Thursday, 2023-10-05

opendevreviewOpenStack Proposal Bot proposed openstack/project-config master: Normalize projects.yaml
opendevreviewMerged openstack/project-config master: Revert "Temporarily remove release docs semaphores"
fricklernot sure why the proposal bot sent a no-change update on this one, normally it should detect this and not send a new patchset?
opendevreviewMerged openstack/project-config master: Add nested-virt-debian-bookworm
fricklerdoes anyone understand when gerrit shows merge conflicts and when and why it does not? on I was being shown five conflicts after submitting a +2, after a reload, they are no longer shown, same when I initially looked at it02:47
fricklerinfra-root: I'm hesitating to abandon because it sounds like this might actually still be relevant?03:05
*** diablo_rojo is now known as Guest221904:05
opendevreviewMerged openstack/project-config master: Clean up vestigial default-branch
fungifrickler: it has seemed to me like gerrit shows merge conflict on newly uploaded changes until it gets a chance to run its merge conflict check in the background. i'm not really sure though12:00
fungion 897273, yes the "New patch set was added with same tree, parent tree, and commit message..." is strange, i'll hunt down the build log for that in zuul to see if it explains why it did that12:02
fungifrickler: so, it did think there was a difference...
fungieven though gerrit then warned at push "no changes between prior commit f3b7bf3 and new commit 54cd8a1"12:09
fungiseeing if i can recreate that locally12:11
fungilooking through the build log, i don't see any indication that it actually checked out the previous change. it queries it and lists details, but doesn't seem to actually retrieve it12:19
fungiyeah, it should probably be an extra step in
fungiby comparison, does a `git review -d` of the existing change12:25
tkajinamI've not yet checked details but git repository in gerrit responds quite slowly for me, it seems13:13
tkajinamwondering if anyone else is seeing the same problem13:13
fungitkajinam: slow webui? slow to push changes to it? slow in some other way?13:16
fungibrowsing around different changes and queries, it's as snappy as usual for me13:16
fungiare you connecting to it over ipv4 or ipv6?13:17
tkajinamwebui works quite smooth. the problem is seen only in CLI so far, when git cli tires to fetch contents from git in review.opendev.org13:18
fungii just did a `git remote update` in a weeks-old openstack/nova checkout and it took roughly 3 seconds to complete (pulling in new tags for train-eol and the bobcat release)13:20
fungii'm connecting over ipv4 though, because of a long-standing routing issue in vexxhost where it seems like their bgp tables are missing a route to my isp13:20
fungimissing the ipv6 route i mean13:21
tkajinamgit pull gerrit master -4  completes in less than 1 second but -6 takes a few minutes. so the issue seems to be specific to ipv613:21
tkajinamI manipulated my local ssh config to use IPv4 to connect, which seems to be a valid solution so far. this might be an issue with network connectivity (in my local env or probably in vexxhost which you mentioned ?), instead of one in infra servers13:26
fungiyes, or anywhere on the internet in between, are you seeing packet loss or high round-trip latency when you ping6
fungi(compared to ipv4 ping)13:28
tkajinamaha. 100% packet loss when I ping by IPv6 :-D13:30
fungimnaser: guilhermesp: on a related note, has anyone had a chance to check your bgp tables and see if you're missing as20115's announcement for 2600:6c60::/32 ?13:31
tkajinamping -4 responds normally. I can ping other sites such as google by ping6 so I guess something with ipv6 route between me and is messed up I guess13:32
fungitkajinam: if you feel comfortable messaging me privately with your ipv6 address, i can test from the gerrit server end to see if it's acting like vexxhost is missing a return route to you13:32
fungiif so, you may have the same issue i've been seeing from home13:33
fungii opened a support ticket with vexxhost a month or so ago about it, but if there are more networks affected than just my isp's, that may indicate a more serious problem13:34
tkajinamI've not seen this issue earlier this month in my env. I replaced my laptop last week but afair I didn't face this problem when I was setting up the new one initially.13:40
fungimnaser: guilhermesp: i checked trying to reach tkajinam's ipv6 address from two of our servers in vexxhost ca-ymq-1 and it's behaving the same as trying to reach my home ipv6 address (routing loop at the next hop, back and forth between 2604:e100:1::1 and 2604:e100:1::2)13:43
fungibut from a server in vexxhost sjc1 it's fine, like for me13:44
*** blarnath is now known as d34dh0r5313:49
clarkbtoomer: its best to just ask here. Typically we try to have two core reviewers on system-config changes, but that has been more difficult these days. I will defer to fungi on whether or not he is comfortable landing that change with only my review14:40
clarkbfungi: question is when are we ready to merge ?14:41
fungiclarkb: toomer: i went ahead and approved it14:47
fungii just wanted to make sure i had some free time to keep an eye on the deploy job in case it doesn't go as expected14:47
fungiand yes, normally we want two sysadmins who aren't the author to review changes like that one, but we did announce it during the sysadmins meeting on tuesday and there's been time for folks to object14:49
clarkbGerrit User Community Meeting was short. Not sure I'm a fan of discord for voice calls yet, but it did work. Sounds like they are happy to ahve anyone edit the agenda going forward though so we can add our stuff there and I can try to attend (it is monthly first Thursday of the month at 8am pacific time)15:18
clarkbSounded like its a quiet time for the project as everyone recovers from post summit stuff. Hopefully we get more movement on docs updates and bugfixes in a week or two15:19
opendevreviewMerged opendev/system-config master: Add mailing list for Nordix environment
funginow to make sure that ^ doesn't encounter deployment issues15:22
toomerclarkb: Thanks for a help with pushing that trough. When the mailing list is up I will try to enrol on it. 15:28
fungitoomer: once zuul reports on that change with success from the deploy pipeline, i'll give it a quick look-over and let you know15:30
fungiinfra-prod-service-lists3 is running for it now15:40
fungitoomer: exists and i was able to subscribe to it15:46
fungiif you create an account with the sign-in button in the top-right corner, then you should automatically end up with list owner status for it once you've confirmed your account15:46
fungier, i guess the button says "sign up"15:47
fungii should have brought "Add Element X Ignition support for Matrix" in tuesday's meeting, i'll add it to next week's agenda so i don't forget16:01
opendevreviewClark Boylan proposed opendev/system-config master: Update gitea base OS during image builds
clarkbfungi: what is the tool you use for your mute button? its some X thing that listens for the button event then fires a script?18:24
fungiyep, xbindkeys18:25
fungiin my ~/.xbindkeysrc i just list the script i want run followed by its language for specifying the keyboard scancode to listen for18:27
fungialso i run xbindkeys at login from my ~/.xinitrc but your window manager may have its own methods for doing that18:28
fungior session manager more like18:28
toomerfungi: Thank you! I crated an account and I'm both the owner and moderator for the Nordix list18:57
toomerDo we have an OpenInfra Europe website ?18:59
Clark[m] I think. Note the OpenDev team doesn't run that19:04
fungitoomer: the foundation staff probably contract out management of that site, but i can ask around if you need to know how to make changes19:05
opendevreviewDr. Jens Harbott proposed openstack/project-config master: Retire opendev/puppet-openstack_health: remove project from infa
clarkbfungi: I went ahead and approved the gitea container image update. I think if you want to approve the element X change behind that that is fine then maybe aim for 1.20.5 tomorrow? (a day later than anticipated, but I'm realizing I have enough other things to do that two gitea updatse is plenty )19:49
opendevreviewMerged opendev/system-config master: Update gitea base OS during image builds
opendevreviewMerged opendev/system-config master: Add Element X Ignition support for Matrix
clarkbfungi: the matrix client file being served now is updated22:09
clarkbfwiw that didn't actually go in the image so I was mistaken22:09
clarkbwe're just serving that off of disk22:09
fungioh, yeah there's an ansible task to install it to the servers22:10
fungianyway, i'll take it back off next week's agenda22:10
opendevreviewClark Boylan proposed opendev/system-config master: Update the apache us filter set
fungispotted some new ones, eh?22:48
clarkbya when verifying the updates we just did I went "why is gitea09 so slow" -> "wow why si system load ~100" -> "I know probably the same old thing" sure enough found evidence of new friends22:50
clarkbthere were a bunch of git processes running to cat files and the like which maps pretty cleanly to commit specific requests that we get hit with22:51
clarkbnote that change adds regexes and exact matches. Should probably be checked carefully to make sure I didnd't do anything stupid adding regexes22:51
clarkbin particular I'm not sure if I need to escape some things now that I think of it22:51
fungii wonder if we could make a filter test with a file that's populated with all the bogus agent strings we see from our logs over time22:52
clarkbya I suspect the (s and )s will need escaping? or are they matched literally and \( and \) are used for groups?22:52
fungibasically stuff requests through with each string and make sure we get a 403 back22:53
clarkb`RewriteCond  "%{HTTP_USER_AGENT}"  "(iPhone|Blackberry|Android)"` <- is the example from the mod rewrite docs. Based on that I bet I have to escape the (s and )s22:53
opendevreviewClark Boylan proposed opendev/system-config master: Update the apache ua filter set
clarkbI read the docs more closely and I'm 99% sure that update is necessary23:03
fungiaha, you also fixed up the commit title, so i can delete my nit comment i hadn't saved yet23:03
clarkbya that bothered me as soon as I pushed it :)23:04
clarkbI'm going to have to punt the caracal signing key checking until tomorrow23:06
clarkbbut hoping to get to it then between the other stuff23:06
fungino worries, the release team will also make a test release as soon as it merges anyway, so we can double-check things are still working and validate as expected23:07
clarkbfwiw I think the gitea updates we landed applied properly across all of the servers despite these annoyances23:08
fungiload average of 100 is no match for ansible23:09
clarkbI also think it was largely disk bound which kept cpus free to do useful work23:10
clarkbfungi: I was looking to see if I could switch to element x and it seems to be mobile os only for now23:10
clarkbthere is no web app update yet.23:10
clarkbnot sure yet if I want to install yet another app to test things23:11
fungiprobably only apps for android and iphone too, nothing for mobian/phosh/etc23:12
clarkbya ios and android only23:12
clarkbyou might be able to emulate the android app on other platforms though23:13
clarkbwow when did apache docs get colorful diagrams23:15
fungithat's pretty slick23:15
fungii shouldn't have peeked behind the curtain23:20
fungiapache docs are assembled from xml sources with a mashup of perl and java23:20
fungiall in subversion, of course23:21
fungii suddenly no longer care to find out what tool they generated those diagrams with (if they are indeed generated at all)23:23
fungixmp:CreatorTool="Adobe Photoshop CS5.1 Windows"23:24
*** cloudnull7 is now known as cloudnull23:45

Generated by 2.17.3 by Marius Gedminas - find it at!