原生的ansible不支持service.status,在Google之后,发现有人提交了一个patch,可以支持status选项.见https://github.com/ritzk/ansible-modules-core/commit/ab4c9c226e94371c2b60aab57ef5b0a76f75a74b
就此使用一段时间后,在centos 7使用时,出现了报错.
root@HOSTXXXXXXX001:~# ansible HOSTXXXXXXX003 -m service -a 'name="zabbix-agent" state="status"' HOSTXXXXXXX003 | FAILED >> { "failed": true, "msg": "value of state must be one of: running,started,stopped,restarted,reloaded, got: status" }
以前都是在centos 6使用的,正常,如今在centos 7出现报错,就打开源码查看了一下.
发现centos 6是/usr/local/lib/python3.6/dist-packages/ansible/modules/system/service.py
而centos 7使用的则是/usr/local/lib/python3.6/dist-packages/ansible/modules/system/systemd.py , 该文件并没有进行patch
于是对该文件进行了修改,如下.
#!/usr/bin/python # -*- coding: utf-8 -*- # Copyright: (c) 2016, Brian Coca <bcoca@ansible.com> # GNU General Public License v3.0+ (see COPYING or https://www.gnu.org/licenses/gpl-3.0.txt) # patch by Redheat # email qjyyn@qq.com from __future__ import absolute_import, division, print_function __metaclass__ = type ANSIBLE_METADATA = {'metadata_version': '1.1', 'status': ['stableinterface'], 'supported_by': 'core'} DOCUMENTATION = ''' module: systemd author: - Ansible Core Team version_added: "2.2" short_description: Manage services description: - Controls systemd services on remote hosts. options: name: description: - Name of the service. When using in a chroot environment you always need to specify the full name i.e. (crond.service). aliases: [ service, unit ] state: description: - C(started)/C(stopped) are idempotent actions that will not run commands unless necessary. C(restarted) will always bounce the service. C(reloaded) will always reload. choices: [ reloaded, restarted, started, stopped ] enabled: description: - Whether the service should start on boot. B(At least one of state and enabled are required.) type: bool masked: description: - Whether the unit should be masked or not, a masked unit is impossible to start. type: bool daemon_reload: description: - run daemon-reload before doing any other operations, to make sure systemd has read any changes. type: bool default: 'no' aliases: [ daemon-reload ] user: description: - run systemctl talking to the service manager of the calling user, rather than the service manager of the system. type: bool default: 'no' no_block: description: - Do not synchronously wait for the requested operation to finish. Enqueued job will continue without Ansible blocking on its completion. type: bool default: 'no' version_added: "2.3" notes: - Since 2.4, one of the following options is required 'state', 'enabled', 'masked', 'daemon_reload', and all except 'daemon_reload' also require 'name'. - Before 2.4 you always required 'name'. requirements: - A system managed by systemd. ''' EXAMPLES = ''' - name: Make sure a service is running systemd: state=started name=httpd - name: stop service cron on debian, if running systemd: name=cron state=stopped - name: restart service cron on centos, in all cases, also issue daemon-reload to pick up config changes systemd: state: restarted daemon_reload: yes name: crond - name: reload service httpd, in all cases systemd: name: httpd state: reloaded - name: enable service httpd and ensure it is not masked systemd: name: httpd enabled: yes masked: no - name: enable a timer for dnf-automatic systemd: name: dnf-automatic.timer state: started enabled: True - name: just force systemd to reread configs (2.4 and above) systemd: daemon_reload=yes ''' RETURN = ''' status: description: A dictionary with the key=value pairs returned from `systemctl show` returned: success type: complex contains: { "ActiveEnterTimestamp": "Sun 2016-05-15 18:28:49 EDT", "ActiveEnterTimestampMonotonic": "8135942", "ActiveExitTimestampMonotonic": "0", "ActiveState": "active", "After": "auditd.service systemd-user-sessions.service time-sync.target systemd-journald.socket basic.target system.slice", "AllowIsolate": "no", "Before": "shutdown.target multi-user.target", "BlockIOAccounting": "no", "BlockIOWeight": "1000", "CPUAccounting": "no", "CPUSchedulingPolicy": "0", "CPUSchedulingPriority": "0", "CPUSchedulingResetOnFork": "no", "CPUShares": "1024", "CanIsolate": "no", "CanReload": "yes", "CanStart": "yes", "CanStop": "yes", "CapabilityBoundingSet": "18446744073709551615", "ConditionResult": "yes", "ConditionTimestamp": "Sun 2016-05-15 18:28:49 EDT", "ConditionTimestampMonotonic": "7902742", "Conflicts": "shutdown.target", "ControlGroup": "/system.slice/crond.service", "ControlPID": "0", "DefaultDependencies": "yes", "Delegate": "no", "Description": "Command Scheduler", "DevicePolicy": "auto", "EnvironmentFile": "/etc/sysconfig/crond (ignore_errors=no)", "ExecMainCode": "0", "ExecMainExitTimestampMonotonic": "0", "ExecMainPID": "595", "ExecMainStartTimestamp": "Sun 2016-05-15 18:28:49 EDT", "ExecMainStartTimestampMonotonic": "8134990", "ExecMainStatus": "0", "ExecReload": "{ path=/bin/kill ; argv[]=/bin/kill -HUP $MAINPID ; ignore_errors=no ; start_time=[n/a] ; stop_time=[n/a] ; pid=0 ; code=(null) ; status=0/0 }", "ExecStart": "{ path=/usr/sbin/crond ; argv[]=/usr/sbin/crond -n $CRONDARGS ; ignore_errors=no ; start_time=[n/a] ; stop_time=[n/a] ; pid=0 ; code=(null) ; status=0/0 }", "FragmentPath": "/usr/lib/systemd/system/crond.service", "GuessMainPID": "yes", "IOScheduling": "0", "Id": "crond.service", "IgnoreOnIsolate": "no", "IgnoreOnSnapshot": "no", "IgnoreSIGPIPE": "yes", "InactiveEnterTimestampMonotonic": "0", "InactiveExitTimestamp": "Sun 2016-05-15 18:28:49 EDT", "InactiveExitTimestampMonotonic": "8135942", "JobTimeoutUSec": "0", "KillMode": "process", "KillSignal": "15", "LimitAS": "18446744073709551615", "LimitCORE": "18446744073709551615", "LimitCPU": "18446744073709551615", "LimitDATA": "18446744073709551615", "LimitFSIZE": "18446744073709551615", "LimitLOCKS": "18446744073709551615", "LimitMEMLOCK": "65536", "LimitMSGQUEUE": "819200", "LimitNICE": "0", "LimitNOFILE": "4096", "LimitNPROC": "3902", "LimitRSS": "18446744073709551615", "LimitRTPRIO": "0", "LimitRTTIME": "18446744073709551615", "LimitSIGPENDING": "3902", "LimitSTACK": "18446744073709551615", "LoadState": "loaded", "MainPID": "595", "MemoryAccounting": "no", "MemoryLimit": "18446744073709551615", "MountFlags": "0", "Names": "crond.service", "NeedDaemonReload": "no", "Nice": "0", "NoNewPrivileges": "no", "NonBlocking": "no", "NotifyAccess": "none", "OOMScoreAdjust": "0", "OnFailureIsolate": "no", "PermissionsStartOnly": "no", "PrivateNetwork": "no", "PrivateTmp": "no", "RefuseManualStart": "no", "RefuseManualStop": "no", "RemainAfterExit": "no", "Requires": "basic.target", "Restart": "no", "RestartUSec": "100ms", "Result": "success", "RootDirectoryStartOnly": "no", "SameProcessGroup": "no", "SecureBits": "0", "SendSIGHUP": "no", "SendSIGKILL": "yes", "Slice": "system.slice", "StandardError": "inherit", "StandardInput": "null", "StandardOutput": "journal", "StartLimitAction": "none", "StartLimitBurst": "5", "StartLimitInterval": "10000000", "StatusErrno": "0", "StopWhenUnneeded": "no", "SubState": "running", "SyslogLevelPrefix": "yes", "SyslogPriority": "30", "TTYReset": "no", "TTYVHangup": "no", "TTYVTDisallocate": "no", "TimeoutStartUSec": "1min 30s", "TimeoutStopUSec": "1min 30s", "TimerSlackNSec": "50000", "Transient": "no", "Type": "simple", "UMask": "0022", "UnitFileState": "enabled", "WantedBy": "multi-user.target", "Wants": "system.slice", "WatchdogTimestampMonotonic": "0", "WatchdogUSec": "0", } ''' # NOQA from ansible.module_utils.basic import AnsibleModule from ansible.module_utils.service import sysv_exists, sysv_is_enabled, fail_if_missing from ansible.module_utils._text import to_native def is_running_service(service_status): return service_status['ActiveState'] in set(['active', 'activating']) def request_was_ignored(out): return '=' not in out and 'ignoring request' in out def parse_systemctl_show(lines): # The output of 'systemctl show' can contain values that span multiple lines. At first glance it # appears that such values are always surrounded by {}, so the previous version of this code # assumed that any value starting with { was a multi-line value; it would then consume lines # until it saw a line that ended with }. However, it is possible to have a single-line value # that starts with { but does not end with } (this could happen in the value for Description=, # for example), and the previous version of this code would then consume all remaining lines as # part of that value. Cryptically, this would lead to Ansible reporting that the service file # couldn't be found. # # To avoid this issue, the following code only accepts multi-line values for keys whose names # start with Exec (e.g., ExecStart=), since these are the only keys whose values are known to # span multiple lines. parsed = {} multival = [] k = None for line in lines: if k is None: if '=' in line: k, v = line.split('=', 1) if k.startswith('Exec') and v.lstrip().startswith('{'): if not v.rstrip().endswith('}'): multival.append(v) continue parsed[k] = v.strip() k = None else: multival.append(line) if line.rstrip().endswith('}'): parsed[k] = ' '.join(multival).strip() multival = [] k = None return parsed # =========================================== # Main control flow def main(): # initialize module = AnsibleModule( argument_spec=dict( name=dict(type='str', aliases=['service', 'unit']),
#choices里如果没有对应的操作就会报value of state must be one of: reloaded, restarted, started, stopped, got: status
#报错文件位于ansible/module_utils/basic.py 1810行
state=dict(type='str', choices=['reloaded', 'restarted', 'started', 'stopped', 'status']),
enabled=dict(type='bool'), masked=dict(type='bool'), daemon_reload=dict(type='bool', default=False, aliases=['daemon-reload']), user=dict(type='bool', default=False), no_block=dict(type='bool', default=False), ), supports_check_mode=True, required_one_of=[['state', 'enabled', 'masked', 'daemon_reload']], ) systemctl = module.get_bin_path('systemctl', True) if module.params['user']: systemctl = systemctl + " --user" if module.params['no_block']: systemctl = systemctl + " --no-block" unit = module.params['name'] rc = 0 out = err = '' result = dict( name=unit, changed=False, status=dict(), ) for requires in ('state', 'enabled', 'masked'): if module.params[requires] is not None and unit is None: module.fail_json(msg="name is also required when specifying %s" % requires) # Run daemon-reload first, if requested if module.params['daemon_reload'] and not module.check_mode: (rc, out, err) = module.run_command("%s daemon-reload" % (systemctl)) if rc != 0: module.fail_json(msg='failure %d during daemon-reload: %s' % (rc, err)) if unit: found = False is_initd = sysv_exists(unit) is_systemd = False # check service data, cannot error out on rc as it changes across versions, assume not found
#这里执行的命令是类似systemctl show zabbix-agent (rc, out, err) = module.run_command("%s show '%s'" % (systemctl, unit)) if request_was_ignored(out) or request_was_ignored(err): # fallback list-unit-files as show does not work on some systems (chroot) # not used as primary as it skips some services (like those using init.d) and requires .service/etc notation (rc, out, err) = module.run_command("%s list-unit-files '%s'" % (systemctl, unit)) if rc == 0: is_systemd = True elif rc == 0: # load return of systemctl show into dictionary for easy access and return if out:
#把返回值都扔进result['status']去了 result['status'] = parse_systemctl_show(to_native(out).split(' ')) is_systemd = 'LoadState' in result['status'] and result['status']['LoadState'] != 'not-found' # Check for loading error if is_systemd and 'LoadError' in result['status']: module.fail_json(msg="Error loading unit file '%s': %s" % (unit, result['status']['LoadError'])) else: # Check for systemctl command module.run_command(systemctl, check_rc=True) # Does service exist? found = is_systemd or is_initd if is_initd and not is_systemd: module.warn('The service (%s) is actually an init script but the system is managed by systemd' % unit) # mask/unmask the service, if requested, can operate on services before they are installed if module.params['masked'] is not None: # state is not masked unless systemd affirms otherwise masked = ('LoadState' in result['status'] and result['status']['LoadState'] == 'masked') if masked != module.params['masked']: result['changed'] = True if module.params['masked']: action = 'mask' else: action = 'unmask' if not module.check_mode: (rc, out, err) = module.run_command("%s %s '%s'" % (systemctl, action, unit)) if rc != 0: # some versions of system CAN mask/unmask non existing services, we only fail on missing if they don't fail_if_missing(module, found, unit, msg='host') # if module.params['choices'] == 'status': # Enable/disable service startup at boot if requested if module.params['enabled'] is not None: if module.params['enabled']: action = 'enable' else: action = 'disable' fail_if_missing(module, found, unit, msg='host') # do we need to enable the service? enabled = False (rc, out, err) = module.run_command("%s is-enabled '%s'" % (systemctl, unit)) # check systemctl result or if it is a init script if rc == 0: enabled = True elif rc == 1: # if not a user service and both init script and unit file exist stdout should have enabled/disabled, otherwise use rc entries if not module.params['user'] and is_initd and (not out.strip().endswith('disabled') or sysv_is_enabled(unit)): enabled = True # default to current state result['enabled'] = enabled # Change enable/disable if needed if enabled != module.params['enabled']: result['changed'] = True if not module.check_mode: (rc, out, err) = module.run_command("%s %s '%s'" % (systemctl, action, unit)) if rc != 0: module.fail_json(msg="Unable to %s service %s: %s" % (action, unit, out + err)) result['enabled'] = not enabled # set service state if requested if module.params['state'] is not None: fail_if_missing(module, found, unit, msg="host") # default to desired state result['state'] = module.params['state'] # What is current service state? if 'ActiveState' in result['status']: action = None if module.params['state'] == 'started': if not is_running_service(result['status']): action = 'start' elif module.params['state'] == 'stopped': if is_running_service(result['status']): action = 'stop'
# patch status
# 如果是status,则判断返回的result['status'],并标记result['state']
# action其实是个flag,影响的是接下来的操作.完全可以不对action赋值,就不会影响下面的操作
# 仿照service.py这个patch,result['status']被我删了
elif module.params['state'] == 'status': if is_running_service(result['status']): result['state'] = True else: result['state'] = False del result['status']
else: if not is_running_service(result['status']): action = 'start' else: action = module.params['state'][:-2] # remove 'ed' from restarted/reloaded result['state'] = 'started' if action: result['changed'] = True if not module.check_mode: (rc, out, err) = module.run_command("%s %s '%s'" % (systemctl, action, unit)) if rc != 0: module.fail_json(msg="Unable to %s service %s: %s" % (action, unit, err)) else: # this should not happen? module.fail_json(msg="Service is in unknown state", status=result['status']) module.exit_json(**result) if __name__ == '__main__': main()
染红加粗的几行,就是对status的支持和我写的几行简单说明.
我在git对源码进行了fork,见https://github.com/RedheatWei/ansible-modules-core