| opendevreview | Merged zuul/zuul-operator master: Support zuul-preview https://review.opendev.org/c/zuul/zuul-operator/+/785760 | 00:02 |
|---|---|---|
| opendevreview | Merged zuul/zuul-operator master: Add support for zuul-registry https://review.opendev.org/c/zuul/zuul-operator/+/785761 | 00:02 |
| opendevreview | Merged zuul/zuul-operator master: Remove extra 2 minute wait from tests https://review.opendev.org/c/zuul/zuul-operator/+/785762 | 00:05 |
| opendevreview | Merged zuul/zuul-operator master: Add allowUnsafeConfig database setting https://review.opendev.org/c/zuul/zuul-operator/+/785764 | 00:05 |
| opendevreview | Merged zuul/zuul-operator master: Pass through environment to scheduler, web and launcher https://review.opendev.org/c/zuul/zuul-operator/+/785988 | 00:05 |
| opendevreview | Merged zuul/zuul-operator master: Allow terminationGracePeriodSeconds to be configurable https://review.opendev.org/c/zuul/zuul-operator/+/785989 | 00:06 |
| opendevreview | Merged zuul/zuul-operator master: Flake8 cleanups https://review.opendev.org/c/zuul/zuul-operator/+/786349 | 00:06 |
| opendevreview | Merged zuul/zuul-operator master: Fix error with multiple nodepool providers https://review.opendev.org/c/zuul/zuul-operator/+/799917 | 00:06 |
| opendevreview | Merged zuul/zuul-operator master: Add instructions and tools for running tests with kind https://review.opendev.org/c/zuul/zuul-operator/+/785763 | 00:06 |
| opendevreview | Merged zuul/zuul master: Route streams to different zones via finger gateway https://review.opendev.org/c/zuul/zuul/+/664965 | 00:51 |
| opendevreview | Merged zuul/zuul master: Use component registry in fingergw routing https://review.opendev.org/c/zuul/zuul/+/793666 | 00:51 |
| opendevreview | Ian Wienand proposed zuul/zuul-jobs master: configure-mirrors: don't install wheel mirror on CentOS 8 Stream https://review.opendev.org/c/zuul/zuul-jobs/+/802981 | 01:56 |
| *** marios is now known as marios|ruck | 05:05 | |
| *** rpittau|afk is now known as rpittau | 07:21 | |
| *** jcapitao is now known as jcapitao_lunch | 10:31 | |
| *** jcapitao_lunch is now known as jcapitao | 12:09 | |
| swest | corvus: found another bug in kazoo (read/write lock recipe) https://github.com/python-zk/kazoo/issues/649 | 12:38 |
| corvus | swest: neat -- are you working on a fix, or should i start on that? | 13:13 |
| swest | If we are ok with no prio for write locks the fix should be pretty straight forward. In case that's something we want, it get's a lot more complicated | 13:15 |
| corvus | swest: you mean priority between write locks (ie, strict ordering of write locks) or priority of write locks over read locks? | 13:18 |
| swest | corvus: prio of write locks over read locks | 13:18 |
| corvus | hrm, at least for our case, write-over-read priority does sound like it would be better doesn't it? | 13:19 |
| swest | talking about this now, there seems to be no real priority anyways as there can be multiple readers at the same time | 13:23 |
| swest | strike that. there is a priority in case there are multiple write locks waiting behind some waiting read locks | 13:24 |
| swest | so those read/write locks would then be processed strictly in order. | 13:25 |
| corvus | swest: well, maybe having write priority isn't that important, maybe it is better for read=write pr | 13:29 |
| corvus | er | 13:29 |
| corvus | sorry i was try to revise that to say that just getting anything that doesn't deadlock is a good start :) | 13:30 |
| swest | yea, I agree | 13:30 |
| swest | I'll try to come up with a fix | 13:31 |
| swest | maybe something we have to vendor similar to the watchers | 13:31 |
| corvus | apparently i wrote the original read/write lock impl; i can't tell if the bug was in the original or not. it's changed a bit since then. | 13:31 |
| corvus | well, i guess i didn't write it, i just took it over. | 13:32 |
| swest | I think the bug is also in the official Zookeeper recipe description for the "shared lock" | 13:32 |
| corvus | yeah, istr it followed it pretty closely | 13:33 |
| corvus | swest: it looks like the intent is that the read lock shouldn't be considering the second write lock as a predecessor, since it appears after the read lock. | 13:46 |
| swest | corvus: I wasn't sure about that as this is not documented anywhere. | 13:49 |
| corvus | swest: i think there has been an important change since the original | 13:50 |
| swest | reading the Zookeeper recipe doc https://zookeeper.apache.org/doc/r3.1.2/recipes.html#Shared+Locks it sounds like write locks should have priority over read locks | 13:51 |
| corvus | swest: somehow the current code (which does very complicated sorting/filtering with regex) returns contenders after the current contender. but i don't think the original code could do that. | 13:55 |
| corvus | swest: no -- in the recipe, the procedure for both the read and write lock include "having a lower sequence number" in the requirements | 13:57 |
| corvus | swest: so in all cases, a lock of any type should only consider a contender with a lower sequence to have the lock. | 13:58 |
| swest | corvus: true | 13:58 |
| corvus | i think the recipe description is correct -- the bug is that your contender #2 considers contender #3 to be a predecessor | 13:59 |
| swest | yes, I think the check for contender seq# > own seq# is missing | 13:59 |
| corvus | swest: it looks like this is supposed to handle it, but i suspect it's not working as expected? https://github.com/python-zk/kazoo/blob/master/kazoo/recipe/lock.py#L319-L321 | 14:01 |
| corvus | bit hard to say since it wall works on regexes (that's the major change since the original version i worked on) | 14:02 |
| corvus | * bit hard to say since it all works on regexes (that's the major change since the original version i worked on) | 14:02 |
| swest | I think the read lock will never be in the list of contenders (and with that found_self == False) as the regex never matches | 14:05 |
| corvus | won't that raise the forceretryerror? | 14:05 |
| swest | no, as it is checking for "is False" and found_self will be None in that case | 14:06 |
| *** rpittau is now known as rpittau|afk | 14:07 | |
| corvus | ah i see | 14:07 |
| corvus | then that seems likely to be the issue | 14:07 |
| corvus | that code is definitely a bear trap :) | 14:08 |
| -opendevstatus- NOTICE: There will be a brief outage of the Gerrit service on review.opendev.org starting at 15:00 UTC today as part of a routine project rename maintenance: http://lists.opendev.org/pipermail/service-announce/2021-July/000023.html | 14:13 | |
| opendevreview | Pierre-Louis Bonicoli proposed zuul/zuul master: smtp reporter: authenticate with client certificate https://review.opendev.org/c/zuul/zuul/+/755668 | 14:27 |
| -opendevstatus- NOTICE: There will be a brief outage of the Gerrit service on review.opendev.org in the next few minutes as part of a routine project rename maintenance: http://lists.opendev.org/pipermail/service-announce/2021-July/000023.html | 15:02 | |
| *** marios|ruck is now known as marios|out | 16:30 | |
| opendevreview | Matthieu Huin proposed zuul/zuul master: web UI: user login with OpenID Connect https://review.opendev.org/c/zuul/zuul/+/734082 | 17:43 |
| *** melwitt is now known as jgwentworth | 17:55 | |
Generated by irclog2html.py 2.17.2 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!