*** bhagyashris_ is now known as bhagyashris|rover | 07:08 | |
*** jpena|off is now known as jpena | 07:31 | |
*** jpena is now known as jpena|lunch | 11:29 | |
*** dviroel|out is now known as dviroel | 11:43 | |
*** jpena|lunch is now known as jpena | 12:22 | |
opendevreview | Martin Kopec proposed openstack/tempest master: Mark unstable test_create_object_with_transfer_encoding https://review.opendev.org/c/openstack/tempest/+/812256 | 12:30 |
---|---|---|
opendevreview | Lee Yarwood proposed openstack/devstack master: WIP libvirt: Increase keepalive_interval to workaround bug #1912310 https://review.opendev.org/c/openstack/devstack/+/798067 | 13:00 |
opendevreview | Lee Yarwood proposed openstack/devstack master: WIP zuul: Add openstack-two-node-centos-8-stream https://review.opendev.org/c/openstack/devstack/+/812350 | 13:07 |
opendevreview | Lee Yarwood proposed openstack/devstack master: WIP libvirt: Increase keepalive_interval to workaround bug #1912310 https://review.opendev.org/c/openstack/devstack/+/798067 | 13:09 |
*** haleyb|out is now known as haleyb | 13:25 | |
*** TheJulia is now known as needsleep | 13:49 | |
*** needsleep is now known as TheJulia | 13:49 | |
*** dviroel is now known as dviroel|doc_apt | 13:55 | |
*** pojadhav- is now known as pojadhav | 13:59 | |
*** akekane__ is now known as abhishekk | 14:42 | |
opendevreview | Dr. Jens Harbott proposed openstack/devstack master: Enable running on Debian Bullseye https://review.opendev.org/c/openstack/devstack/+/789083 | 15:33 |
opendevreview | Dr. Jens Harbott proposed openstack/devstack master: DNM: debugging ooms https://review.opendev.org/c/openstack/devstack/+/811203 | 15:33 |
*** pojadhav- is now known as pojadhav | 16:14 | |
*** dviroel|doc_apt is now known as dviroel | 16:16 | |
*** akekane_ is now known as abhishekk | 16:28 | |
*** jpena is now known as jpena|off | 16:34 | |
opendevreview | Lee Yarwood proposed openstack/devstack master: WIP nova: Ensure each compute uses a unique iSCSI initiator https://review.opendev.org/c/openstack/devstack/+/812391 | 17:08 |
dansmith | gmann: yoctozepto so... I'm trying to track down an issue which we actually saw a while ago, but I just realized is a problem | 17:31 |
dansmith | gmann: when using devstack and tls-proxy, when an upload to glance triggers a quota failure, it return 413, and for some reason apache is turning that into a 502 Proxy error | 17:31 |
dansmith | I think it's probably doing that because it's expecting something back from the proxy-ee, like a header or something, but when I curl reproduce the problem it looks totally legit to me | 17:32 |
dansmith | know anything about this or where I should start looking? | 17:32 |
yoctozepto | hmm | 17:32 |
yoctozepto | I wonder if it's some bug in apache | 17:33 |
yoctozepto | like, it can only generate 413 itself (which it could as a proxy) | 17:33 |
yoctozepto | interesting | 17:33 |
yoctozepto | anyhow, how's life without facebook? ;d | 17:34 |
dansmith | um :) | 17:35 |
dansmith | So you actually think that apache cares about 413 specifically or you're just speculating? | 17:37 |
opendevreview | Merged openstack/tempest master: Mark unstable test_create_object_with_transfer_encoding https://review.opendev.org/c/openstack/tempest/+/812256 | 17:41 |
yoctozepto | dansmith: I've seen enough to consider this a possibility; proxies tend to interpret errors from backends :-( | 18:02 |
dansmith | hrm | 18:03 |
dansmith | yoctozepto: looks specific to POST with a data stream hitting an error: https://bz.apache.org/bugzilla/show_bug.cgi?id=61090 | 18:25 |
dansmith | and sounds like resolving it is not straightforward | 18:25 |
dansmith | I wanted to make sure glanceclient and/or devstack aren't at fault here, but sounds like they're not | 18:26 |
dansmith | I sure wish glance would (have originally decided to) require declaring image sizes upfront before upload, but alas :/ | 18:26 |
yoctozepto | dansmith: nginx tends to be a better proxy; if you have spare cycles to provide this alternative, we would appreciate it :-) | 18:39 |
dansmith | uh huh :) | 18:40 |
dansmith | yoctozepto: maybe facebook should have used nginx? | 18:43 |
yoctozepto | dansmith: I doubt their issue is with http proxy; we know they stopped routing (sending bgp peerings) and that's the reason of outage | 18:48 |
dansmith | t'was just trying to loop back to your original jab | 18:48 |
dansmith | and was planning a response for you saying "they do use nginix" | 18:49 |
dansmith | and one for if you said "they also use apache" | 18:49 |
yoctozepto | lol | 18:49 |
dansmith | all the way to me suggesting Windows NT 4.0's IIS | 18:49 |
yoctozepto | oh my, I remember that monstrosity | 18:50 |
yoctozepto | and people were attaching php to that | 18:50 |
yoctozepto | kill it with fire | 18:50 |
dansmith | I bet you it's still running in places | 18:50 |
yoctozepto | I know | 18:50 |
yoctozepto | re nginx vs apache though - it's true; nginx does not carry legacy and its proxying capabilities are simply superior thanks to that | 18:51 |
yoctozepto | but we did not really require them until you discovered the bug | 18:52 |
dansmith | well, again from the bug discussion, I think it's a concern over http semantics | 18:52 |
dansmith | it sounds like if requests supported 100-continue we could get the proper behavio | 18:52 |
yoctozepto | hmm | 18:53 |
opendevreview | Lee Yarwood proposed openstack/devstack master: WIP nova: Ensure each compute uses a unique iSCSI initiator https://review.opendev.org/c/openstack/devstack/+/812391 | 19:18 |
opendevreview | Ade Lee proposed openstack/tempest master: WIP Add support for ecdsa keys https://review.opendev.org/c/openstack/tempest/+/807465 | 20:11 |
opendevreview | Ade Lee proposed openstack/tempest master: Add skip for tests that wont work under fips https://review.opendev.org/c/openstack/tempest/+/810808 | 20:11 |
opendevreview | Ade Lee proposed openstack/tempest master: DNM - Add temporary link to patched paramiko https://review.opendev.org/c/openstack/tempest/+/810809 | 20:11 |
Generated by irclog2html.py 2.17.2 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!