particular file (e.g. a Makefile) you need to insert a character.

Press Ctrl+V TAB to insert a literal tab character.

Or you can disable tab expansion altogether with:

:set expandtab!

Changing 2 space indent to 4 space indent (e.g. for python files)

:%s/^\s*/&&/g

For more information see here.

Recording and replaying a macro

To record a macro press 'q' and then a number between 1 and 9. E.g. press "q1". The macro is now recording. When you've finished issuing your commands press 'q' again to finish recording. To replay a macro press '@' followed by the number of the macro. That is, if you pressed "q1" to record the macro, press "@1" to replay the macro. To replay the last macro again press "@@".

Deleting to end of line

d$

Deleting to beginning of line

d^

Finding text

To search forward for "text":

/text

To search backward for "text":

?text

To repeat the last search in a forward direction press 'n', or to search again backwards press 'N'.

Finding and replacing text

To replace the first instance of "search" on the current line with "destroy":

:s/search/destroy/

To replace all instances of "search" on the current line with "destroy":

:s/search/destroy/g

To replace all instances of "search" on lines 13 to 37 with "destroy":

:13,37 s/search/destroy/g

To replace all instances of "search" in the entire file with "destroy":

:%s/search/destroy/g

Changing DOS/Windows line-endings (CRLF) to Unix line-endings

To set the line-ending to Unix line endings run the command:

:setlocal ff=unix

More information on managing file formats available here.

Disabling auto-indent etc. to paste from clipboard

To disable smart indenting when you're going to paste in text:

:set paste

To turn it off again:

:set nopaste

There's more info in this article: Toggle auto-indenting for code paste

Positioning windows

Use -o for horizontal split, e.g.:

vim -o a.txt b.txt

Use -O for vertical split, e.g.:

vim -o a.txt b.txt

Use ^W to navigate windows then use directional keys h, j, k, l, etc.

Use ^W and < or > to resize windows.

To indent a block of text in Vim

Use the > command. E.g. to indent five lines:

5 > >

Press . (dot) to keep indenting.

Or inside a block (e.g. curly brace, HTML/XML element, etc.) you can put your cursor in the element on on the curly brace and then:

> %

See here for more.

Open a file in a new window/tab

To open a file on the left hand side:

:vert new filename.ext

Note: ':vnew filename.ext' and ':vsp filename.ext' also work.

To open a file at the top:

:new filename.ext

See here for more.

Explore files in Vim

Enter:

:Explore

Switch between Vim tabs

Use gt and gT.

Switch between Vim windows

To toggle between open windows use:

Ctrl+W W

To move in a direction use:

Ctrl+W h/j/k/l

See here for more.

Insert block comment in Vim

See here for line-commenting.

So it's:

  1. Ctrl+V (Note: not Shift+V!)
  2. Up/Down to select rows
  3. Shift+I
  4. Enter your text, e.g. '#' or '//'
  5. Ctrl+[ (or 'Esc')

Navigate to matching tag

To navigate to the matching beginning or end tag use '%'.

You can also use e.g. '[{' to match the previous '{', or e.g. '])' to match the next ')'.

Auto-format HTML tags

Stolen from here.

  1. first join all the lines - ggVGgJ
  2. Now break tags to new lines - :%s/>\s*</>\r</g
  3. Now set filetype - :set ft=html (you can do this before too)
  4. Now Indent - ggVG=

Links

Write

Talking to other users on the system

write is a unix command for talking to other users on the system. To use write:

1. SSH to <username>@<hostname> and login with your username and password.

2. Issue the following command to find out who is logged onto the system:

$ who

3. Issue the following command to talk to a specific user:

$ write <username>

4. Enter the message you'd like to send the user, followed by Ctrl+C to send. Press Ctrl+D to cancel.

Date

Reporting the time on the server

$ date

Reporting UTC time

$ date --utc

Getting the date in yyyy-MM-dd-hhmmss format

$ date="`date +%F-%H%M%S`"

Getting the year in four digits

$ year="`date +%Y`"

Getting the month in two digits

$ month="`date +%m`"

Getting the day of the month in two digits

$ day="`date +%d`"

Getting yesterday's date

$ date --date='1 day ago' +%Y-%m-%d

Converting Unix time (seconds since epoch)

For timestamp '1501370200':

$ date -d @1501370200 +%F-%H%M%S

Running timedatectl from systemd

There's a new command bundled with systmed:

# timedatectl

It reports on (and controls) how the system time is configured.

MySQL

Run mysql without authentication/authorisation

# service mysql stop
# mysqld_safe --skip-grant-tables &

Then you can connect without a password, e.g.:

# mysql -u root mysql

To stop the unauthenticated service:

# mysqladmin shutdown

Then restart a normal service:

# service mysql start

Logging all database queries

# vim /etc/mysql/my.cnf

In the [mysqld] section add:

log=/tmp/mysql.log

Then:

# service mysql restart

Watch the log with:

# tail -f /tmp/mysql.log

Dumping a MySQL database

You can dump the database into a file using:

$ mysqldump -h hostname -u user --password=password databasename > filename

Loading a MySQL database from a dump file

You can create a database using:

$ echo create database databasename | mysql -h hostname -u user -p

You can restore a database using:

$ mysql -h hostname -u user --password=password databasename < filename

Creating a MySQL user

# mysql -h localhost -u root --password=<password>
mysql> create user 'username'@'localhost' identified by '<password>';

Granting all MySQL user permissions

# mysql -h localhost -u root --password=<password>
mysql> grant all privileges on dbname.* to user@host;

Select domain name from email address

SELECT SUBSTR( email, INSTR( email, '@' ) + 1 )

Check if MySQL connection is encrypted with TLS/SSL

Check the SSL version in use:

show status like 'Ssl_version';

Or check the cipher in use:

show status like 'Ssl_cipher';

Apache

Reporting loaded Apache modules

# apache2ctl -M

Maintaining .htaccess passwords

To add or modify the password for a user:

$ htpasswd /etc/apache2/passwd username

Configuring PHP session timeout in .htaccess

For a session timeout of 9 hours:

php_value session.cookie_lifetime 32400
php_value session.gc_maxlifetime 32400

Disabling PHP magic quotes in .htaccess

php_flag magic_quotes_gpc Off

Requiring HTTP Auth in .htaccess

AuthType Basic
AuthName "Speak Friend And Enter"
AuthUserFile /home/jj5/.htpasswd
Require valid-user

Restarting Apache

The hard way

$ sudo /etc/init.d/apache2 restart

The graceful way (avoids dropping active connections)

$ sudo apache2ctl graceful

Allowing directory browsing

To show directory index pages, in the apache config file:

<Directory /var/www/data>
  Options Indexes
</Directory>

C

Locating memset function

The memset function is in <string.h> as described in this article Using memset(), memcpy(), and memmove() in C

Links

PHP

Including a file relative to the including file

require_once( dirname( __FILE__ ) . '/relative/path/to.php' );

Enabling error reporting

error_reporting( E_ALL | E_STRICT );
ini_set( 'display_errors', 'On' );

Setting an error handler

set_error_handler( "error_handler", E_ALL | E_STRICT );
function error_handler( $error_code, $error_message, $error_file, $error_line, $error_context ) {
  // ...
}

Disable HTML content in var_dump

ini_set( 'html_errors', 'off' );

Report PHP modules

$ php -m

PHP Security Best Practices For Sys Admins

See Linux 25 PHP Security Best Practices For Sys Admins.

BASH scripting

For a primer on bash scripting see TFM: Erotic Fantasy: /bin/sh Programming.

Telling a script to run in bash

The first line of the file should be:

#!/bin/bash

Checking if a command-line argument was passed in

if [ -n "$1" ]; then
  echo "Missing parameter 1.";
  exit 1;
fi

Checking if a command-line argument was not passed in

if [ "$1" = "" ]; then
  echo "Missing parameter 1.";
  exit 1;
fi

Or:

if [ -z "$1" ]; then
  echo "Missing parameter 1.";
  exit 1;
fi

Checking command exit status

cd /my/path
if [ "$?" -ne "0" ]; then
  echo "Cannot change dir.";
  exit 1;
fi

Checking if a file does/doesn't exist

Check if file exists:

if [ -f "/my/file" ]; then
  cat /my/file
fi

Check if file doesn't exist:

if [ ! -f "/my/file" ]; then
  touch /my/file
fi

Checking if a directory does/doesn't exist

Check if directory exists:

if [ -d "/my/dir" ]; then
  rmdir /my/dir
fi

Check if directory doesn't exist:

if [ ! -d "/my/dir" ]; then
  mkdir /my/dir
fi

Deleting old backups

To keep only the latest five backups:

find . -maxdepth 1 -type f -printf '%T@ %p\0' | sort -r -z -n | awk 'BEGIN { RS="\0"; ORS="\0"; FS="" } NR > 5 { sub("^[0-9]*(.[0-9]*)? ", ""); print }' | xargs -0 rm -f

This script stolen from stackoverflow.

Requires GNU find for -printf, GNU sort for -z, GNU awk for "\0" and GNU xargs for -0, but handles files with embedded newlines or spaces.

Changing into the script's directory

cd "`dirname $0`"

Getting the absolute path of a relative path

readlink -f ./some/path

Creating a temp directory

dir=`mktemp -d` && cd $dir

Reading secret input from stdin

You can read a secret, such as a password, like this:

echo -n "Enter passphrase: "
stty -echo
read passphrase;
stty echo
echo ""

After running the above the secret will be in the $passphrase environment variable.

String replacements in bash

See the string manipulation doco. Basically, to replace first occurrence:

result=${var/find/replace}

To replace all occurrences:

result=${var//find/replace}

A practical example, get an ISO date and turn it into a path:

date="$(date +%Y-%m-%d)"
work_dir=${date//-//}

Sending a HEREDOC to a file

cat << EOF > /tmp/yourfilehere
These contents will be written to the file.
        This line is indented.
EOF

Bash case/switch statement

See using case statements, e.g.:

case $space in
[1-6]*)
  Message="All is quiet."
  ;;
[7-8]*)
  Message="Start thinking about cleaning out some stuff.  There's a partition that is $space % full."
  ;;
9[1-8])
  Message="Better hurry with that new disk...  One partition is $space % full."
  ;;
99)
  Message="I'm drowning here!  There's a partition at $space %!"
  ;;
*)
  Message="I seem to be running with an nonexistent amount of disk space..."
  ;;
esac

Using dotglob shopt to match dot-files

To enable dot-file matching in globs, set the dotglob shell option:

$ shopt -s dotglob

Stopping a script from running if it previously exited due to error

persistentDataDir=/var/lib/something
alarm() {
  touch $persistentDataDir/alarm
}
trap alarm ERR
[ -f $persistentDataDir/alarm ] && exit 1

Make sure only one instance of a script is running at a time

ephemeralDataDir=/var/run/something
unlock() {
  rmdir $ephemeralDataDir/lock
}
mkdir $ephemeralDataDir/lock || exit 1;
trap unlock EXIT

Sed

Find and replace with sed

To update the current file use '-i'. E.g.:

sed -i 's/search-text/replace-text/' file

Awk

Listing IP addresses in an Apache web log

awk '/GET \/path\/for\/url/ { print $1 }' /var/log/apache2/access.log | sort | uniq

Printing space-separated field

echo 'no no yes no' | awk '{print $3}'

Printing delimited field

echo 'no:no:yes:no' | awk -F ':' '{print $3}'

Subversion

Setting svn:externals from the command-line

See here.

To set an svn:externals from the command-line:

svn propset svn:externals 'rdfind-php https://www.progclub.org/svn/pcrepo/rdfind.php/branches/0.1' .
svn ci -m 'Adding svn:externals for rdfind-php...'
svn up

Or to use a file:

svn propset svn:externals -F svn.externals .

Setting svn:ignore from the command line

See here.

$ svn propset svn:ignore [file|folder] [path]

Or use a file and apply recursively:

$ svn propset svn:ignore -RF ./svn-ignore-list.txt .

Git

Showing status of working copy

git status

Showing repo history

git log

Showing remote repositories (including 'origin')

git remote -v

Handy git aliases

Save these to your ~/.gitconfig file.

For a nicer view of history than standard 'git log' -- colourful, one-line-per commit, etc:

 graph = !git log --all --graph --color --abbrev-commit --pretty=oneline

To show only the files that have changed, rather than the full line-by-line content:

 dif   = !git diff --name-status

IRC

Instructing ChanServ to op an admin

/msg ChanServ op #channel user

E.g.

/msg ChanServ op #gnurc jj5

Sub 'op' for 'deop' to remove op privilege.

C++

C++ books

Books I want

Books I own

Books I'm not reading

Books I've read

C++ blogs/articles

C++ performance tips

  • ++c can be faster than c++.
  • use const for everything that you possibly can.
  • use 'inline' when you need to define a function in a header. Typically only do that if it's small and the increase in code size from inlining is worth the cost to avoid the cost of a function call. For anything except trivially small functions you'll probably need to profile to know if it's worth it.
  • don't use registers.
  • const rarely affects performance.
  • debunking a number of C++ myths that won't die.
  • std::sort<> is typically faster than qsort() because it can avoid indirection at runtime.
  • if you've got parallelisation going on, you may be able to just replace a std::for_each with a parallel equivalent.
  • read about performance cost of RTTI (Run Time Type Information) and how to disable it
  • don't use dynamic_cast because it is slow (typeid is faster but still relies on RTTI)
  • prefer unique_ptr to shared_ptr when possible. unique_ptr has less overhead.
  • Which is better, static or dynamic linking?
  • Integer vs Floating-Point performance

systemd

systemd is an init system used in most Linux distributions to bootstrap the user space and manage all processes subsequently.

Following a service log

e.g. for bind9:

# journalctl -f -u bind9

or for everything:

# journalctl -f

System status

To see spawned services hierarchy:

# systemctl status

Or for a specific service e.g.:

# systemctl status networking

SaltStack

Running a command on specified minions

salt 'host' cmd.run 'update-locale'

Running a command on all minions

salt '*' cmd.run 'update-locale'

Listing active jobs

salt-run jobs.active

Listing available grains

salt 'example' grains.items

Listing available pillar

salt 'example' pillar.items

Reporting a grain value

e.g. for the 'mem_total' grain:

salt '*' grains.item mem_total

KDE

Running user login script (X11/XOrg/XWindows)

A way to run user login scripts which works for KDE Plasma (and apparently other X.Org Server X Window System environments) is to create a *.desktop file in ~/.config/autostart/. For example I have a ~/.config/autostart/ssh-add.desktop file with the following contents to register my SSH key in the SSH Agent:

[Desktop Entry]
Type=Application
Name=ssh-add
Comment=Adds my private key to my session.
Exec=/usr/bin/konsole -e 'ssh-add /home/$USER/.ssh/id_rsa'

VirtualBox

See here:

Install qemu if necessary:

# apt install qemu

Then you'll need to load the network block device module:

# rmmod nbd
# modprobe nbd max_part=16

Attach the .vdi image to one of the nbd you just created:

# qemu-nbd -c /dev/nbd0 drive.vdi

Now you will get a /dev/nbd0 block device, along with several /dev/nbd0p* partition device nodes.

# mount /dev/nbd0p1 /mnt

Once you are done, unmount everything and disconnect the device:

# qemu-nbd -d /dev/nbd0