| ianw | I just installed "Logitech Options +" on osx to check the firmware version on my mouse for https://bugzilla.redhat.com/show_bug.cgi?id=2274393. It was 250Mb, wanted permissions for everything, and includes a "AI prompt generator" tool to ... ease my ability to make AI prompts. You can't even install a mouse driver without AI these days. It can do all this, but not scroll up and down properly! | 00:45 |
|---|---|---|
| ianw | this has nothing to do with anything. i just thought you'd like something to ponder while everyone is up waiting for the Tay Tay album to drop | 00:45 |
| fungi | sounds artificial to me | 01:14 |
| fungi | not intelligent, just artificial | 01:14 |
| opendevreview | Alberto Gonzalez proposed openstack/project-config master: Add new repository powertrain-build https://review.opendev.org/c/openstack/project-config/+/916038 | 07:12 |
| opendevreview | Alberto Gonzalez proposed openstack/project-config master: Add new repository powertrain-build https://review.opendev.org/c/openstack/project-config/+/916038 | 07:30 |
| * frickler had to look up what Tay Tay is ... and in fact I was waiting for a new album, but not that one, thx for the reminder anyway ;) | 07:48 | |
| ianw | :) | 07:49 |
| amoralej | hi, i'm hitting some issues while trying to pull content from opendev.org | 09:05 |
| amoralej | Downloading: http://opendev.org/openstack/os-refresh-config/archive/4b510162e573dbde407bc362b9cdfae74397d322.tar.gz#/os-refresh-config-4b51016.tar.gz | 09:05 |
| amoralej | Download failed: | 09:05 |
| amoralej | ('Connection aborted.', RemoteDisconnected('Remote end closed connection without response')) | 09:05 |
| amoralej | i got the same error several times | 09:06 |
| amoralej | Location: https://opendev.org/openstack/os-refresh-config/archive/4b510162e573dbde407bc362b9cdfae74397d322.tar.gz [following] | 09:10 |
| amoralej | --2024-04-19 05:07:31-- https://opendev.org/openstack/os-refresh-config/archive/4b510162e573dbde407bc362b9cdfae74397d322.tar.gz | 09:10 |
| amoralej | Connecting to opendev.org (opendev.org)|38.108.68.66|:443... connected. | 09:10 |
| amoralej | HTTP request sent, awaiting response... Read error (The TLS connection was non-properly terminated.) in headers. | 09:10 |
| amoralej | Retrying. | 09:10 |
| frickler | amoralej: I'm not sure we support this archive functionality. is there a reason that tarballs.opendev.org isn't enough for you? | 09:10 |
| frickler | other than that, this sounds like possibly an MTU issue, can you try to do a tracepath from your host to the site? | 09:11 |
| amoralej | https://review.opendev.org/c/openstack/releases/+/915987 i'm waiting for it ... | 09:11 |
| amoralej | oh, it's done now, sorry for the noise | 09:11 |
| amoralej | sure, lemme try | 09:11 |
| amoralej | it was working some minutes ago https://review.rdoproject.org/paste/show/b0IKr91IcwsyviXRwt4z/ | 09:13 |
| amoralej | anyway, i will move to the tarball | 09:13 |
| frickler | hmm, I'm astonished by how much bot traffic we actually see on gitea. we also don't seem to have any robots.txt file on that site? | 09:22 |
| amoralej | must be AI models learning from opendev :) | 09:23 |
| frickler | at least it looks like we have enough free space now on all gitea instances | 09:26 |
| amoralej | i'm having different results from differen places, so the issue may be in our infra side | 09:30 |
| amoralej | thanks frickler for your help!, i'll report this to our infra team | 09:31 |
| amoralej | it's strange | 09:32 |
| *** liuxie is now known as liushy | 09:33 | |
| fungi | frickler: which site don't we have a robots.txt on? do you mean https://opendev.org/robots.txt isn't loading for you? | 11:06 |
| fungi | also it explicitly includes "Disallow: /*/*/archive/" but it seems like some bots these days ignore robots.txt | 11:08 |
| frickler | fungi: https://opendev.org/robots.txt gives me a 404. also http://opendev.org/* redirects with a 307, not sure why this isn't a 302? | 11:13 |
| fungi | odd, https://opendev.org/robots.txt loads for me. maybe it's missing on some backends... | 11:14 |
| fungi | i'm able to load it successfully from each individual backend like https://gitea09.opendev.org:3000/robots.txt too | 11:18 |
| frickler | weird, it works in a browser for me, too. not sure what's different about curl, will need to dig closer | 11:22 |
| fungi | if i `curl https://opendev.org/robots.txt` i get the expected content back too | 11:25 |
| fungi | i agree i see the 307 temporary redirect response for http, likely needs an adjustment in haproxy config | 11:27 |
| opendevreview | Alberto Gonzalez proposed openstack/project-config master: Add new repository powertrain-build https://review.opendev.org/c/openstack/project-config/+/916038 | 11:27 |
| opendevreview | Alberto Gonzalez proposed openstack/project-config master: Add new repository powertrain-build https://review.opendev.org/c/openstack/project-config/+/916038 | 11:29 |
| opendevreview | Alberto Gonzalez proposed openstack/project-config master: Add new repository powertrain-build https://review.opendev.org/c/openstack/project-config/+/916038 | 11:32 |
| frickler | oh, the issue only happens with "curl -I" | 11:37 |
| fungi | that's odd | 11:38 |
| fungi | i confirm that behavior | 11:38 |
| opendevreview | Alberto Gonzalez proposed openstack/project-config master: Add new repository powertrain-build https://review.opendev.org/c/openstack/project-config/+/916038 | 11:47 |
| frickler | so since this happens both on the :3000 and :3081 URLs, it must be something gitea does. not sure though if it might be a regression or has always been that way? | 11:49 |
| frickler | it also doesn't affect other paths like / or /opendev | 11:50 |
| frickler | unrelated: does "copyCondition = changekind:NO_CODE_CHANGE OR is:ANY" make any sense? isn't that equivalent to just "is:ANY"? | 12:00 |
| frickler | cf. e.g. https://opendev.org/openstack/project-config/src/branch/master/gerrit/acls/openstack/kolla.config#L28 | 12:01 |
| frickler | the gitea/curl issue also doesn't affect other static files like https://gitea13.opendev.org:3081/assets/img/git.svg | 12:03 |
| frickler | o.k., we need to move the file, gitea logs: 2024/04/17 19:30:56 cmd/web.go:191:serveInstalled() [E] Found legacy public asset "robots.txt" in CustomPath. Please move it to /custom/public/robots.txt | 12:13 |
| frickler | also they have a special function to serve this file, not sure if that is related https://github.com/go-gitea/gitea/blob/eb24d973b036e4dddf505d8c12e905ecb1a688f9/routers/web/misc/misc.go#L36-L43 | 12:14 |
| frickler | so much fun on a friday, yay | 12:14 |
| opendevreview | Dr. Jens Harbott proposed opendev/system-config master: gitea: move robots.txt to public directory https://review.opendev.org/c/opendev/system-config/+/916420 | 12:25 |
| frickler | o.k., same effect on https://try.gitea.io/robots.txt , so nothing specific to our installation, guess I'll just report this as upstream issue | 12:25 |
| fungi | since we frontend them all with apache, we could serve it from there insteam | 12:30 |
| frickler | that could be a workaround, but let's see what they respond on https://github.com/go-gitea/gitea/issues/30601 first I'd say | 12:33 |
| fungi | er, instead | 12:34 |
| fungi | we could also use apache for the http redirect if we want | 12:34 |
| frickler | some robot seems to fetch the file every couple of seconds to make sure it didn't change :-/ | 12:34 |
| frickler | ah, interesting, http://try.gitea.io/ gives a "308 Permanent Redirect" | 12:35 |
| fungi | unrelated, we did get the 30-day warning for the wiki cert, as anticipated | 12:38 |
| frickler | seems gitea is hardcoded to use the 307, makes me wonder how they do different on their try instance https://github.com/go-gitea/gitea/blob/main/cmd/web.go#L85 | 12:47 |
| frickler | I wonder whether we should do both the redirect and the robot stuff directly on the lb https://cbonte.github.io/haproxy-dconv/2.2/configuration.html#4.2-http-request%20return | 12:54 |
| opendevreview | Dr. Jens Harbott proposed openstack/project-config master: Make Review-Priority sticky for QA repos https://review.opendev.org/c/openstack/project-config/+/916423 | 13:14 |
| frickler | kopecmartin: ^^ | 13:14 |
| frickler | that went fast ;) https://github.com/go-gitea/gitea/pull/30603 | 13:16 |
| fungi | frickler: we could do them on the lb or in the apache service that's running on each backend where we enforce the ua filters | 13:51 |
| * fungi disappears for a few hours, not really supposed to be here, don't tell my wife! ;) | 13:58 | |
| Clark[m] | We shouldn't host robots.txt upstream of gitea because gitea is still accessible and we want the same rules to apply all the way back to the source | 13:58 |
| Clark[m] | We could temporarily do the hosting upstream in addition to the back ends until they fix it. However the main problems are bad bots so not sure it will help us much | 14:00 |
| opendevreview | Takashi Kajinami proposed opendev/system-config master: reprepro: mirror Ubuntu UCA Caracal for Ubuntu Jammy https://review.opendev.org/c/opendev/system-config/+/916529 | 16:02 |
| opendevreview | Merged openstack/project-config master: Make Review-Priority sticky for QA repos https://review.opendev.org/c/openstack/project-config/+/916423 | 16:18 |
| opendevreview | Clark Boylan proposed openstack/project-config master: Move vexxhost/ansible-role-frrouting to openstack namespace https://review.opendev.org/c/openstack/project-config/+/910018 | 16:18 |
| clarkb | ok I think I got that rebase correct | 16:18 |
| clarkb | please double check if you have a chance cc noonedeadpunk | 16:18 |
| clarkb | https://etherpad.opendev.org/p/opendev-project-renames-20240422 I've put this process plan document together for renames on monday | 16:43 |
| opendevreview | Merged zuul/zuul-jobs master: Fix collection import error https://review.opendev.org/c/zuul/zuul-jobs/+/916303 | 17:30 |
| opendevreview | Rico Lin proposed zuul/zuul-jobs master: bump ensure-chart-testing dependencies https://review.opendev.org/c/zuul/zuul-jobs/+/916531 | 18:01 |
| diablo_rojo_phone | I'm seeing a timeout trying to access https://lists.opendev.org/mailman3/lists/ is that happening to anyone else? | 20:12 |
| fungi | diablo_rojo_phone: same, load average and memory utilization aren't high though, and i can ssh in easily enough | 20:16 |
| fungi | and now it's loading again | 20:16 |
| fungi | not sure what changed, any better for you? | 20:16 |
| clarkb | *** uWSGI listen queue of socket "0.0.0.0:8080" (fd: 8) full !!! (101/100) | 20:27 |
| clarkb | I think this is the issue | 20:27 |
| clarkb | and I notice we don't have a robots.txt but similar to other services have a number of crawlers poking at it | 20:27 |
| clarkb | I suspect that we want to add a robots.txt, add the UA filter, and then possibly increase the queue length for uwsgi | 20:28 |
| diablo_rojo_phone | Checking now fungi | 20:29 |
| diablo_rojo_phone | All good! | 20:30 |
| clarkb | looks like we bake the uwsgi.ini into our docker container image (to be inline with upstream). We could modify that to incrase the queue length. Though thinking about it more if we add a robots.txt and filter the bad ones at all then maybe we won't need a longer queue | 20:31 |
| fungi | i don't think we want to dissuade web search indexing of the ml archives, but otherwise sounds good | 20:32 |
| clarkb | agreed, I think its more to stop 500 erroring on the request so that the bots will cache the data rather than requestingi to ver and over again filling the queue | 20:33 |
| clarkb | the upstream mailing list server doesn't serve a robots.txt either though | 20:34 |
| opendevreview | Clark Boylan proposed opendev/system-config master: Add robots.txt to our list servers https://review.opendev.org/c/opendev/system-config/+/916543 | 20:51 |
| opendevreview | Clark Boylan proposed opendev/system-config master: Use UA filter in mailman apache vhost https://review.opendev.org/c/opendev/system-config/+/916544 | 20:51 |
| clarkb | something like that maybe. I've just noticed that my realted zuul chagnes are still not landed. We could do a big robots and UA filter update day soon | 20:51 |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!
