| *** amandap has quit IRC | 00:22 | |
| *** amandap has joined #openstack-swift | 00:23 | |
| *** amandap has quit IRC | 00:31 | |
| *** germs has joined #openstack-swift | 00:45 | |
| *** germs has quit IRC | 00:50 | |
| *** germs has joined #openstack-swift | 00:54 | |
| *** threestrands has joined #openstack-swift | 01:12 | |
| *** amandap has joined #openstack-swift | 01:23 | |
| *** armaan has quit IRC | 01:30 | |
| *** armaan has joined #openstack-swift | 01:31 | |
| *** armaan has quit IRC | 02:21 | |
| *** armaan has joined #openstack-swift | 02:21 | |
| *** germs has quit IRC | 02:38 | |
| *** psachin has joined #openstack-swift | 03:10 | |
| *** amandap has quit IRC | 03:10 | |
| *** amandap has joined #openstack-swift | 03:17 | |
| *** germs has joined #openstack-swift | 04:10 | |
| *** germs has quit IRC | 04:15 | |
| *** threestrands has quit IRC | 04:18 | |
| *** threestrands has joined #openstack-swift | 04:22 | |
| *** threestrands has quit IRC | 04:22 | |
| *** threestrands has joined #openstack-swift | 04:22 | |
| *** threestrands has quit IRC | 04:23 | |
| *** threestrands has joined #openstack-swift | 04:23 | |
| *** amandap has quit IRC | 04:38 | |
| *** amandap has joined #openstack-swift | 04:39 | |
| *** cshastri has joined #openstack-swift | 04:45 | |
| *** geaaru_ has quit IRC | 05:36 | |
| *** armaan_ has joined #openstack-swift | 05:57 | |
| *** armaan has quit IRC | 05:57 | |
| *** links has joined #openstack-swift | 05:57 | |
| *** germs has joined #openstack-swift | 06:11 | |
| *** germs has quit IRC | 06:11 | |
| *** germs has joined #openstack-swift | 06:11 | |
| *** germs has quit IRC | 06:16 | |
| *** amandap has quit IRC | 06:20 | |
| *** itlinux has joined #openstack-swift | 06:33 | |
| *** saltsa has joined #openstack-swift | 06:47 | |
| *** pcaruana has joined #openstack-swift | 06:47 | |
| *** ntt has quit IRC | 07:00 | |
| *** ntt has joined #openstack-swift | 07:04 | |
| *** rcernin has quit IRC | 07:06 | |
| *** ccamacho has joined #openstack-swift | 07:10 | |
| *** tesseract has joined #openstack-swift | 07:10 | |
| *** ccamacho has quit IRC | 07:15 | |
| *** ccamacho has joined #openstack-swift | 07:16 | |
| *** amandap has joined #openstack-swift | 07:25 | |
| acoles | good morning | 07:32 |
|---|---|---|
| acoles | hope everyone in Vancouver has a great week | 07:33 |
| *** geaaru has joined #openstack-swift | 07:37 | |
| openstackgerrit | Alistair Coles proposed openstack/swift master: Use correct policy when faking misplaced objects in probe test https://review.openstack.org/569737 | 07:38 |
| acoles | timburke: that's a fix for the problem you ran into with probe test failing ^^ | 07:39 |
| *** pcaruana has quit IRC | 07:40 | |
| *** pcaruana has joined #openstack-swift | 07:41 | |
| *** amandap has quit IRC | 07:48 | |
| *** amandap has joined #openstack-swift | 07:48 | |
| *** silor has joined #openstack-swift | 07:59 | |
| *** threestrands has quit IRC | 08:00 | |
| *** itlinux has quit IRC | 08:04 | |
| *** germs has joined #openstack-swift | 08:12 | |
| *** germs has quit IRC | 08:12 | |
| *** germs has joined #openstack-swift | 08:12 | |
| *** germs has quit IRC | 08:17 | |
| *** amandap has quit IRC | 08:22 | |
| *** amandap has joined #openstack-swift | 08:22 | |
| *** amandap has quit IRC | 08:28 | |
| *** amandap has joined #openstack-swift | 08:29 | |
| *** silor has quit IRC | 08:33 | |
| *** bkopilov has joined #openstack-swift | 08:34 | |
| *** amandap has quit IRC | 08:35 | |
| *** amandap has joined #openstack-swift | 08:52 | |
| *** amandap has quit IRC | 08:57 | |
| *** armaan_ has quit IRC | 09:37 | |
| *** armaan has joined #openstack-swift | 09:38 | |
| *** bkopilov has quit IRC | 10:08 | |
| *** mikecmpbll has joined #openstack-swift | 10:11 | |
| *** germs has joined #openstack-swift | 10:13 | |
| *** armaan has quit IRC | 10:17 | |
| *** armaan has joined #openstack-swift | 10:17 | |
| *** germs has quit IRC | 10:18 | |
| *** armaan has quit IRC | 10:22 | |
| *** armaan has joined #openstack-swift | 10:23 | |
| *** kei_yama has quit IRC | 10:24 | |
| openstackgerrit | Alistair Coles proposed openstack/swift master: Container sharding doc fixes and clarifications https://review.openstack.org/569773 | 10:37 |
| *** psachin has quit IRC | 10:43 | |
| openstackgerrit | Alistair Coles proposed openstack/swift master: Container sharding doc fixes and clarifications https://review.openstack.org/569773 | 10:59 |
| *** psachin has joined #openstack-swift | 11:04 | |
| *** armaan has quit IRC | 11:22 | |
| *** armaan has joined #openstack-swift | 11:22 | |
| *** bkopilov has joined #openstack-swift | 11:28 | |
| *** amandap_ has joined #openstack-swift | 11:53 | |
| *** NM has joined #openstack-swift | 12:10 | |
| remix_tj | acoles: if i want to store a custom metadata on an object, which name should i use? X-Object-Systemeta-* ? or i can specify my own X-whatever? | 12:13 |
| *** germs has joined #openstack-swift | 12:14 | |
| remix_tj | or X-Meta? | 12:14 |
| remix_tj | X-Object-Meta-* ? | 12:14 |
| *** mikecmpbll has quit IRC | 12:18 | |
| *** germs has quit IRC | 12:18 | |
| acoles | remix_tj: that depends upon the context - a client can only set x-object-meta, whereas middleware can set x-object-sysmeta. Also x-object-sysmeta is not visible to clients. So x-object-sysmeta can be used for internal metadata, whereas x-object-meta is for clients. | 12:18 |
| remix_tj | i want give user the ability to say "This file cannot be altered or deleted for X time" | 12:19 |
| acoles | remix_tj: https://docs.openstack.org/swift/latest/development_middleware.html#swift-metadata | 12:19 |
| remix_tj | so i think X-Object-Meta is the appropriate | 12:19 |
| remix_tj | i'll take a look. I need also to find when the file has been created, so i can calculate when a file can be deleted. I've seen that X-Delete-After is converted to X-Delete-At converting the seconds to a timestamp, am i right? | 12:21 |
| *** mikecmpbll has joined #openstack-swift | 12:23 | |
| acoles | remix_tj: for that you might want to consider translating a client header e.g. 'X-do-such-and-such' to x-object-sysmeta-do-such-and-such in middleware. Otherwise you have to 'reserve' a user metadata name for internal purposes. There's examples of this pattern in other middleware e.g. https://github.com/openstack/swift/blob/097e975938befb939fa6821f50c061e2c7f42cef/swift/common/middleware/symlink.py#L417-L454 | 12:24 |
| remix_tj | acoles: thank you | 12:25 |
| acoles | Also, you should be aware of the persistence properties of various object metadata - see that doc ^^ | 12:25 |
| remix_tj | yes, that's what i'm reading now | 12:26 |
| acoles | remix_tj: re x-delete-at - correct, internally x-delete-after is converted to absolute x-delete-at time | 12:26 |
| remix_tj | anyway in my idea if an object hasn't expired it's retention time, any altering operation is denied | 12:27 |
| *** psachin has quit IRC | 12:29 | |
| *** cshastri has quit IRC | 12:31 | |
| *** psachin has joined #openstack-swift | 12:37 | |
| *** ndk_ has quit IRC | 12:45 | |
| *** ndk_ has joined #openstack-swift | 12:46 | |
| *** vinsh has quit IRC | 12:55 | |
| *** vinsh has joined #openstack-swift | 12:55 | |
| *** spsurya has joined #openstack-swift | 12:58 | |
| *** portdirect has quit IRC | 13:04 | |
| *** portdirect has joined #openstack-swift | 13:04 | |
| openstackgerrit | Alistair Coles proposed openstack/swift master: Improve building listings from shards https://review.openstack.org/569796 | 13:23 |
| *** silor has joined #openstack-swift | 13:26 | |
| *** mikecmpbll has quit IRC | 13:26 | |
| *** silor has quit IRC | 13:32 | |
| *** NM has quit IRC | 13:36 | |
| *** cshastri has joined #openstack-swift | 13:38 | |
| *** NM has joined #openstack-swift | 13:38 | |
| *** NM has quit IRC | 13:39 | |
| *** mvenesio has joined #openstack-swift | 13:54 | |
| openstackgerrit | Alistair Coles proposed openstack/swift master: Maintain manage-shard-ranges docs in its module https://review.openstack.org/569799 | 14:01 |
| acoles | kota_: ^^ this is something you suggested during review | 14:01 |
| *** armaan has quit IRC | 14:02 | |
| *** armaan has joined #openstack-swift | 14:03 | |
| *** tesseract-RH has joined #openstack-swift | 14:09 | |
| *** tesseract has quit IRC | 14:12 | |
| *** germs has joined #openstack-swift | 14:15 | |
| *** germs has joined #openstack-swift | 14:15 | |
| *** armaan has quit IRC | 14:17 | |
| *** armaan has joined #openstack-swift | 14:17 | |
| *** zhongjun_ has quit IRC | 14:18 | |
| *** germs has quit IRC | 14:20 | |
| *** amandap_ has quit IRC | 14:23 | |
| *** NM has joined #openstack-swift | 14:30 | |
| *** NM has quit IRC | 14:33 | |
| *** NM has joined #openstack-swift | 14:36 | |
| remix_tj | acoles: last question for today and then i'll will not bother you for some days: is there any metadata reporting when an object has been created? I see X-Object-Meta-mtime but to me "mtime" means "modification time". I'm looking for something like "ctime" | 14:37 |
| *** NM has quit IRC | 14:40 | |
| *** germs has joined #openstack-swift | 14:41 | |
| *** SkyRocknRoll has joined #openstack-swift | 14:41 | |
| *** NM has joined #openstack-swift | 14:44 | |
| *** NM has joined #openstack-swift | 14:45 | |
| acoles | remix_tj: IIRC X-Object-Meta-mtime is actually sent by swiftclient. Internal to swift the X-Backend-Data-Timestamp is the time at which the object was PUT. X-Backend-Timestamp is the time at which the last POST or PUT occurred. The two will be equal if there has not been a POST. You should cast the string header values to a utils.Timestamp instance. | 14:45 |
| *** NM has quit IRC | 14:47 | |
| *** SkyRocknRoll has quit IRC | 14:47 | |
| remix_tj | acoles: in my idea is that there is a container level default retention value. So if there is no specific metadata set on the object, then the retention time is defaulted to object_creation_time + container_default_retention_value | 14:49 |
| remix_tj | but i don't want to set metadata on each object, i want to be able also to compute on the fly | 14:49 |
| remix_tj | i'm getting the object headers with: swift.proxy.controllers.base.get_object_info | 14:50 |
| *** itlinux has joined #openstack-swift | 14:51 | |
| *** links has quit IRC | 14:51 | |
| *** NM has joined #openstack-swift | 14:56 | |
| *** SkyRocknRoll has joined #openstack-swift | 14:56 | |
| *** itlinux has quit IRC | 14:59 | |
| *** germs has quit IRC | 15:13 | |
| *** tesseract-RH has quit IRC | 15:16 | |
| *** pcaruana has quit IRC | 15:18 | |
| *** cshastri has quit IRC | 15:20 | |
| acoles | remix_tj: hmm, looks like the backend timestamps don't make it into the object info returned by get_object_info | 15:24 |
| remix_tj | acoles: no | 15:24 |
| remix_tj | but if there is another function to get it, i can use | 15:24 |
| *** armaan has quit IRC | 15:25 | |
| *** armaan has joined #openstack-swift | 15:26 | |
| acoles | you could stash a copy of x-timestamp during each PUT in object sysmeta, under your own sysmeta key, then you'd be able to access it from the get_object_info, under sysmeta. or maybe there's a way to get to it - AFAICT it comes back in the HEAD response that populated object info, just isn't copied into the cached dict | 15:26 |
| remix_tj | ok, i'll check if i can get those infos | 15:31 |
| *** tesseract has joined #openstack-swift | 15:31 | |
| *** gyee has joined #openstack-swift | 16:02 | |
| notmyname | good monring | 16:11 |
| notmyname | FYI, now that sharding has merged, I'm planning on waiting until the end of this week (or potentially the start of next) to do the 2.18.0 tag | 16:12 |
| notmyname | DHE: ^ | 16:12 |
| DHE | :) | 16:13 |
| notmyname | note for everyone, though, sharding's merged, so that means master is unfrozen. other stuff can land now | 16:13 |
| *** cshastri has joined #openstack-swift | 16:13 | |
| notmyname | I'll be going to vancouver tomorrow and be there through thursday evening | 16:15 |
| notmyname | kota_: mattoliverau: cschwede: mahatic: if you're available, I'd love to go to dinner with you on wednesday evening | 16:15 |
| * notmyname is not sure who else is at the summit | 16:15 | |
| *** gyee has quit IRC | 16:31 | |
| *** gyee has joined #openstack-swift | 16:35 | |
| *** mikecmpbll has joined #openstack-swift | 16:51 | |
| *** silor has joined #openstack-swift | 16:53 | |
| *** psachin has quit IRC | 16:56 | |
| *** pcaruana has joined #openstack-swift | 17:05 | |
| *** germs has joined #openstack-swift | 17:13 | |
| *** germs has joined #openstack-swift | 17:13 | |
| *** germs has quit IRC | 17:17 | |
| *** armaan has quit IRC | 17:21 | |
| *** armaan has joined #openstack-swift | 17:21 | |
| *** spsurya has quit IRC | 17:27 | |
| *** germs has joined #openstack-swift | 17:27 | |
| *** tesseract has quit IRC | 17:28 | |
| timburke | acoles: so even with that patch, i'm still getting a failure around https://github.com/openstack/swift/blob/master/test/probe/test_sharder.py#L1817 | 17:35 |
| timburke | all replicas are at collapsed, with two deleted shard ranges plus their own shard range | 17:36 |
| acoles | timburke: can you take a peek at the partitions for root and shards? using shard-info? | 17:38 |
| acoles | timburke: if a shard is in same partition as root then it may be that it is shrunk sooner than the test expects | 17:38 |
| timburke | ah, that's probably it -- i see two dbs on 197, one on 98 | 17:38 |
| timburke | i guess we could add a --suffix option? *shrug* | 17:39 |
| acoles | if one on 197 is second shard then that would explain it | 17:39 |
| timburke | will re-run and check that it passes | 17:39 |
| openstackgerrit | Samuel Merritt proposed openstack/swift master: Add support for PROXY protocol v1 (only) https://review.openstack.org/373563 | 17:39 |
| acoles | or a --db_file option ? | 17:39 |
| acoles | timburke: or perhaps the test doesn't need to empty the second shard range quite so soon? | 17:40 |
| acoles | what's your EC ring part power? | 17:40 |
| timburke | both are 8 | 17:41 |
| acoles | hmm, maybe its the two shards that would need to be in same partition | 17:42 |
| timburke | root and shard index 1 had the same partition | 17:44 |
| acoles | ah, line 1814 will be the culprit then, running sharder on shard index 1 to get updates inadvertently runs sharder on root which causes shard index 1 to shrink (assuming it runs on the root first) | 17:45 |
| *** silor has quit IRC | 17:46 | |
| *** NM has quit IRC | 17:46 | |
| *** cshastri has quit IRC | 17:48 | |
| openstackgerrit | Alistair Coles proposed openstack/swift master: WIP Reclaim expired cleaving context metadata https://review.openstack.org/569832 | 17:49 |
| *** amandap has joined #openstack-swift | 17:49 | |
| acoles | timburke: I'll look tomorrow at if the object deletes can be reordered to avoid that glitch, if you don't beat me to it. I need to go now. | 17:50 |
| acoles | timburke: I'll look tomorrow at if the object deletes can be reordered to avoid that glitch, if you don't beat me to it. I need to go now. | 17:50 |
| *** NM has joined #openstack-swift | 17:53 | |
| *** geaaru has quit IRC | 17:54 | |
| *** amandap has quit IRC | 17:54 | |
| *** armaan has quit IRC | 18:12 | |
| *** armaan has joined #openstack-swift | 18:13 | |
| openstackgerrit | Tim Burke proposed openstack/swift master: Use assertRaises more https://review.openstack.org/567933 | 18:13 |
| torgomatic | interesting; test/unit/common/test_utils.py:TestUtils.test_get_logger_sysloghandler_plumbing fails on python 3.5 on Ubuntu 18.04, but passes on Ubuntu 16.04 | 18:15 |
| openstackgerrit | Tim Burke proposed openstack/swift master: Make the decision between primary/handoff sets more obvious https://review.openstack.org/567940 | 18:18 |
| openstackgerrit | Tim Burke proposed openstack/swift master: Fix-up signal handling https://review.openstack.org/569837 | 18:21 |
| openstackgerrit | Tim Burke proposed openstack/swift master: Handle SIGINT more like we do SIGHUP and SIGTERM https://review.openstack.org/569838 | 18:21 |
| openstackgerrit | Tim Burke proposed openstack/swift master: Remove exclude_states from get_shard_ranges https://review.openstack.org/569840 | 18:25 |
| *** SkyRocknRoll has quit IRC | 18:29 | |
| notmyname | timburke: or torgomatic: if either of you could take some time to look at https://review.openstack.org/#/c/555245/ in the next couple of days, I'd appreciate it | 18:29 |
| patchbot | patch 555245 - swift - Fix versioned writes error with url-encoded object... | 18:29 |
| notmyname | a while back (before the feature/deep crunch), we'd said we'd like to have it for 2.18.0 | 18:29 |
| notmyname | so if it were to land this week, that would be great | 18:29 |
| timburke | oh yeah, i'd started looking at it a while ago... i should at least leave my comments so far... | 18:30 |
| notmyname | thanks | 18:30 |
| notmyname | rememeber, kota_ is at the summit, so he likely won't have a lot of chance for gerrit back-and-forth | 18:31 |
| *** NM has quit IRC | 18:31 | |
| *** NM has joined #openstack-swift | 18:35 | |
| *** Guest74322 is now known as mgagne | 18:41 | |
| *** mgagne has quit IRC | 18:42 | |
| *** mgagne has joined #openstack-swift | 18:42 | |
| openstackgerrit | Tim Burke proposed openstack/swift master: Plumb includes down into _get_shard_range_rows https://review.openstack.org/569847 | 18:50 |
| *** mikecmpbll has quit IRC | 19:03 | |
| *** NM has quit IRC | 19:06 | |
| openstackgerrit | Tim Burke proposed openstack/swift master: container/backend: Always call transform_func when building listings https://review.openstack.org/569850 | 19:22 |
| *** zaitcev has joined #openstack-swift | 19:28 | |
| *** ChanServ sets mode: +v zaitcev | 19:28 | |
| *** amandap has joined #openstack-swift | 19:50 | |
| *** amandap has quit IRC | 19:54 | |
| *** NM has joined #openstack-swift | 19:57 | |
| clayg | timburke: ok, I see the race now in https://review.openstack.org/#/c/569837/1/swift/common/wsgi.py@949 | 20:05 |
| patchbot | patch 569837 - swift - Fix-up signal handling | 20:05 |
| clayg | THANK YOU! | 20:05 |
| clayg | But I'm not sure the solution is just to make it so small it's atomic - I think it'd be to design it so the context is "sigterm recieved = True" and we just never let sighup overwrite that | 20:06 |
| clayg | the problem is I didn't see the race the first go 'round so reusing context[1] = signal wasn't obviously wrong | 20:06 |
| clayg | THANKS for seeing what I could not! | 20:06 |
| timburke | i'm not sure it's a *bad* idea to have multiple SIGTERMs trigger multiple killpg()s though -- like, i *told you* i wanted you to go murdering processes, and i have have reason to believe that *i need to tell you again* -- just GO DO IT AGAIN | 20:08 |
| clayg | hrm... interesting... | 20:09 |
| clayg | timburke: it was under sam's advice that pulled out the syscalls from the handler - it sure made sense to me - maybe we could ask his opinion? | 20:09 |
| clayg | torgomatic: you have AF's for patch 569837 :D | 20:10 |
| patchbot | https://review.openstack.org/#/c/569837/ - swift - Fix-up signal handling | 20:10 |
| torgomatic | AF's? | 20:10 |
| clayg | oh.. I was thinking "any" - i guess it's typically just "as 4$@%" | 20:11 |
| timburke | i always trust torgomatic on these things -- i'll defer to his judgement! | 20:11 |
| clayg | torgomatic: you've been nominated ruler of all signal handling :D | 20:12 |
| torgomatic | oh joy :) | 20:13 |
| torgomatic | I left a comment with my reasoning on that patch | 20:14 |
| timburke | hmmm... so this is weird -- when i go to run *all* probe tests, they go along fine until test_object_partpower_increase, at which point i hit an error down in resetswift: "umount: /mnt/swift-disk: target is busy" | 20:16 |
| clayg | torgomatic: but we already knew that logging would bork us - timburke pointed out that syscalls probably wouldn't bork us - but we're less clear if moving them back into the signal handler is "better"? I guess it makes the race smaller. That's probably better. | 20:18 |
| torgomatic | I should probably go read up on signal handlers and such again. I have a vague idea that maybe you can't get a signal while you're in a signal handler, so if you manage to infinite-loop in a signal handler then SIGKILL is your only way out, but I could easily be wrong about that | 20:19 |
| openstackgerrit | Kota Tsuyuzaki proposed openstack/swift master: Maintain manage-shard-ranges docs in its module https://review.openstack.org/569799 | 20:21 |
| *** germs has quit IRC | 20:21 | |
| openstackgerrit | Merged openstack/swift master: Use correct policy when faking misplaced objects in probe test https://review.openstack.org/569737 | 20:23 |
| openstackgerrit | Merged openstack/swift master: Use assertRaises more https://review.openstack.org/567933 | 20:23 |
| kota_ | notmyname: unfortunately, wednesday is only unavailable day through the summit for dinner :/ | 20:24 |
| kota_ | notmyname: today, Tuesday, Thursday can work for me. | 20:25 |
| kota_ | and congrats container-sharding is now in the master. | 20:25 |
| openstackgerrit | Tim Burke proposed openstack/swift master: Pre-storage-policy-index tests shouldn't have shard_range tables https://review.openstack.org/569856 | 20:36 |
| *** pcaruana has quit IRC | 20:37 | |
| timburke | i'd really like to take those container db migration tests and make them much more declarative... | 20:37 |
| *** mvenesio has quit IRC | 20:40 | |
| openstackgerrit | Tim Burke proposed openstack/swift master: Respect auto_create_account_prefix in container PUT path https://review.openstack.org/569860 | 20:47 |
| notmyname | wow. 2.18.0 will be swift's 49th release! | 20:51 |
| clayg | whoot! let's do 3.0.0 next!? | 20:54 |
| clayg | big five-oh!? | 20:54 |
| openstackgerrit | Tim Burke proposed openstack/swift master: Use sets instead of dicts when we only care about the keys https://review.openstack.org/569861 | 21:01 |
| clayg | timburke: what's the story with patch 569860 - can you put it in the commit message for me? | 21:07 |
| patchbot | https://review.openstack.org/#/c/569860/ - swift - Respect auto_create_account_prefix in container PU... | 21:07 |
| clayg | it's not obvious to me at all why we'd want to start doing that? Do we HAVE to? | 21:08 |
| timburke | nothing much -- it just seemed weird to me that we respect that flag on GET (check up around L90) to differentiate between 401/403/404 (i guess?) but we don't look at it on PUT | 21:09 |
| timburke | like -- i can do a GET on the container and get back a 404... but then i do a PUT and i get back... another 404? i guess it's telling me what needs to happen at least... it's just weird that we've got an option that says that account should get auto created, but we're not auto creating it | 21:12 |
| clayg | I still don't get it... i thought all that dot-account stuff was for internal daemons only - you're saying some of the proxy's controllers make special accommodations for dot-accounts and this is the only operation that doesn't? | 21:14 |
| timburke | iirc we need the code in GET so we can get shards before the container-updater goes to report stats back to the .sharded_ account which pops it into existence | 21:18 |
| timburke | POST and DELETE don't need it -- if the account doesn't exist, the container can't exist, so 404 is appropriate | 21:19 |
| timburke | i suppose there's a similar oddity in the account APIs... there, we don't look at the prefixes *at all* | 21:20 |
| clayg | i see, so we snuck it in so internal client can get ahead of the container-updater... and I guess maybe we can't just skip the check on GET because of how account-reaping works... difference between 404 and 410 maybe? | 21:24 |
| clayg | I mean what you're saying about POST & DELETE can't be true or we wouldn't need the hack in GET ;) | 21:24 |
| clayg | if we can't loose the check on GET maybe internal client should just get special privledge to make whatever requests to the container layer it damn well pleases regardless of what the account layer thinks? | 21:25 |
| timburke | bah, you're right! need it in POST and DELETE, too. | 21:28 |
| timburke | this isn't even for internal client! it's for client-facing proxy-server! as soon as a container starts responding that it's sharding and got an active range, we want to be able to go querying that shard db | 21:29 |
| clayg | oh, gross - does container_info have this problem? maybe the flag is "hi, I'm a client, please don't trust me" vs. "I am the $%^&ing proxy dude, back off!" | 21:33 |
| clayg | or maybe it's the combination of 404 and config_option_1 + config_option_2 ??? | 21:33 |
| *** NM has quit IRC | 21:33 | |
| *** NM has joined #openstack-swift | 21:34 | |
| clayg | you could factor it out to a single source of truth... IDK, i wonder if it's more "this makes my OCD itch" more than "this is technical investment to make doing other things later cheaper" | 21:35 |
| timburke | clayg: probably more the former than the latter -- just something that i noticed in review that i figured i could push up a patch for... i'll see about getting it better formed (or just let it get forgotten in my long long list of open patches) | 21:38 |
| clayg | 👍 | 21:38 |
| *** NM has quit IRC | 21:38 | |
| *** Guest83262 has joined #openstack-swift | 21:46 | |
| *** germs has joined #openstack-swift | 21:48 | |
| *** Guest83262 is now known as _alastor_ | 21:49 | |
| *** germs has quit IRC | 21:50 | |
| *** ianychoi has quit IRC | 22:06 | |
| *** ianychoi has joined #openstack-swift | 22:06 | |
| *** germs has joined #openstack-swift | 22:13 | |
| *** germs has quit IRC | 22:13 | |
| *** germs has joined #openstack-swift | 22:13 | |
| *** rcernin has joined #openstack-swift | 22:16 | |
| *** germs has quit IRC | 22:17 | |
| openstackgerrit | Merged openstack/swift master: Container sharding doc fixes and clarifications https://review.openstack.org/569773 | 22:29 |
| openstackgerrit | Merged openstack/swift master: Pre-storage-policy-index tests shouldn't have shard_range tables https://review.openstack.org/569856 | 22:29 |
| *** mikecmpbll has joined #openstack-swift | 22:34 | |
| timburke | oh yay! someone already wrote up https://bugs.launchpad.net/openstack-gate/+bug/1708704 | 22:46 |
| openstack | Launchpad bug 1708704 in OpenStack-Gate "yum client: [Errno 256] No more mirrors to try" [Undecided,New] | 22:46 |
| timburke | looks like manila's already found a workaround!? sweet! https://review.openstack.org/#/c/569360/ | 22:47 |
| patchbot | patch 569360 - manila - Fix manila-tempest-*-centos-7 jobs (MERGED) | 22:47 |
| openstackgerrit | Tim Burke proposed openstack/swift master: Fix probe tests in the gate https://review.openstack.org/569882 | 22:52 |
| clayg | timburke: you're a mad man! 569882 | 22:53 |
| timburke | clayg: the commit message may be more hope than fact ;-) | 22:53 |
| clayg | if the gate passes can we get the abridged story in the commit message? I'll pour a glass of wine and snuggle up with my phone .. | 22:54 |
| timburke | ...it's not super-interesting. the manila folks did all the heavy lifting and spotted the centos-devel message: https://lists.centos.org/pipermail/centos-devel/2018-March/016542.html | 22:57 |
| *** amandap has joined #openstack-swift | 22:58 | |
| timburke | basically, we need /etc/yum/vars/contentdir to say "centos" and it either wasn't getting set, or was saying "altarch", or *something* | 22:58 |
| timburke | maybe related? https://bugs.centos.org/view.php?id=14795 | 22:59 |
| timburke | already doing loads better: http://zuul.openstack.org/stream.html?uuid=e42544d0e627464faae31646eabd4563&logfile=console.log | 22:59 |
| *** amandap has quit IRC | 23:02 | |
| *** threestrands has joined #openstack-swift | 23:05 | |
| *** amandap has joined #openstack-swift | 23:08 | |
| timburke | clayg: how much do you really care about the story? or is the link to someone else's write-up (which in turn includes links to how kolla and tripleo fixed it) enough? | 23:11 |
| timburke | 'cause those probe tests look like they're gonna pass, and i'd rather not waste wall time if i'm gonna need to push up an updated commit message | 23:12 |
| notmyname | timburke: great work. IMO a link to someone else's writeup is find | 23:13 |
| notmyname | *fine | 23:13 |
| notmyname | ah, the related-change id. got it | 23:14 |
| notmyname | timburke: I'm personally good with it, as-is | 23:14 |
| clayg | probably i'm just not good at reading playbooks! | 23:15 |
| clayg | ... and I think I missed the related change - cause I hadn't looked at it before | 23:15 |
| clayg | i wonder why whoever is producing these images wouldn't want contentdir to just be populated to start with? | 23:20 |
| timburke | no idea. something tells me eventually this would have resolved itself, and the base images would start including it... but i want working probe tests in the gate faster than "eventually" | 23:28 |
| *** armaan has quit IRC | 23:28 | |
| clayg | 👍 | 23:28 |
| clayg | when do we get to find out if it works!!! | 23:29 |
| *** armaan has joined #openstack-swift | 23:29 | |
| timburke | clayg: we already know it does :-) http://zuul.openstack.org/ shows that change just waiting on the two long-running jobs -- everything else (including probe tests) passed | 23:36 |
| clayg | :shipit: | 23:37 |
| openstackgerrit | Tim Burke proposed openstack/swift master: Clean up direct_client's gen_headers https://review.openstack.org/569884 | 23:40 |
| openstackgerrit | Merged openstack/swift master: Maintain manage-shard-ranges docs in its module https://review.openstack.org/569799 | 23:41 |
| clayg | +19, -78 <- LIKE A BOSS! | 23:45 |
| *** amandap has quit IRC | 23:49 | |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!