git_commit_behind

#! /usr/bin/env python3

"""=cut
=head1 NAME

git_commit_behind - Munin plugin to monitor local git repositories and report how many commits behind their remote they are

=head1 NOTES

This plugin is similar to how apt_all works for apt packages.

To be able to check how behind a git repository is, we need to run git fetch.
To avoid fetching all repos every 5 minutes (for each munin period) and thus
slowing down the data collection, the git fetch operation is only randomly
triggered (based on env.update.probability).
In case of very time-consuming update operations, you can run them in a
separate cron job.

=head1 REQUIREMENTS

=over 4

=item Python3

=item Git

=back


=head1 INSTALLATION

Link this plugin, as usual.
For example :

  ln -s /path/to/git_commit_behind /etc/munin/plugins/git_commit_behind

If you wish to update the repositories via cron and not during the plugin
execution (cf CONFIGURATION section), you need a dedicated cron job.

For example, you can use the following cron :

 # If the git_commit_behind plugin is enabled, fetch git repositories randomly
 # according to the plugin configuration.
 # By default : once an hour (12 invocations an hour, 1 in 12 chance that the
 # update will happen), but ensure that there will never be more than two hours
 # (7200 seconds) interval between updates.
 */5 * * * * root if [ -x /etc/munin/plugins/git_commit_behind ]; then /usr/sbin/munin-run git_commit_behind update >/dev/null; fi

=head1 CONFIGURATION

Use your "/etc/munin/plugin-conf.d/munin-node" to configure this plugin.

    [git_commit_behind]
    user [user]
    env.git_path /path/to/git
    env.update.mode [munin|cron]
    env.update.probability 12
    env.update.maxinterval 7200

user [user] : required, the owner of the repository checkouts
    in case of multiple different owners, use root
env.git_path : optional (default : /usr/bin/git), the path to the git binary.
env.update.mode : optional (default : munin), the update mode.
    munin : repositories are git fetched during the pugin execution
    cron : a dedicated cron job needs to be used to update the repositories
env.update.probability : optional (default : 12),
    runs the update randomly (1 in <probability> chances)
env.update.maxinterval : optional (default : 7200),
    ensures that the update is run at least every <maxinterval> seconds


Then, for each repository you want to check, you need the following
configuration block under the git_commit_behind section:

    env.repo.[repoCode].path /path/to/local/repo
    env.repo.[repoCode].name Repo Name
    env.repo.[repoCode].user user
    env.repo.[repoCode].warning 10
    env.repo.[repoCode].critical 100

[repoCode] can only contain letters, numbers and underscores.

path : mandatory, the local path to your git repository
name : optional (default : [repoCode]), a cleaner name that will be displayed
user : optional (default : empty), the owner of the repository
       if set and different from the user running the plugin, the git commands
       will be executed as this user
warning : optional (default 10), the warning threshold
critical : optional (default 100), the critical threshold

For example :

    [git_commit_behind]
    user root

    env.repo.munin_contrib.path /opt/munin-contrib
    env.repo.munin_contrib.name Munin Contrib

    env.repo.other_repo.path /path/to/other-repo
    env.repo.other_repo.name Other Repo

=head1 MAGIC MARKERS

  #%# family=auto
  #%# capabilities=autoconf

=head1 VERSION

1.0.0

=head1 AUTHOR

Neraud (https://github.com/Neraud)

=head1 LICENSE

GPLv2

=cut"""


import logging
import os
from pathlib import Path
import pwd
from random import randint
import re
from shlex import quote
from subprocess import check_output, call, DEVNULL, CalledProcessError
import sys
import time


plugin_version = "1.0.0"

if int(os.getenv('MUNIN_DEBUG', 0)) > 0:
    logging.basicConfig(level=logging.DEBUG,
                        format='%(asctime)s %(levelname)-7s %(message)s')

current_user = pwd.getpwuid(os.geteuid())[0]

conf = {
    'git_path':            os.getenv('git_path', '/usr/bin/git'),
    'state_file':          os.getenv('MUNIN_STATEFILE'),
    'update_mode':         os.getenv('update.mode', 'munin'),
    'update_probability':  int(os.getenv('update.probability', '12')),
    'update_maxinterval':  int(os.getenv('update.maxinterval', '7200'))
}

repo_codes = set(re.search('repo\.([^.]+)\..*', elem).group(1)
                 for elem in os.environ.keys() if elem.startswith('repo.'))

repos_conf = {}
for code in repo_codes:
    repos_conf[code] = {
        'name':        os.getenv('repo.%s.name' % code, code),
        'path':        os.getenv('repo.%s.path' % code, None),
        'user':        os.getenv('repo.%s.user' % code, None),
        'warning':     os.getenv('repo.%s.warning' % code, '10'),
        'critical':    os.getenv('repo.%s.critical' % code, '100')
    }


def print_config():
    print('graph_title Git repositories - Commits behind')

    print('graph_args --base 1000 -r --lower-limit 0')
    print('graph_vlabel number of commits behind')
    print('graph_scale yes')
    print('graph_info This graph shows the number of commits behind' +
          ' for each configured git repository')
    print('graph_category file_transfer')

    print('graph_order %s' % ' '.join(repo_codes))

    for repo_code in repos_conf.keys():
        print('%s.label %s' % (repo_code, repos_conf[repo_code]['name']))
        print('%s.warning %s' % (repo_code, repos_conf[repo_code]['warning']))
        print('%s.critical %s' %
              (repo_code, repos_conf[repo_code]['critical']))


def generate_git_command(repo_conf, git_command):
    if not repo_conf['user'] or repo_conf['user'] == current_user:
        cmd = [quote(conf['git_path'])] + git_command
    else:
        shell_cmd = 'cd %s ; %s %s' % (
            quote(repo_conf['path']),
            quote(conf['git_path']),
            ' '.join(git_command))
        cmd = ['su', '-', repo_conf['user'], '-s', '/bin/sh', '-c', shell_cmd]
    return cmd


def execute_git_command(repo_conf, git_command):
    cmd = generate_git_command(repo_conf, git_command)
    return check_output(cmd, cwd=repo_conf['path']).decode('utf-8').rstrip()


def print_info():
    if not os.access(conf['git_path'], os.X_OK):
        print('Git (%s) is missing, or not executable !' %
              conf['git_path'], file=sys.stderr)
        sys.exit(1)

    for repo_code in repos_conf.keys():
        logging.debug(' - %s' % repo_code)
        try:
            remote_branch = execute_git_command(
                repos_conf[repo_code],
                ['rev-parse', '--abbrev-ref', '--symbolic-full-name', '@{u}'])
            logging.debug('remote_branch = %s' % remote_branch)

            commits_behind = execute_git_command(
                repos_conf[repo_code],
                ['rev-list', 'HEAD..%s' % remote_branch, '--count'])

            print('%s.value %d' % (repo_code, int(commits_behind)))
        except CalledProcessError as e:
            logging.error('Error executing git command : %s', e)
        except FileNotFoundError as e:
            logging.error('Repo not found at path %s' %
                          repos_conf[repo_code]['path'])


def check_update_repos(mode):
    if not conf['state_file']:
        logging.error('Munin state file unavailable')
        sys.exit(1)

    if mode != conf['update_mode']:
        logging.debug('Wrong mode, skipping')
        return

    if not os.path.isfile(conf['state_file']):
        logging.debug('No state file -> updating')
        do_update_repos()
    elif (os.path.getmtime(conf['state_file']) + conf['update_maxinterval']
            < time.time()):
        logging.debug('State file last modified too long ago -> updating')
        do_update_repos()
    elif randint(1, conf['update_probability']) == 1:
        logging.debug('Recent state, but random matched -> updating')
        do_update_repos()
    else:
        logging.debug('Recent state and random missed -> skipping')


def do_update_repos():
    for repo_code in repos_conf.keys():
        try:
            logging.info('Fetching repo %s' % repo_code)
            execute_git_command(repos_conf[repo_code], ['fetch'])
        except CalledProcessError as e:
            logging.error('Error executing git command : %s', e)
        except FileNotFoundError as e:
            logging.error('Repo not found at path %s' %
                          repos_conf[repo_code]['path'])
    logging.debug('Updating the state file')

    # 'touch' the state file to update its last modified date
    Path(conf['state_file']).touch()


if len(sys.argv) > 1:
    action = sys.argv[1]
    if action == 'config':
        print_config()
    elif action == 'autoconf':
        errors = []

        if not conf['state_file']:
            errors.append('munin state file unavailable')

        if os.access(conf['git_path'], os.X_OK):
            test_git = call([conf['git_path'], '--version'], stdout=DEVNULL)
            if test_git != 0:
                errors.append('git seems to be broken ?!')
        else:
            errors.append('git is missing or not executable')

        if errors:
            print('no (%s)' % ', '.join(errors))
        else:
            print('yes')
    elif action == 'version':
        print('Git commit behind Munin plugin, version {0}'.format(
            plugin_version))
    elif action == 'update':
        check_update_repos('cron')
    else:
        logging.warn("Unknown argument '%s'" % action)
        sys.exit(1)
else:
    if conf['update_mode'] == 'munin':
        check_update_repos('munin')
    print_info()