| *** rcernin has quit IRC | 02:48 | |
| *** hoonetorg has quit IRC | 02:51 | |
| *** rcernin has joined #openstack-swift | 02:59 | |
| *** rcernin has quit IRC | 03:05 | |
| *** rcernin has joined #openstack-swift | 03:20 | |
| *** rcernin has quit IRC | 03:22 | |
| *** rcernin has joined #openstack-swift | 03:22 | |
| *** manuvakery has joined #openstack-swift | 03:27 | |
| manuvakery | timburke: as we discussed I proposed the basic tagging support here https://review.opendev.org/735173 | 03:29 |
|---|---|---|
| patchbot | patch 735173 - swift - s3api: Add basic support for ?tagging requests - 3 patch sets | 03:29 |
| timburke | manuvakery, yes, i saw! thank you! i've been meaning to give it a spin but got side-tracked by gate issues | 03:30 |
| manuvakery | Ok thanks | 03:30 |
| *** ravsingh has joined #openstack-swift | 03:30 | |
| *** psachin has joined #openstack-swift | 03:37 | |
| openstackgerrit | Tim Burke proposed openstack/python-swiftclient master: Clean up some warnings https://review.opendev.org/736426 | 04:05 |
| *** evrardjp has quit IRC | 04:33 | |
| *** evrardjp has joined #openstack-swift | 04:33 | |
| *** hoonetorg has joined #openstack-swift | 04:35 | |
| *** m75abrams has joined #openstack-swift | 06:24 | |
| *** ravsingh has left #openstack-swift | 06:28 | |
| *** noonedeadpunk has joined #openstack-swift | 06:41 | |
| *** rpittau|afk is now known as rpittau | 06:43 | |
| *** rcernin has quit IRC | 07:54 | |
| *** rcernin_ has joined #openstack-swift | 07:54 | |
| *** rcernin_ has quit IRC | 08:20 | |
| *** manuvakery has quit IRC | 09:00 | |
| *** ccamacho has quit IRC | 09:16 | |
| *** tkajinam has quit IRC | 10:05 | |
| *** rpittau is now known as rpittau|bbl | 10:06 | |
| *** manuvakery has joined #openstack-swift | 10:34 | |
| *** rcernin_ has joined #openstack-swift | 10:48 | |
| *** ccamacho has joined #openstack-swift | 11:22 | |
| *** m75abrams has quit IRC | 12:17 | |
| *** rpittau|bbl is now known as rpittau | 12:20 | |
| *** noonedeadpunk is now known as noonedeadpunk_ | 12:57 | |
| *** psachin has quit IRC | 13:20 | |
| *** m75abrams has joined #openstack-swift | 14:05 | |
| *** rcernin_ has quit IRC | 14:14 | |
| *** noonedeadpunk_ is now known as noonedeadpunk | 14:29 | |
| *** m75abrams has quit IRC | 14:55 | |
| *** jv has quit IRC | 15:16 | |
| *** jv has joined #openstack-swift | 15:31 | |
| clayg | timburke: didn't you have a change related to p 728298 that tried used the most recent backend_path instead of the first one? | 15:43 |
| patchbot | https://review.opendev.org/#/c/728298/ - swift - proxy-logging: Use swift.backend_path if available - 3 patch sets | 15:43 |
| *** gyee has joined #openstack-swift | 15:50 | |
| timburke | good morning | 16:04 |
| timburke | clayg, yeah -- https://review.opendev.org/#/c/735221/ | 16:04 |
| patchbot | patch 735221 - swift - s3api: Set swift.backend_path to the last-used pat... - 1 patch set | 16:04 |
| timburke | i should get rledisez / ovh's take on it | 16:05 |
| timburke | idk where people usually have ceilometer middleware in the pipeline -- i've never used it | 16:05 |
| *** manuvakery has quit IRC | 16:33 | |
| openstackgerrit | Tim Burke proposed openstack/swift feature/losf: Merge remote-tracking branch 'gerrit/master' into feature/losf https://review.opendev.org/735381 | 16:42 |
| *** rpittau is now known as rpittau|afk | 17:00 | |
| openstackgerrit | Tim Burke proposed openstack/swift master: Rip out pickle support in our memcached client https://review.opendev.org/736787 | 17:32 |
| clayg | timburke: where do you come up with these ideas? | 17:33 |
| timburke | i get tired of having useless knobs :P | 17:33 |
| timburke | speaking of, how's the retry-abort thing looking today? | 17:33 |
| timburke | er, retry-complete, i mean | 17:34 |
| timburke | 8 years seems like a long-enough deprecation period, though ;-) | 17:35 |
| rledisez | timburke: we usually place ceilometermiddleware before any middleware than can generate subrequests, so it's on the left side of bulk and staticweb | 17:53 |
| clayg | timburke: still going! it'll be 24 hours very soon | 17:57 |
| *** manuvakery has joined #openstack-swift | 18:04 | |
| openstackgerrit | Tim Burke proposed openstack/swift master: wip: memcache: Config option to add a chance of skipping memcache https://review.opendev.org/736802 | 18:48 |
| clayg | timburke: ok, my complete multi part is still working 24 hours after the fact | 18:55 |
| clayg | well shit, no my experiment was botched - so ignore everything I've said 😞 | 18:58 |
| clayg | ok, i have the loop making the right api calls now - | 19:19 |
| clayg | @timburke abort after complete returns 204 | 19:21 |
| clayg | timburke: and you can still keep sending complete | 19:22 |
| clayg | I'll try again to figure out how long 😞 | 19:22 |
| *** gmann is now known as gmann_afk | 19:22 | |
| timburke | and you can still read the object? seems kinda wild that you can get a successful abort that doesn't actually clean up any data... | 19:22 |
| clayg | FWIW doing a list parts after complete fails immediately with NoSuchUpload | 19:24 |
| clayg | i can still see it listed it in the bucket - i'll try to download | 19:24 |
| clayg | yeah download seems to work | 19:25 |
| clayg | was abort/complete/list-parts the main api calls we were wondering about after a successful complete? | 19:26 |
| timburke | yeah, pretty sure | 19:26 |
| clayg | ok, well I have at least one example of a complete failing after 24 hours - so that's something | 19:27 |
| clayg | I have another completed upload calling complete every 5 mins (for realzy this time) and we'll find out how long it lasts 🤞 | 19:27 |
| clayg | I would assume abort would behave similarly... maybe I can convince my loop to do both 🤔 | 19:28 |
| openstackgerrit | Tim Burke proposed openstack/swift stable/ussuri: Fix pep8 job https://review.opendev.org/736814 | 19:31 |
| *** gmann_afk is now known as gmann | 19:37 | |
| *** gmann is now known as gmann_afk | 20:10 | |
| openstackgerrit | Tim Burke proposed openstack/swift stable/ussuri: Fix stable gate https://review.opendev.org/736829 | 20:17 |
| *** manuvakery has quit IRC | 20:31 | |
| timburke | what do people usually call their networks? i know clayg, tdasilva, and i use "outward-facing", "cluster-facing", and "replication" but idk how standard that is... | 20:44 |
| timburke | also i'm thinking about https://review.opendev.org/#/c/735991/ and how it'd really be preferable to have reconciler on the replication network -- but still want to provide an option since that may not actually be accessible from where people currently have their reconcilers deployed | 20:46 |
| patchbot | patch 735991 - swift - Add X-Backend-Use-Replication-Network header - 1 patch set | 20:46 |
| timburke | having a `use_cluster_facing_network = True` option feels stupid and wrong... but use_client_traffic_network isn't great, either (though it seems demonstrably better than use_client_network!) | 20:47 |
| timburke | anybody want to help me pick a name? | 20:48 |
| timburke | use_primary_network, maybe? | 20:49 |
| rledisez | FWIW, I use public, data and replication (data is proxy to storage node) | 20:57 |
| openstackgerrit | Tim Burke proposed openstack/swift stable/train: Use ensure-pip role https://review.opendev.org/736845 | 21:42 |
| *** lxkong has joined #openstack-swift | 21:45 | |
| lxkong | hi swift team, is there some way for a cloud admin to check (or better, delete) the orphan containers for the tenants that are removed from keystone? | 21:46 |
| clayg | does swift client ever decide to use a *LO on it's on? w/o me setting -S ? | 21:48 |
| timburke | lxkong, a reseller admin should be able to delete the account and let the account-reaper clean up all the old data -- see https://docs.openstack.org/swift/latest/overview_reaper.html | 21:48 |
| timburke | clayg, i think maybe when uploading from stdin? don't think so when you're uploading a file, though | 21:49 |
| clayg | yeah i'm upload from stdin! is there anyway to disable that? "trust me it's less than 5GB" | 21:50 |
| clayg | timburke: so this is a reconciler option? do you want the default to be existing behavior or just code the option as a safety latch JIC and immediately deprecate (pls configure your reconciler to have access to replication network) | 21:52 |
| clayg | and the sharder we're only not worried about because it's already on the container nodes next to the replicator ... which always use the replication network | 21:52 |
| *** rcernin_ has joined #openstack-swift | 21:55 | |
| openstackgerrit | Clay Gerrard proposed openstack/swift master: Add concurrent primary feeder to EC GET requests https://review.opendev.org/711342 | 21:57 |
| timburke | clayg, nah, don't think timur included a clean way to do that in https://opendev.org/openstack/python-swiftclient/commit/2faea932 -- you *could* set the segment-size arbitrarily high, but it'd mean buffering the whole thing in memory | 22:00 |
| timburke | yeah, i was thinking that reconciler should default to using the replication network, but i ought to include a way to get back to the old behavior since there's no guarantee that the reconciler's running on a system with access to that net. idk about deprecating... hadn't thought that far ahead. could there be a legit reason to put them on the same net? 🤔 | 22:02 |
| timburke | as for the sharder, it *already* uses the replication network for a bunch of stuff -- the fact that it *also* uses the primary network sometimes is a pretty straight-forward bug with no real upgrade consequences | 22:03 |
| timburke | like, it doesn't just live *next to* the replicator, it *is* a replicator (just a very special-case one) | 22:04 |
| *** gmann_afk is now known as gmann | 22:04 | |
| *** tonyb has joined #openstack-swift | 22:05 | |
| clayg | Ok, so name it use_legacy_node_ip | 22:11 |
| clayg | _and_port if you’re feeling verbose | 22:12 |
| *** rcernin_ has quit IRC | 22:13 | |
| *** tkajinam has joined #openstack-swift | 23:00 | |
| *** rcernin has joined #openstack-swift | 23:16 | |
| timburke | account-reaper gets a little hairier -- like, i *expect* that it'd have replication-network access (since it's walking account-server drives), but it hasn't been explicitly *required* before... | 23:27 |
| timburke | ditto container-sync | 23:29 |
| openstackgerrit | Tim Burke proposed openstack/swift master: Allow direct and internal clients to use the replication network https://review.opendev.org/735751 | 23:52 |
Generated by irclog2html.py 2.17.2 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!