| *** ramishra_ is now known as ramishra | 04:37 | |
| *** ralonsoh is now known as ralonsoh_ooo | 16:33 | |
| elodilles | hi infa team o/ I'm planning to clean up some EOL'd branch if nothing is against it | 16:38 |
|---|---|---|
| clarkb | elodilles: we just upgraded our zookeeper cluster behind zuul to a new version. But that seems to be working as we are done with the upgrade. I have no other planned activity that might conflict | 16:39 |
| clarkb | I think its fine to proceed | 16:39 |
| elodilles | clarkb: thanks! then i'll run the script some time today | 16:39 |
| elodilles | actually, started it now :) let's see :) | 16:41 |
| elodilles | and these were deleted this time: https://paste.opendev.org/show/btPOii9kPp95SQMBP9rE/ | 17:27 |
| clarkb | elodilles: please see the note in #opendev about potentially optimizing the deletions to reduce zuul churn. It isn't critical but would probably be nice to have | 17:28 |
| clarkb | basically having the delete events happen as closely together in sequence as possible allows zuul to deduplicate them reducing config regeneration churn | 17:28 |
| JayF | I'll note that if my inbox's gerrit folder is any indication; the branch deletions for Ironic projects were landed in pretty close successsion | 17:29 |
| elodilles | clarkb: hmmm, well, that would need to refactor the script as it is looping sequentially over the deliverables that are marked as EOL and whenever there is any still open, then the script asks if that is OK to delete. so there are sime delay between deletions | 17:33 |
| elodilles | at least a couple of sec, and sometimes minutes | 17:33 |
| clarkb | elodilles: the issue is that every reconfiguration takes aobut 6-8 minutes judging from grafana | 17:33 |
| clarkb | so the fewer of those we can do the better | 17:33 |
| elodilles | JayF: this is not fully automated, we run a script that iterates over the eol'd deliverables and looks for still open branches | 17:34 |
| JayF | aha; I see, it's not on merge | 17:34 |
| elodilles | clarkb: hmmm. so then the script should somehow collect all the deletions and do it in one go at the end of the run, right? | 17:40 |
| clarkb | elodilles: that would be better for zuul responsiveness yes | 17:41 |
| clarkb | maybe there is some middle ground of doing smaller batches if we need to make it easier for humans to process the list though | 17:42 |
| clarkb | so instead of one branch deletion after another slowly happening we can do 5 at a time or something | 17:42 |
| elodilles | OK, I'll think about this & add a TODO to my list to refactor the script | 17:43 |
| wncslln | a | 18:32 |
| wncslln | a | 18:32 |
| wncslln | hi guys! in my infrastructure i want to enable instances with SGX, but im dont know a way to make it work with kolla-ansible. someone can help me? | 18:39 |
| *** haleyb is now known as haleyb|out | 22:30 | |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!