[openstack-dev] [oslo] How should oslo.service handle SIGINT?

Davanum Srinivas davanum at gmail.com
Tue Jul 21 10:57:49 UTC 2015


Elena,

The cleaner output was the norm before oslo.service was merged and if
there are no downsides, then we should fix oslo.service to do the
same.

thanks,
dims

On Tue, Jul 21, 2015 at 6:49 AM, Elena Ezhova <eezhova at mirantis.com> wrote:
> @Dims,
>
> The main advantage of having a handler is a more clean output in case a
> service is killed, which means there wouldn't be a stacktrace with
> KeyboardInterrupt exception, just messages like we have in nova now (nova
> "Switch to oslo.service patch" hadn't merged yet):
>
> 2015-07-21 10:46:06.765 INFO nova.openstack.common.service
> [req-b65dd702-4407-4442-941b-bdb45d935cfa None None] Caught SIGINT, stopping
> children
> 2015-07-21 10:46:06.779 INFO nova.openstack.common.service
> [req-b65dd702-4407-4442-941b-bdb45d935cfa None None] Waiting on 2 children
> to exit
> 2015-07-21 10:46:06.823 INFO nova.openstack.common.service
> [req-b65dd702-4407-4442-941b-bdb45d935cfa None None] Child 14490 killed by
> signal 15
> 2015-07-21 10:46:06.830 INFO nova.openstack.common.service
> [req-b65dd702-4407-4442-941b-bdb45d935cfa None None] Child 14491 killed by
> signal 15
>
> That looks more clean than KeyboardInterrupt:
> http://paste.openstack.org/show/395226/
>
>
> Elena
>
>
>
>
> On Mon, Jul 20, 2015 at 8:22 PM, Davanum Srinivas <davanum at gmail.com> wrote:
>>
>> @Elena,
>>
>> What are the pro's and con's of "having a handler" as this was the
>> existing behavior for all the projects that picked up code from
>> oslo-incubator for this behavior?
>>
>> -- dims
>>
>> On Mon, Jul 20, 2015 at 1:12 PM, Elena Ezhova <eezhova at mirantis.com>
>> wrote:
>> > Hi!
>> >
>> > Not so long ago oslo.service had a handler for SIGINT and on receiving
>> > this
>> > signal a process killed all children with SIGTERM and exited.
>> > Change [1], that added graceful shutdown on SIGTERM, removed all SIGINT
>> > handlers and currently all services that consume oslo.service generate
>> > KeyboardInterrupt exceptions on receiving SIGINT.
>> >
>> > My question is whether such behavior is what users and operators expect
>> > or
>> > having a handler is still the preferred way?
>> >
>> >
>> > Thanks, Elena
>> >
>> > [1]
>> >
>> > https://github.com/openstack/oslo.service/commit/fa9aa6b665f75e610f2b91a7d310f6499bd71770
>> >
>> >
>> > __________________________________________________________________________
>> > OpenStack Development Mailing List (not for usage questions)
>> > Unsubscribe:
>> > OpenStack-dev-request at lists.openstack.org?subject:unsubscribe
>> > http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
>> >
>>
>>
>>
>> --
>> Davanum Srinivas :: https://twitter.com/dims
>>
>> __________________________________________________________________________
>> OpenStack Development Mailing List (not for usage questions)
>> Unsubscribe: OpenStack-dev-request at lists.openstack.org?subject:unsubscribe
>> http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
>
>
>
> __________________________________________________________________________
> OpenStack Development Mailing List (not for usage questions)
> Unsubscribe: OpenStack-dev-request at lists.openstack.org?subject:unsubscribe
> http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
>



-- 
Davanum Srinivas :: https://twitter.com/dims



More information about the OpenStack-dev mailing list