Skip to content
This repository has been archived by the owner on Mar 30, 2020. It is now read-only.
/ sysadmin-util Public archive

Tools for Linux/Unix sysadmins.

License

Notifications You must be signed in to change notification settings

skx/sysadmin-util

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

sysadmin-utils

This repository contains a small collection of scripts that might be useful to sysadmins. I put it together myself to centralise the small tools that I find useful, and it seems to be popular.

I used to solicit the inclusion of new tools, but have slowly come to realize that "less is more". I love the idea of sysadmins, developers, and other people building up their own toolkits, but also find that people submit things that I just don't understand the appeal of.

It makes sense that personal-tools are very personal, but it does mean rejection is almost always the default behaviour and that makes me feel bad.

Instead of adding things here consider this repository a small collection of things that I use, and if you want to take some/all of utilities into your own use then please do so. If not then I would strongly encourage you to consider what tools would make your daily-life more useful and then collect them, document them, and publish them yourself.

In short the value here is the idea of collecting your commonmost utilities and making them easy to install and update from one central-source. Not the specific tools themselves.

There is a replacement repository which is still open, and which new additions can be made more freely:

ago

Show how long ago a file/directory was modified in a human-readable fashion.

Example:

 $ ./ago /etc/passwd
 /etc/passwd 15 weeks ago

Alternatives:

  • stat and ls both show ages, but not in a human-readable fashion.

chronic

Run a command, hiding STDOUT and STDERR if it completes successfully.

Example:

 ./chronic cp /etc/passwd /tmp/not/found

This is designed to be used for cron-jobs, where output is generally ignored in the case of success.

This was written by Joey Hess and is part of moreutils.

cidr2ip

Given a set of CIDR ranges output the individual IPs in the range(s).

Example:

 $ ./cidr2ip 192.168.0.0/24
 192.168.0.0
 192.168.0.1
 192.168.0.2
 192.168.0.3
 192.168.0.4
 ..

collapse

Remove extraneous whitespace from lines, and remove empty-lines entirely.

Example:

 $ echo -e "Test1\n    f  \n\nTest2\n\n\n\n" | ./collapse
 Test1
 f
 Test2

Alternatives:

  • tr
  • ...

dupes

Report on duplicate files, via a SHA1 hash of the contents, recursively.

Example:

$ dupes
./.git/logs/HEAD
./.git/logs/refs/heads/master
./.git/refs/heads/master
./.git/refs/remotes/origin/master

Alternatives:

empty-dir

Indicate, via return code, whether a given directory is empty or not.

Example:

if empty-dir /etc; then echo "We're broken" ; fi

expand-ipv6

Expand an abbreviated/compressed IPv6 address to the full-form.

Example:

 ./expand-ipv6 fe80::1 2001:41c8:10b:103::111
 fe80:0000:0000:0000:0000:0000:0001
 2001:41c8:010b:0103:0000:0000:0111

Alternatives:

  • sipcalc
  • ...

graphite_send

A simple script to send a consistent set of metrics & values to a remote graphite instance.

The metrics may be extended via small "plugins", which are nothing more than shell/perl/ruby/python scripts in a particular directory.

Example:

 graphite_send -v

NOTE Some metrics will only be sent if the invoking user is root.

Alternatives

ipaddr

Get IP addresses easily, either all IPs, all those which are IPv4/IPv6, or those for a device. Designed primarily for scripting.

Example:

  $ ./ips -4
  lo 127.0.0.1
  eth0 80.68.84.102
  eth0 80.68.84.104

Or to see all IPv6 addreses on eth0:

  $ ipaddr -6 -d eth0
  eth0 2001:41c8:10b:102::10
  eth0 fe80::216:3eff:fe08:16a4

NOTE Requires compilation via make build.

Alternatives:

  • ip -[46] addr show
  • ifconfig -a

maybe

In a similar vein to true and false the maybe command exits with a status code of zero or one, depending on a random number.

It can be useful in scripts which need to test-failures, or which benefit from randomness:

Example:

 maybe && echo "I pass"

 maybe || echo "I fail"

multi-ping

Ping a host, regardless of whether it is an IPv6 or IPv4 host.

Example:

 $ multi-ping steve.org.uk
 Host steve.org.uk - 80.68.85.46 alive
 Host steve.org.uk - 2001:41c8:125:46:0:0:0:10 alive

As a convenience you may also specify URIs as arguments, for example:

 $ multi-ping http://steve.org.uk/foo/bar
 Host steve.org.uk - 80.68.85.46 alive
 Host steve.org.uk - 2001:41c8:125:46:0:0:0:10 alive

Requirements:

  • The Net::DNS perl module.
  • The ping + ping6 binaries.

mysql-slave-check

If the current host is a MySQL slave this script will test that the slave replication is still working.

Replication is regarded as being OK if the following three conditions are true:

  • The output of "SHOW SLAVE STATUS" includes: Slave_IO_Running: Yes
  • The output of "SHOW SLAVE STATUS" includes: Slave_SQL_Runing: Yes
  • The slave is less than 24 hours behind the master.

Example:

 # ./mysql-slave-check
 The replication appears to show an error:
 ..
 Master_Host: da-db1
 Master_User: slave
 Master_Port: 3306
 Connect_Retry: 60
 Master_Log_File: mysql-bin.000124
 Read_Master_Log_Pos: 65667
 Relay_Log_File: relay-log.001139
 Relay_Log_Pos: 27251
 Relay_Master_Log_File: mysql-bin.000124
 Slave_IO_Running: No
 Slave_SQL_Running: No
 ..

The script exits silently if all is well, unless you add "--verbose":

 # ./mysql-slave-check -v
 The slave is running, successfully.
 Replication lag: 0 seconds

Requirements:

  • There must be a file /etc/mysql/debian.cnf with valid "user=" and "password=" lines.

pyhttpd

A simple Python HTTP server, which has been updated to allow it to bind to arbitrary IP addresses, specifically to allow you to bind to localhost.

Example:

 $ ./pyhttpd 127.0.0.1:8080
 Serving HTTP on 127.0.0.1 port 8080 ...

or

 $ ./pyhttpd 8080
 Serving HTTP on 0.0.0.0 port 8080 ...

randpass

Generate a single random password via /dev/urandom.

Example:

  $ ./randpass
  cT3j7Zp6
  $ ./randpass -n 10
  ulHrNvYLaa
  $ ./randpass -n 20 -f
  oe[d7+e.{Uw=L'RO~[]{

(Adding "-f" uses the full alphabet of possible symbols, otherwise only alpha-numeric values are shown. "-n" sets the length of the generated password.)

Existing alternatives:

  • apg
  • gpw
  • pwgen
  • ...

since

Show the new output since previously reading a file. This is useful for keeping track of logfile updates.

Example:

   $ ./since /var/log/messages >/dev/null
   $ logger "testing the log"
   $ ./since /var/log/messages
   Apr 20 11:24:37 precious skx: testing the log

Existing alternatives:

  • logtail

ssh-auth-types

Show the authentication types presented by a remote SSH server.

Example:

 $ ./ssh-auth-types precious
 publickey password

 $ ./ssh-auth-types ssh.example.com
 publickey

ssh-test

Test whether ssh connections to a list of hosts will succeed, by testing each in order.

Example:

 $ ./ssh-test host.list.txt
 ssh.steve.org.uk    ... OK
 www.steve.org.uk    ... OK
 foo.example.com:222 ... OK

 $ cat host.list.txt
 ssh.steve.org.uk
 www.steve.org.uk
 foo.example.com:222

The format of the input-file is:

[user@]hostname1[:port]
[user@]hostname2[:port]
..

splay

Sleep for a random amount of time, limited by the given max value. (Default is 5 minutes).

Example:

  $ ./splay -v
  Sleeping for 77 seconds from max splay-time of 300 seconds

  $ ./splay -v -m 20
  Sleeping for 7 seconds from max splay-time of 20 seconds

Existing alternatives:

ssl-expiry-date

Report the date, and number of days, until the given SSL certificate expires. Multiple domain-names may be accepted and each is tested in turn.

The default output is "noisy", but you may add "-d" to simplify this to the domain-name and the number of days remaining on the certificate.

Example:

  ./ssl-expiry-date  bbc.co.uk
  bbc.co.uk
      Expires: Sep 18 13:50:57 2016 GMT
      Days: 266

  ./ssl-expiry-date -d bbc.co.uk steve.org.uk
  bbc.co.uk: 266
  steve.org.uk: 82

timeout

Timeout allows you to run a command which will be killed after the given number of seconds.

Example:

    # Kill the command after 63 seconds.
    ./timeout -t 63 top

    # Kill the command after two minutes, five seconds.
    ./timeout -t 2:5 top

    # Kill the command after three hours, five minutes, and seven seconds
    ./timeout -t 3:5:7 top

until-success

Repeat the specific command until it succeeds - run at least once always.

Example:

     ./until-success ssh example.com -l root -i ~/.ssh/example.com.key

Trivial (ba)sh alternatives:

  • while true ; do $cmd; done
  • watch -n 2 $cmd

when-up

Waits until a given host is online, determined by ping, until executing a given command.

Example:

 $ ./when-up 1.2.3.4 ssh [email protected]
 Waiting for 1.2.3.4 to come online...
 Last login: Sat Dec 28 23:25:01 2013 from 5.6.7.8
 [email protected]:~#

Alternatives:

until-error

Repeat the specific command until it fails - run at least once always.

Example:

     ./until-error ssh example.com -l root -i ~/.ssh/example.com.key

Trivial (ba)sh alternatives:

  • while true ; do $cmd; done
  • watch -n 2 $cmd

when-down

Waits until a given host is down

Example:

 $ ./when-down 1.2.3.4 echo "down"
 Waiting for 1.2.3.4 to get down...
 down

Alternatives:

  • until-error ping -c 1 -W 1 1.2.3.4; echo "down"

which-shell

Identify the shell we're running under.

For example:

 $ which-shell
 dash

Existing alternatives:

  • ls -l /bin/sh

with-lock

Run a command, unless an existing copy of that command is already running, via the creation of a temporary lockfile.

For example:

 with-lock rsync ...

The lockfile-name is based upon the SHA1 hash of the command to be executed and the current User-ID.

Existing alternatives:

  • lckdo - Requires you to build your own lockfile name.
  • flock - Requires you to build your own lockfile name.

Problems

Please report any issue/suggestions via the github repository:

Author

Steve Kemp [email protected]