[openstack-dev] [tripleo] Log collection of overcloud nodes
flavio at redhat.com
Wed Mar 29 13:13:10 UTC 2017
On 29/03/17 12:10 +0200, Bogdan Dobrelya wrote:
>On 29.03.2017 00:57, Alex Schultz wrote:
>> Hey folks,
>> So as part of the capture-environment-status-and-logs blueprint,
>> I'm working on adding a single command to perform status and log
>> collection of the overcloud nodes via the tripleoclient that can be
>> used on the undercloud. I would like to bring up switching over to
>> this as part of our CI log collection activities as many of the
>That's a great effort, thank you for doing that! Personally, that is the
>number one topic I'd like to have smooth UX with. Not even to mention
>ops and production environments, when tripleo CI fails, devs get stuck
>with patches and would be nice them to get the way out *fast* and
>*easy*. Note, there is a CI scripts for getting some of the logs as
>well, and this improvement  for you kind review please.
>> relevant logs we want are already captured via the sosreport tool.
>> Additionally this is the way many operators are collecting and
>> reporting their logs when submitting issues.
>> I think this would benefit us to switch as sosreports also capture
>> additional status of the services at the time of the report and we
>> can improve sosreports via plugins to help diagnose frequent service
>> related problems. I believe we're duplicating some of the items
>> already covered via sosreport in tripleo-quickstart-extras and I
>> think it would be beneficial to not continue to duplicate this work
>> but rather use already available tooling. For CI once we have these
>> sosreport bundles, it would be fairly straight forward to only extract
>++ couldn't agree more on that (did you miss the link for the sosreport
>> relevant information for debugging use.
>> If you have some time, please review the outstanding reviews and
>> provide concerns around possibly switching over to relying on
>> sosreport for our log collection.
>I encourage everyone to join you and to do review as well. It is hard to
>overestimate the importance of logs collection for both dev and ops worlds!
>PS. Would be nice to adapt those sosreports and GSS diag tools w/
>plugins to automate collection and submission of known CI bugs for the
>elastic-recheck  status for tripleo. Not sure it that falls into
>that topic's scope though. But it is yet another important development
>workflow shortcut for getting out of (known) troubles fast.
+1 on all of the above! I think we're even just hitting some issues with the
inconsistencies between triple.sh and oooq in the different containers gates.
I love this effort, thanks for working on it.
-------------- next part --------------
A non-text attachment was scrubbed...
Size: 862 bytes
Desc: not available
More information about the OpenStack-dev