| zaitcev | So, we can settle on representing both keys and values as native strings, it seems. | 00:00 |
|---|---|---|
| *** rcernin has quit IRC | 00:07 | |
| timburke | but there *isn't* a native-string representation of b'\xff' under py3 :-( i think that commit message is saying we standardized on *bytes*, and if we happen to unpickle some unicode, we'll encode it to bytes | 00:07 |
| timburke | or, we assume that people probably aren't stuffing non-utf8 metadata on objects, call the ability to do so a bug, and fix the bug | 00:08 |
| *** rcernin has joined #openstack-swift | 00:09 | |
| timburke | looks like acoles first noticed this around http://eavesdrop.openstack.org/irclogs/%23openstack-swift/%23openstack-swift.2016-02-26.log.html#t2016-02-26T17:14:25 -- we were looking at it in the context of https://review.openstack.org/#/c/284822/ (which was a follow-up for https://review.openstack.org/#/c/282363/ ) | 00:16 |
| patchbot | patch 284822 - python-swiftclient - Fix test for redacting sensitive data in client.ht... (MERGED) - 3 patch sets | 00:16 |
| patchbot | patch 282363 - python-swiftclient - Do not reveal auth token in swiftclient log messag... (MERGED) - 6 patch sets | 00:16 |
| zaitcev | it's fine, it's fine. I only need to make a decision on representation, and someone has done it for me already | 00:19 |
| zaitcev | https://github.com/openstack/swift/blob/2.20.0/swift/obj/diskfile.py#L156 | 00:19 |
| *** itlinux has joined #openstack-swift | 01:03 | |
| *** itlinux has quit IRC | 01:19 | |
| *** baojg has quit IRC | 01:24 | |
| *** itlinux has joined #openstack-swift | 02:11 | |
| *** itlinux has quit IRC | 02:13 | |
| *** psachin has joined #openstack-swift | 02:48 | |
| openstackgerrit | Merged openstack/swift master: Modify the description content https://review.openstack.org/623650 | 03:02 |
| *** gyee has quit IRC | 03:10 | |
| kota_ | hello | 03:21 |
| *** itlinux has joined #openstack-swift | 03:33 | |
| *** spsurya has joined #openstack-swift | 03:46 | |
| mattoliverau | kota_: o/ | 03:50 |
| kota_ | mattoliverau: o/ | 03:50 |
| zaitcev | AttributeError: 'tuple_iterator' object has no attribute 'next' | 03:56 |
| zaitcev | ato chotto | 03:56 |
| *** lifeless_ has joined #openstack-swift | 04:57 | |
| *** psachin has quit IRC | 05:02 | |
| *** joeljwright has quit IRC | 05:02 | |
| *** lifeless has quit IRC | 05:02 | |
| *** joeljwright has joined #openstack-swift | 05:03 | |
| *** ChanServ sets mode: +v joeljwright | 05:03 | |
| *** e0ne has joined #openstack-swift | 06:22 | |
| *** e0ne has quit IRC | 06:23 | |
| *** mikecmpbll has quit IRC | 06:36 | |
| *** mikecmpbll has joined #openstack-swift | 06:39 | |
| *** pcaruana has joined #openstack-swift | 07:06 | |
| *** e0ne has joined #openstack-swift | 07:18 | |
| *** e0ne has quit IRC | 07:22 | |
| *** rcernin has quit IRC | 07:26 | |
| *** e0ne has joined #openstack-swift | 07:47 | |
| *** itlinux has quit IRC | 07:48 | |
| *** mikecmpbll has quit IRC | 07:49 | |
| *** hseipp has joined #openstack-swift | 07:52 | |
| *** e0ne has quit IRC | 08:47 | |
| *** mikecmpbll has joined #openstack-swift | 08:52 | |
| *** mikecmpbll has quit IRC | 08:54 | |
| *** mikecmpbll has joined #openstack-swift | 09:13 | |
| *** ccamacho has joined #openstack-swift | 09:16 | |
| *** pavelkv has joined #openstack-swift | 09:59 | |
| pavelkv | Thanks @timburke for slow object expirator patch; we need it after Pike upgrade to resolve 450M objects today :-) because we PUT objects much faster than expirer deletes them | 10:03 |
| pavelkv | can somebody ack me cherry-pick to Pike #629402 ? | 10:04 |
| pavelkv | https://review.openstack.org/#/c/629402/ | 10:05 |
| patchbot | patch 629402 - swift (stable/pike) - object-server can 409 in response to x-if-delete-at - 1 patch set | 10:05 |
| *** e0ne has joined #openstack-swift | 10:21 | |
| *** e0ne has quit IRC | 12:30 | |
| *** ybunker has joined #openstack-swift | 12:52 | |
| *** e0ne has joined #openstack-swift | 13:24 | |
| *** ybunker has quit IRC | 15:56 | |
| *** ybunker has joined #openstack-swift | 15:57 | |
| *** ybunker has quit IRC | 16:01 | |
| *** ybunker has joined #openstack-swift | 16:02 | |
| *** pcaruana has quit IRC | 16:03 | |
| openstackgerrit | zhulingjie proposed openstack/swift master: Update hacking version https://review.openstack.org/628515 | 16:05 |
| *** gyee has joined #openstack-swift | 16:18 | |
| *** e0ne has quit IRC | 17:03 | |
| *** itlinux has joined #openstack-swift | 17:05 | |
| openstackgerrit | 98k proposed openstack/python-swiftclient master: Add doc/requirements.txt to releasenotes tox environment https://review.openstack.org/629578 | 17:12 |
| *** hseipp has quit IRC | 17:25 | |
| tdasilva | timburke: just trying to understand a bit more about s3api code, the reason for verifying the etag and size on a MPU upload is so that it can return a "s3" error message as opposed to an "slo" error message, correct. Ultimately slo does the verification for of the manifest anyway...this is re. p 610747 | 17:27 |
| patchbot | https://review.openstack.org/#/c/610747/ - swift - s3api: Stop relying on container listings during m... - 9 patch sets | 17:27 |
| tdasilva | s/correct./correct? | 17:27 |
| *** mikecmpbll has quit IRC | 17:30 | |
| *** mikecmpbll has joined #openstack-swift | 17:43 | |
| openstackgerrit | Merged openstack/swift master: Add slo_manifest_hook callback https://review.openstack.org/609860 | 17:56 |
| openstackgerrit | Pete Zaitcev proposed openstack/swift master: WIP py3: object server https://review.openstack.org/629610 | 18:03 |
| *** zaitcev has quit IRC | 18:05 | |
| timburke | zaitcev: oh jeeze... we do .next() all over the place in tests, don't we... | 18:25 |
| timburke | tdasilva: yeah, more or less. there's some historical context going on, too, though. it used to be that SLO required that you specify all of path, etag, and size_bytes in the manifest. swift3 could generate the path from the part number and upload id, and etag is provided by the client, but it still needed size_bytes or SLO would reject the manifest | 18:32 |
| timburke | following https://review.openstack.org/#/c/184479/ etag and size_bytes became optional (and following https://review.openstack.org/#/c/391090/ you could omit them completely), but swift3 still had to work with old swift, so it did a container listing to get all the sizes | 18:32 |
| patchbot | patch 184479 - swift - Allow SLO PUTs to forgo per-segment integrity checks (MERGED) - 6 patch sets | 18:32 |
| patchbot | patch 391090 - swift - SLO: Make etag and size_bytes fully optional (MERGED) - 6 patch sets | 18:32 |
| timburke | as long as it had the listing in hand, it could check etag, too, and avoid all the HEADs | 18:32 |
| *** e0ne has joined #openstack-swift | 18:35 | |
| *** itlinux has quit IRC | 18:36 | |
| *** itlinux has joined #openstack-swift | 18:38 | |
| *** e0ne has quit IRC | 19:02 | |
| *** e0ne has joined #openstack-swift | 19:05 | |
| *** ybunker has quit IRC | 19:43 | |
| *** e0ne has quit IRC | 20:08 | |
| cwright | Hi everyone, I've had a situation where account quotas stopped getting updated. Container counts continued to be updated quickly. | 20:35 |
| cwright | I restarted all services, and the quotas began updating again. All processes were still running before the restart. | 20:35 |
| cwright | From what I've read, the container-updater is responsible for updating account bytes used. Is there any explanation for it to have stopped updating? | 20:35 |
| cwright | by "quotas began updating again", I mean "bytes-used" started getting updated again | 20:36 |
| openstackgerrit | Tim Burke proposed openstack/swift master: slo_manifest_hook follow-up https://review.openstack.org/629659 | 20:40 |
| timburke | cwright: sounds like https://bugs.launchpad.net/swift/+bug/1722951 | 20:45 |
| openstack | Launchpad bug 1722951 in swift (Ubuntu) "Container updater may be stuck and not make progress" [Undecided,New] | 20:45 |
| timburke | should be fixed with https://github.com/openstack/swift/commit/dc8da5b | 20:47 |
| mattoliverau | morning | 20:56 |
| kota_ | morning | 20:57 |
| notmyname | hello mattoliverau and kota_ | 20:58 |
| notmyname | mattoliverau: kota_: timburke: rledisez: clayg: tdasilva: zaitcev: swift team meeting in 2 minutes in #openstack-meeting | 20:59 |
| notmyname | of course, anyone else is welcome too :-) | 21:00 |
| *** m_kazuhiro has joined #openstack-swift | 21:01 | |
| cwright | timburke: thanks for the link to that bug, it does sound like what I'm experiencing | 21:23 |
| timburke | good news is that it's fixed in newer swift, and even backported to affected stable branches. bad news is that we haven't had new tags on the stable branches in a while, but i think notmyname is planning on fixing that soon :-) | 21:25 |
| notmyname | ...soon... | 21:26 |
| openstackgerrit | Tim Burke proposed openstack/swift master: WIP py3: object server https://review.openstack.org/629610 | 21:27 |
| tdasilva | mattoliverau: I think we even documented that behavior: https://github.com/openstack/swift/blob/master/swift/common/middleware/versioned_writes.py#L107,L111 | 21:48 |
| mattoliverau | Yeah, that's why I added it was known in the bugm | 21:50 |
| mattoliverau | But figured it was still a problem | 21:50 |
| mattoliverau | And now that we have a warning people want to remove it | 21:50 |
| clayg | rledisez: the main issue I have with ssync is that it tries to act like it's http-ish - which is broken for streaming requests when an expect-100 continue returns an error | 21:55 |
| clayg | PUT content-length: X (w/ expect: 100-contineu); get_expect -> 409 => HTTP IS BROKEN | 21:56 |
| clayg | because the server HAS to get those X bytes before it can handle another request | 21:56 |
| rledisez | clayg: yeah, ok, you're right, i'm not fixing that for now :D | 21:57 |
| clayg | ssync is weird in that it has some "failure count" non-sense in there, because we don't get many failures in the final status *after* we send the bytes | 21:57 |
| clayg | even the whole "here's all the objects i have tell me what you want" is so stupid! if we could do errors the conversation would be "you want this one? nope. This one? sure. This one? nah. ..." | 21:57 |
| clayg | i don't much care for ssync | 21:58 |
| clayg | let's just get get in gRPC and then rewrite the protocol | 21:58 |
| *** m_kazuhiro has quit IRC | 21:59 | |
| rledisez | i'm working on ssync because it's here, and it (finally) works, it's not perfect, but it can be improved a lot with few simple patches. gRPC is a long term goal IMHO, SSYNC is the quick win. I take every win I can :) | 22:00 |
| clayg | 👍 | 22:05 |
| *** ccamacho has quit IRC | 22:08 | |
| *** ccamacho has joined #openstack-swift | 22:27 | |
| *** ccamacho has quit IRC | 22:28 | |
| *** rcernin has joined #openstack-swift | 22:52 | |
| *** pavelkv has quit IRC | 22:59 | |
| openstackgerrit | Tim Burke proposed openstack/swift master: slo_manifest_hook follow-up https://review.openstack.org/629659 | 23:03 |
| *** itlinux has quit IRC | 23:06 | |
| openstackgerrit | Clay Gerrard proposed openstack/swift master: wip: rebuild frags for unmounted disks https://review.openstack.org/629056 | 23:12 |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!