Fix race conditions where a process gets in state "Running"

between the restart/start command and the summary command.

Refactor to avoid repeating the status, and fail if a given call to monit fails.
This commit is contained in:
Brian Brazil 2014-04-16 18:04:19 +01:00
parent 36b9bc4d40
commit 8ec400669f

View file

@ -47,8 +47,6 @@ EXAMPLES = '''
- monit: name=httpd state=started
'''
import pipes
def main():
arg_spec = dict(
name=dict(required=True),
@ -68,8 +66,25 @@ def main():
rc, out, err = module.run_command('%s reload' % MONIT)
module.exit_json(changed=True, name=name, state=state)
rc, out, err = module.run_command('%s summary | grep "Process \'%s\'"' % (MONIT, pipes.quote(name)), use_unsafe_shell=True)
present = name in out
def status():
"""Return the status of the process in monit, or None if not present."""
rc, out, err = module.run_command('%s summary' % MONIT, check_rc=True)
for line in out.split('\n'):
# Sample output lines:
# Process 'name' Running
# Process 'name' Running - restart pending
parts = line.lower().split()
if len(parts) > 2 and parts[0] == 'process' and parts[1] == "'%s'" % name:
return ' '.join(parts[2:])
else:
return None
def run_command(command):
"""Runs a monit command, and returns the new status."""
module.run_command('%s %s %s' % (MONIT, command, name), check_rc=True)
return status()
present = status() is not None
if not present and not state == 'present':
module.fail_json(msg='%s process not presently configured with monit' % name, name=name, state=state)
@ -78,69 +93,57 @@ def main():
if not present:
if module.check_mode:
module.exit_json(changed=True)
module.run_command('%s reload' % MONIT, check_rc=True)
rc, out, err = module.run_command('%s summary | grep %s' % (MONIT, pipes.quote(name)), use_unsafe_shell=True)
if name in out:
module.exit_json(changed=True, name=name, state=state)
status = run_command('reload')
if status is None:
module.fail_json(msg=status, name=name, state=state)
else:
module.fail_json(msg=out, name=name, state=state)
module.exit_json(changed=True, name=name, state=state)
module.exit_json(changed=False, name=name, state=state)
rc, out, err = module.run_command('%s summary | grep %s' % (MONIT, pipes.quote(name)), use_unsafe_shell=True)
running = 'running' in out.lower()
running = 'running' in status()
if running and (state == 'started' or state == 'monitored'):
module.exit_json(changed=False, name=name, state=state)
if running and state == 'monitored':
if running and state in ['started', 'monitored']:
module.exit_json(changed=False, name=name, state=state)
if running and state == 'stopped':
if module.check_mode:
module.exit_json(changed=True)
module.run_command('%s stop %s' % (MONIT, name))
rc, out, err = module.run_command('%s summary | grep %s' % (MONIT, pipes.quote(name)), use_unsafe_shell=True)
if 'not monitored' in out.lower() or 'stop pending' in out.lower():
status = run_command('stop')
if status in ['not monitored'] or 'stop pending' in status:
module.exit_json(changed=True, name=name, state=state)
module.fail_json(msg=out)
module.fail_json(msg=status)
if running and state == 'unmonitored':
if module.check_mode:
module.exit_json(changed=True)
module.run_command('%s unmonitor %s' % (MONIT, name))
# FIXME: DRY FOLKS!
rc, out, err = module.run_command('%s summary | grep %s' % (MONIT, pipes.quote(name)), use_unsafe_shell=True)
if 'not monitored' in out.lower():
status = run_command('unmonitor')
if status in ['not monitored']:
module.exit_json(changed=True, name=name, state=state)
module.fail_json(msg=out)
module.fail_json(msg=status)
elif state == 'restarted':
if module.check_mode:
module.exit_json(changed=True)
module.run_command('%s restart %s' % (MONIT, name))
rc, out, err = module.run_command('%s summary | grep %s' % (MONIT, name))
if 'initializing' in out.lower() or 'restart pending' in out.lower():
status = run_command('restart')
if status in ['initializing', 'running'] or 'restart pending' in status:
module.exit_json(changed=True, name=name, state=state)
module.fail_json(msg=out)
module.fail_json(msg=status)
elif not running and state == 'started':
if module.check_mode:
module.exit_json(changed=True)
module.run_command('%s start %s' % (MONIT, name))
rc, out, err = module.run_command('%s summary | grep %s' % (MONIT, name))
if 'initializing' in out.lower() or 'start pending' in out.lower():
status = run_command('start')
if status in ['initializing', 'running'] or 'start pending' in status:
module.exit_json(changed=True, name=name, state=state)
module.fail_json(msg=out)
module.fail_json(msg=status)
elif not running and state == 'monitored':
if module.check_mode:
module.exit_json(changed=True)
module.run_command('%s monitor %s' % (MONIT, name))
rc, out, err = module.run_command('%s summary | grep %s' % (MONIT, name))
if 'initializing' in out.lower() or 'start pending' in out.lower():
status = run_command('monitor')
if status() not in ['not monitored']:
module.exit_json(changed=True, name=name, state=state)
module.fail_json(msg=out)
module.fail_json(msg=status)
module.exit_json(changed=False, name=name, state=state)