Re: [systemd-devel] Failed state DBUS notification not received on sigkill of service

2018-10-05 Thread Lennart Poettering 
On Do, 27.09.18 14:22, prashantkumar dhotre (prashantkumardho...@gmail.com) 
wrote:

> 1) Is this a bug ?
> 2) Why service is put to failed state on sigkill ?
> 3) if putting service into failed state on sigkill is right thing to do ,
> they why dbus notification  is not being sent in case of service that
> supports restart.

I am not sure I follow. But do note that if Restart= is turned on an
configured properly, then systemd will transition the unit into
"auto-restart" state instead of "failed".

SIGKILL is not consiered a "clean" exit status by default. You can
configure that with SuccessExitStatus= however.

Lennart

-- 
Lennart Poettering, Red Hat
___
systemd-devel mailing list
systemd-devel@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/systemd-devel

[systemd-devel] Failed state DBUS notification not received on sigkill of service

2018-09-27 Thread prashantkumar dhotre 
Hi
Failed state DBUS notification  for  ActiveState property
is not received on  sigkill of service, for the service that supports
restart ('Restart' value in service faile). But i see that unit goes to
failed state in journal logs

++
Sep 25 08:52:00 mynode systemd[1]: alarmd.service: Unit entered failed
state.
Sep 25 08:52:00 mynode systemd[1]: alarmd.service: Triggering OnFailure=
dependencies.
Sep 25 08:52:00 mynode systemd[1]: alarmd.service: Failed with result
'signal'.
Sep 25 08:52:00  mynode   systemd[1]: alarmd.service: Service hold-off time
over, scheduling restart.
Sep 25 08:52:00  mynode   systemd[1]: Stopped "Alarm agent on RE".
Sep 25 08:52:00  mynode   systemd[1]: Started "Alarm agent on RE".
++

Logs from my app which monitors dbus notifications:
+++
Updated Result=signal for service alarmd.service in cache contents:
activeState=active subState=running result=signal" }
Updated ActiveState=deactivating for service alarmd.service in cache
contents: activeState=deactivating subState=running result=signal" }
Updated SubState=stop-post
Updated Result=signal for service alarmd.service in cache contents:
activeState=deactivating subState=stop-post result=signal" }
Updated ActiveState=activating for service alarmd.service in cache
contents: activeState=activating subState=stop-post result=signal" }
Updated SubState=auto-restart for service alarmd.service in cache contents:
activeState=activating subState=auto-restart result=signal" }
Updated Result=success for service alarmd.service in cache contents:
activeState=activating subState=auto-restart result=success" }
Updated ActiveState=active for service alarmd.service in cache contents:
activeState=active subState=auto-restart result=success" }
Updated SubState=running for service alarmd.service in cache contents:
activeState=active subState=running result=success" }
+

If service does not support restart (Restart=no), then I see that we get
 Failed state DBUS notification , as  ' ActiveState=failed '.
Logs from my app which monitors dbus notifications:
++
Updated Result=signal for service alarmd.service in cache contents:
activeState=deactivating subState=stop-post result=signal" }
Updated ActiveState=failed for service alarmd.service in cache contents:
activeState=failed subState=stop-post result=signal" }
Updated SubState=failed for service alarmd.service in cache contents:
activeState=failed subState=failed result=signal" }
++
1) Is this a bug ?
2) Why service is put to failed state on sigkill ?
3) if putting service into failed state on sigkill is right thing to do ,
they why dbus notification  is not being sent in case of service that
supports restart.

Thanks for your help
-Prashant
___
systemd-devel mailing list
systemd-devel@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/systemd-devel