*** rcernin has quit IRC | 01:12 | |
*** rcernin has joined #openstack-swift | 01:17 | |
*** camelCaser has quit IRC | 02:23 | |
*** camelCaser has joined #openstack-swift | 02:25 | |
*** rcernin has quit IRC | 02:34 | |
*** rcernin has joined #openstack-swift | 02:36 | |
openstackgerrit | Merged openstack/swift master: Remove lxml deprecated methods https://review.opendev.org/737468 | 02:46 |
---|---|---|
*** psachin has joined #openstack-swift | 03:43 | |
openstackgerrit | Merged openstack/swift master: Bump up probe test timeout https://review.opendev.org/737385 | 03:44 |
*** rcernin has quit IRC | 03:46 | |
*** rcernin has joined #openstack-swift | 03:55 | |
*** rcernin has quit IRC | 04:04 | |
*** rcernin has joined #openstack-swift | 04:05 | |
*** evrardjp has quit IRC | 04:33 | |
*** evrardjp has joined #openstack-swift | 04:33 | |
*** m75abrams has joined #openstack-swift | 05:23 | |
*** manuvakery has joined #openstack-swift | 06:46 | |
*** gyee has quit IRC | 06:51 | |
*** rpittau|afk is now known as rpittau | 06:57 | |
*** rcernin has quit IRC | 07:39 | |
*** josephillips has quit IRC | 08:47 | |
*** manuvakery has quit IRC | 09:06 | |
*** tkajinam has quit IRC | 10:03 | |
*** rcernin has joined #openstack-swift | 10:10 | |
*** rpittau is now known as rpittau|bbl | 10:11 | |
*** rcernin has quit IRC | 10:15 | |
*** psachin has quit IRC | 10:57 | |
*** rcernin has joined #openstack-swift | 11:27 | |
*** rcernin has quit IRC | 11:34 | |
*** rcernin has joined #openstack-swift | 11:40 | |
*** rcernin has quit IRC | 12:08 | |
*** rpittau|bbl is now known as rpittau | 12:13 | |
*** rcernin has joined #openstack-swift | 12:35 | |
*** rcernin has quit IRC | 13:29 | |
*** ormandj has quit IRC | 13:58 | |
*** ormandj has joined #openstack-swift | 14:13 | |
*** m75abrams has quit IRC | 15:07 | |
timburke | ormandj, just wrote up https://bugs.launchpad.net/swift/+bug/1884991 | 15:47 |
openstack | Launchpad bug 1884991 in OpenStack Object Storage (swift) "s3api on py3 doesn't use the bytes-on-the-wire when calculating string to sign" [Undecided,New] | 15:47 |
*** gyee has joined #openstack-swift | 15:57 | |
*** rpittau is now known as rpittau|afk | 16:41 | |
ormandj | timburke: checking, sounds interesting ;) | 17:04 |
ormandj | yep. thank you! | 17:08 |
ormandj | we'll hold off on train/ussuri on proxies until we have a fix/plan in place | 17:08 |
ormandj | but once you're ready to have someone test for you with real-deal traffic, we're happy to help | 17:09 |
timburke | 👍 | 17:11 |
clayg | man, i'm struggling to load the s3_acl/auth stuff in s3 unittests into my head | 17:15 |
clayg | there's some *crazy* stuff going on down there | 17:15 |
clayg | like - I think the s3acl decorator maybe makes tests run *twice* does that sound even remotely correct?! | 17:15 |
openstackgerrit | Tim Burke proposed openstack/swift master: py3: Stop munging RAW_PATH_INFO https://review.opendev.org/737856 | 18:01 |
timburke | clayg, that's absolutely what it does -- top and bottom of https://github.com/openstack/swift/blob/2.25.0/test/unit/common/middleware/s3api/test_s3_acl.py#L79-L86 | 18:04 |
timburke | it's a little bit crazy | 18:04 |
timburke | and among the many reasons i'd love to find a path to no longer having that option -- to have the swift acl story become a superset of what's expressible via s3. but i've got this feeling like it's going to be a long, hard road -- and we've got a history of not really thinking hard about auth | 18:07 |
clayg | well what is with the fake_authorization - i'm getting a signature does not match error, but then if I set remote_user it's make a get_info request to the null namespace!? | 18:07 |
timburke | i've got this feeling like you're gonna want to take out https://github.com/openstack/swift/blob/2.25.0/test/unit/common/middleware/s3api/helpers.py#L66 -- i think we want to hit https://github.com/openstack/swift/blob/2.25.0/test/unit/common/middleware/s3api/helpers.py#L52 on both calls now | 18:11 |
timburke | but i haven't thought through everything else that might need to change as a result | 18:11 |
openstackgerrit | Tim Burke proposed openstack/swift master: s3api: Allow CompleteMultipartUpload requests to be retried https://review.opendev.org/733919 | 18:13 |
openstackgerrit | Tim Burke proposed openstack/swift master: s3api: test abort after complete https://review.opendev.org/737570 | 18:13 |
openstackgerrit | Tim Burke proposed openstack/swift master: Add py3 probe tests on CentOS 8 https://review.opendev.org/690717 | 18:16 |
clayg | tests are hard; let's go shopping | 18:31 |
*** irclogbot_1 has quit IRC | 18:34 | |
*** irclogbot_1 has joined #openstack-swift | 18:38 | |
clayg | oh, get_container_info does a HEAD requests to the null namespace to get bytes and stuff - so there you go | 18:47 |
timburke | clayg, why, that was more or less my take-away on p 737856, too! i started poking at tests for it... then realized that i wanted to round-trip through eventlet and stdlib, too... | 18:59 |
patchbot | https://review.opendev.org/#/c/737856/ - swift - py3: Stop munging RAW_PATH_INFO - 1 patch set | 18:59 |
timburke | and then i remembered that i have other things to look at like p 722570 | 18:59 |
patchbot | https://review.opendev.org/#/c/722570/ - swift - WIP: PLaying around with a basic shard-cleanup tool - 3 patch sets | 18:59 |
clayg | yay sharding! | 19:00 |
clayg | I think i can see a reasonable path forward for p 735738 - thanks @timburke | 19:01 |
patchbot | https://review.opendev.org/#/c/735738/ - swift - s3api: Don't do naive HEAD request for auth - 1 patch set | 19:01 |
timburke | i'm worried about shrinking :-( | 19:40 |
timburke | if i crank my reclaim_age down to like 10s, i start seeing failures. that alone probably isn't so surprising -- but i didn't expect it to happen at https://github.com/openstack/swift/blob/@%7B2020-06-24%7D/test/probe/test_sharder.py#L1494 | 19:41 |
timburke | swift-container-info on one of the shards gives me output like http://paste.openstack.org/show/795177/ | 19:42 |
timburke | so, account starts with .shards_, db has a delete timestamp, and it's supposedly a root container?? | 19:44 |
timburke | there are no objects in the db -- just a record of the acceptor shard range -- which is what's causing it to think that it has 4 objects | 19:45 |
timburke | almost meeting time! | 20:51 |
seongsoocho | morning~! | 21:00 |
clayg | timburke: i maybe have spoken too soon about s3api tests - I'm still confused about the test path forward | 21:00 |
alecuyer | here's what I have done so far to check the liberasurecode library issue: https://etherpad.opendev.org/p/ec_ubuntu_issue , sorry if that's already known and fixed and if missed something! | 22:09 |
timburke | alecuyer, no, that seems like a very real bug -- might be able to get the object-server working again by using LD_PRELOAD to ensure zlib gets loaded first | 22:14 |
alecuyer | ok I can try that | 22:15 |
timburke | still seems like we might want to look for some env var like LIBERASURECODE_WRITE_LEGACY_CRC; if set, write frags like pre-1.6.0 | 22:16 |
alecuyer | Or document the supported upgrade order? (this happened to us as a result of a broken IPMI and manual install of the wrong version on a lab machine, not planning to mix versions like that :) ) | 22:19 |
timburke | idk -- seems like the sort of thing where there isn't a great upgrade order at the moment. if we upgrade libec on proxy then object, proxy sends frags object can't read. going the other way works better, but as soon as you need to rebuild a frag, object will write a frag that proxy can't read | 22:29 |
*** rcernin has joined #openstack-swift | 22:42 | |
*** tkajinam has joined #openstack-swift | 22:51 |
Generated by irclog2html.py 2.17.2 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!