| *** diablo_rojo has joined #openstack-swift | 00:12 | |
| *** gyee has quit IRC | 00:19 | |
| *** diablo_rojo has quit IRC | 00:25 | |
| *** diablo_rojo__ has joined #openstack-swift | 00:25 | |
| openstackgerrit | Tim Burke proposed openstack/swift master: py3: mostly port s3 func tests https://review.opendev.org/674716 | 00:37 |
|---|---|---|
| *** diablo_rojo__ has quit IRC | 00:54 | |
| openstackgerrit | Tim Burke proposed openstack/swift master: py3: mostly port s3 func tests https://review.opendev.org/674716 | 00:55 |
| *** spsurya has joined #openstack-swift | 01:17 | |
| *** altlogbot_2 has quit IRC | 01:37 | |
| *** altlogbot_3 has joined #openstack-swift | 01:39 | |
| *** BjoernT has joined #openstack-swift | 02:41 | |
| *** zaitcev__ has joined #openstack-swift | 02:45 | |
| *** ChanServ sets mode: +v zaitcev__ | 02:45 | |
| *** zaitcev_ has quit IRC | 02:49 | |
| *** BjoernT_ has joined #openstack-swift | 02:58 | |
| *** BjoernT has quit IRC | 02:59 | |
| *** tdasilva_ has quit IRC | 03:14 | |
| *** tdasilva_ has joined #openstack-swift | 03:14 | |
| *** ChanServ sets mode: +v tdasilva_ | 03:14 | |
| *** zaitcev_ has joined #openstack-swift | 03:45 | |
| *** ChanServ sets mode: +v zaitcev_ | 03:45 | |
| *** zaitcev__ has quit IRC | 03:49 | |
| *** thiago__ has joined #openstack-swift | 04:02 | |
| *** ChanServ sets mode: +v thiago__ | 04:02 | |
| *** tdasilva_ has quit IRC | 04:04 | |
| *** BjoernT_ has quit IRC | 04:12 | |
| *** thiago__ has quit IRC | 05:15 | |
| *** thiago__ has joined #openstack-swift | 05:15 | |
| *** ChanServ sets mode: +v thiago__ | 05:15 | |
| *** tkajinam has quit IRC | 05:38 | |
| *** tkajinam has joined #openstack-swift | 05:38 | |
| *** gkadam_ has joined #openstack-swift | 05:54 | |
| *** gkadam_ has quit IRC | 05:54 | |
| *** gkadam has quit IRC | 05:56 | |
| *** tesseract has joined #openstack-swift | 07:30 | |
| *** zaitcev_ has quit IRC | 07:37 | |
| *** zaitcev_ has joined #openstack-swift | 07:51 | |
| *** ChanServ sets mode: +v zaitcev_ | 07:51 | |
| *** mikecmpbll has joined #openstack-swift | 07:55 | |
| *** rcernin has quit IRC | 08:04 | |
| *** tkajinam has quit IRC | 08:06 | |
| *** ccamacho has joined #openstack-swift | 08:07 | |
| *** ccamacho has quit IRC | 08:08 | |
| *** ccamacho has joined #openstack-swift | 08:08 | |
| *** e0ne has joined #openstack-swift | 08:35 | |
| *** zaitcev_ has quit IRC | 08:56 | |
| *** zaitcev_ has joined #openstack-swift | 09:09 | |
| *** ChanServ sets mode: +v zaitcev_ | 09:09 | |
| *** thiago__ has quit IRC | 09:50 | |
| *** tdasilva has joined #openstack-swift | 11:03 | |
| *** ChanServ sets mode: +v tdasilva | 11:03 | |
| *** henriqueof has joined #openstack-swift | 12:11 | |
| *** mvkr has quit IRC | 12:21 | |
| *** zaitcev_ has quit IRC | 12:47 | |
| *** zaitcev_ has joined #openstack-swift | 12:59 | |
| *** ChanServ sets mode: +v zaitcev_ | 12:59 | |
| *** mvkr has joined #openstack-swift | 13:11 | |
| *** zaitcev_ has quit IRC | 13:17 | |
| *** zaitcev_ has joined #openstack-swift | 13:29 | |
| *** ChanServ sets mode: +v zaitcev_ | 13:29 | |
| *** BjoernT has joined #openstack-swift | 13:40 | |
| *** BjoernT has quit IRC | 13:44 | |
| *** BjoernT has joined #openstack-swift | 13:48 | |
| *** ndk_ has quit IRC | 13:51 | |
| *** altlogbot_3 has quit IRC | 14:12 | |
| *** altlogbot_2 has joined #openstack-swift | 14:15 | |
| *** zaitcev_ has quit IRC | 14:39 | |
| *** BjoernT has quit IRC | 14:42 | |
| *** zaitcev_ has joined #openstack-swift | 14:52 | |
| *** ChanServ sets mode: +v zaitcev_ | 14:52 | |
| *** zaitcev__ has joined #openstack-swift | 15:01 | |
| *** ChanServ sets mode: +v zaitcev__ | 15:01 | |
| *** diablo_rojo has joined #openstack-swift | 15:02 | |
| *** zaitcev_ has quit IRC | 15:05 | |
| *** BjoernT has joined #openstack-swift | 15:10 | |
| *** ccamacho has quit IRC | 15:15 | |
| *** BjoernT has quit IRC | 15:23 | |
| *** hoonetorg has quit IRC | 15:24 | |
| *** hoonetorg has joined #openstack-swift | 15:25 | |
| *** gyee has joined #openstack-swift | 15:29 | |
| *** BjoernT has joined #openstack-swift | 15:33 | |
| *** mrjk__ has quit IRC | 15:53 | |
| *** mvkr has quit IRC | 16:20 | |
| *** mikecmpbll has quit IRC | 16:31 | |
| *** tdasilva has quit IRC | 16:32 | |
| *** e0ne has quit IRC | 16:48 | |
| *** BjoernT has quit IRC | 16:48 | |
| openstackgerrit | Tim Burke proposed openstack/swift master: py3: mostly port s3 func tests https://review.opendev.org/674716 | 16:50 |
| openstackgerrit | Tim Burke proposed openstack/swift master: py3: Add swift-tox-func-encryption-py37 gate job https://review.opendev.org/674878 | 17:01 |
| *** mrjk has joined #openstack-swift | 17:02 | |
| *** BjoernT has joined #openstack-swift | 17:08 | |
| openstackgerrit | Tim Burke proposed openstack/swift master: Use `is` to compare against sentinel object https://review.opendev.org/674883 | 17:15 |
| clayg | timburke: so what do you think about ripping the SLO brains out of symlink - and clients creating hardlinks don't get slo container listing bits in the bucket listing containing the hardlink? | 17:22 |
| clayg | The alternatives being 1) *sometimes* clients creating hardlinks don't get slo container listing bits in the bucket listing or 2) we teach symlinks to get container listing bits from upstream middlewares via callback hooks | 17:23 |
| *** tdasilva has joined #openstack-swift | 17:26 | |
| clayg | my pipeline being `copy slo dlo versioned_writes symlink` means if versioned writes *does* want to get SLO bits in container listing it still has to learn how to handle legacy SLOs or do the callback bits - so i'm not sure "don't make SLO learn about symlinks" really makes the problem simpler as much as moves it around to another middleware | 17:26 |
| *** ChanServ sets mode: +v tdasilva | 17:26 | |
| timburke | i'm partial toward alternative 1, personally | 17:28 |
| timburke | not such a fan of ripping it out if only because the listings feel *so* different between the SLO being pointed at, the client-created hardlink, and the VW-created hardlink | 17:28 |
| timburke | not a fan of adding more callbacks as we know from experience (with authorize and copy_hook) that properly chaining callbacks all the time is hard | 17:28 |
| timburke | but maybe i've just done a poor job of organizing callbacks? *shrug* | 17:30 |
| clayg | no, I 100% agree the callback ordering is *super* hard - it can be a very frail interface | 17:31 |
| clayg | on the "sometimes client's hardlinks ..." - can we define "sometimes" by some quantifiable client oversable difference? | 17:31 |
| clayg | like ... "hey i created a hardlink to this object and it's not doing the listings right" - then we HEAD the object and say "yeah, you need make a COPY request over itself and recreate the hardlink" | 17:32 |
| clayg | hell, at that point maybe SLO's call back is more like "x-newest copy/fix this SLO" 😬 | 17:34 |
| *** diablo_rojo has quit IRC | 17:45 | |
| *** spsurya has quit IRC | 17:52 | |
| *** diablo_rojo has joined #openstack-swift | 17:55 | |
| openstackgerrit | Tim Burke proposed openstack/swift master: py3: decode stdout from backgrounded servers https://review.opendev.org/674705 | 17:58 |
| openstackgerrit | Tim Burke proposed openstack/swift master: s3api: Translate 503 to S3-style 503s https://review.opendev.org/674901 | 18:12 |
| *** e0ne has joined #openstack-swift | 18:17 | |
| * timburke whispers "audit-watcher-driven data migrations" | 18:18 | |
| tdasilva | lol | 18:26 |
| *** e0ne has quit IRC | 18:31 | |
| *** tdasilva has quit IRC | 18:33 | |
| *** tdasilva has joined #openstack-swift | 18:33 | |
| *** ChanServ sets mode: +v tdasilva | 18:33 | |
| *** henriqueof has quit IRC | 18:38 | |
| *** tesseract has quit IRC | 18:52 | |
| *** zaitcev_ has joined #openstack-swift | 18:54 | |
| *** ChanServ sets mode: +v zaitcev_ | 18:54 | |
| *** zaitcev__ has quit IRC | 18:57 | |
| *** e0ne has joined #openstack-swift | 19:04 | |
| *** e0ne has quit IRC | 19:13 | |
| *** e0ne has joined #openstack-swift | 19:14 | |
| *** e0ne has quit IRC | 19:15 | |
| *** zaitcev_ has quit IRC | 19:27 | |
| clayg | timburke: so prior to swift 2.19.0 SLOs didn't put the slo_etag in the container listing | 19:39 |
| clayg | after c4c98eb64d3d99a5fb58f0390600f3e7755da1b2 it did, and it writes it down in X-Object-Sysmeta-Container-Update-Override-Etag - but you can't see that just doing a HEAD on the SLO/manifest | 19:40 |
| *** zaitcev_ has joined #openstack-swift | 19:40 | |
| *** ChanServ sets mode: +v zaitcev_ | 19:40 | |
| clayg | so a client can't know if a SLO has an entry in the listing except for ... going through the listing with prior knowledge of which objects are SLOs and infering from the missing slo_etag key that it was created < 2.19.0 | 19:41 |
| clayg | what's interesting is that even in c4c98eb64d3d99a5fb58f0390600f3e7755da1b2^ we DO have the X-Object-Sysmeta-Slo-Etag and X-Object-Sysmeta-Slo-Size keys in sysmeta, so symlinks could still *build* the X-Object-Sysmeta-Container-Update-Override-Etag if it wanted to 🤷♂️ | 19:42 |
| *** henriqueof has joined #openstack-swift | 19:42 | |
| clayg | that goes all the way back to 2.12 2d25fe6ad3 | 19:44 |
| clayg | and even before then we we might still be able to pull swift_bytes out of content-type 🤔 | 19:48 |
| *** e0ne has joined #openstack-swift | 19:51 | |
| *** diablo_rojo has quit IRC | 20:12 | |
| *** tdasilva has quit IRC | 20:14 | |
| *** tdasilva has joined #openstack-swift | 20:15 | |
| *** ChanServ sets mode: +v tdasilva | 20:15 | |
| timburke | clayg, 😞 https://logs.opendev.org/16/674716/3/check/swift-tox-py37/e2bbe2b/job-output.txt.gz#_2019-08-06_20_24_07_462775 | 20:31 |
| clayg | WOW!!! | 20:31 |
| clayg | so close 2019-08-06 20:24:07.463279 | ubuntu-bionic | cont2 INFO: 127.0.0.1 - - [06/Aug/2019:20:23:30 +0000] "PUT /sdb1/0/a/versions" 201 - "PUT http://localhost/v1/a/versions" "tx881c24bdbaae49e79ecd2-005d49e1ae" "proxy-server 7197" 20.1892 "-" 7197 0 | 20:32 |
| timburke | looks like it was *really* close tho -- 2019-08-06 20:23:10.101014 vs 2019-08-06 20:23:30.328439 | 20:32 |
| clayg | let's just go for 60s and call it a win | 20:33 |
| clayg | at least we know the plumbing is working 🤣 | 20:33 |
| *** zaitcev_ has quit IRC | 20:34 | |
| timburke | oh good, you're right -- obj server *does* send swift_bytes in the content-type, it's the proxy server app that drops it: https://github.com/openstack/swift/blob/2.22.0/swift/proxy/controllers/obj.py#L251-L253 | 20:38 |
| *** zaitcev_ has joined #openstack-swift | 20:48 | |
| *** ChanServ sets mode: +v zaitcev_ | 20:48 | |
| *** e0ne has quit IRC | 20:48 | |
| *** tdasilva_ has joined #openstack-swift | 20:53 | |
| *** ChanServ sets mode: +v tdasilva_ | 20:53 | |
| *** tdasilva has quit IRC | 20:56 | |
| openstackgerrit | Tim Burke proposed openstack/swift master: py3: Add swift-tox-func-encryption-py37 gate job https://review.opendev.org/674878 | 20:57 |
| *** zaitcev_ has quit IRC | 21:05 | |
| openstackgerrit | Tim Burke proposed openstack/python-swiftclient stable/stein: Fix up stable gate https://review.opendev.org/674180 | 21:15 |
| *** zaitcev_ has joined #openstack-swift | 21:17 | |
| *** ChanServ sets mode: +v zaitcev_ | 21:17 | |
| timburke | i think the most compelling reason to finish off the py3 func test series is so we can say, "ok, we have reason to believe they all pass on both -- let's split it up so *these* run on py2 and *those* on py3" | 21:21 |
| timburke | 'cause doubling the number of func tests in the mean time dramatically increases the risk of needing a recheck | 21:21 |
| timburke | ok, maybe not *so* dramatic. but still, roughly doubling the risk sucks | 21:26 |
| *** henriqueof has quit IRC | 21:45 | |
| openstackgerrit | Tim Burke proposed openstack/swift master: s3api: Translate 503 to S3-style 503s https://review.opendev.org/674901 | 21:53 |
| *** henriqueof has joined #openstack-swift | 22:43 | |
| *** tkajinam has joined #openstack-swift | 22:51 | |
| *** threestrands has joined #openstack-swift | 23:24 | |
| *** rcernin has joined #openstack-swift | 23:26 | |
| mattoliverau | morning | 23:45 |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!