| *** thurloat has quit IRC | 00:02 | |
| *** NM has quit IRC | 01:00 | |
| *** gyee has quit IRC | 01:03 | |
| *** tokuzfunpi has joined #openstack-swift | 01:10 | |
| *** openstackgerrit has quit IRC | 01:30 | |
| *** psachin has joined #openstack-swift | 03:01 | |
| *** openstackstatus has quit IRC | 04:35 | |
| *** openstackstatus has joined #openstack-swift | 04:36 | |
| *** ChanServ sets mode: +v openstackstatus | 04:36 | |
| *** pcaruana has joined #openstack-swift | 05:06 | |
| *** tokuzfunpi has quit IRC | 05:27 | |
| *** tokuzfunpi has joined #openstack-swift | 05:52 | |
| *** tokuzfunpi has quit IRC | 05:56 | |
| *** tokuzfunpi has joined #openstack-swift | 05:57 | |
| *** e0ne has joined #openstack-swift | 06:28 | |
| *** e0ne has quit IRC | 06:48 | |
| *** e0ne has joined #openstack-swift | 06:57 | |
| *** openstackgerrit has joined #openstack-swift | 07:25 | |
| openstackgerrit | Matthew Oliver proposed openstack/swift master: py3: port staticweb middleware https://review.openstack.org/643855 | 07:25 |
|---|---|---|
| *** psachin has quit IRC | 07:32 | |
| *** e0ne has quit IRC | 07:33 | |
| *** gkadam has joined #openstack-swift | 07:34 | |
| *** pcaruana has quit IRC | 07:34 | |
| *** gkadam has quit IRC | 07:34 | |
| *** pcaruana has joined #openstack-swift | 07:35 | |
| *** psachin has joined #openstack-swift | 07:36 | |
| *** e0ne has joined #openstack-swift | 07:40 | |
| *** ccamacho has joined #openstack-swift | 07:45 | |
| *** e0ne has quit IRC | 07:48 | |
| *** e0ne has joined #openstack-swift | 07:53 | |
| *** tkajinam has quit IRC | 08:04 | |
| *** e0ne has quit IRC | 08:47 | |
| *** e0ne has joined #openstack-swift | 08:58 | |
| *** e0ne has quit IRC | 09:06 | |
| *** rcernin has quit IRC | 09:47 | |
| *** tokuzfunpi has quit IRC | 10:16 | |
| *** e0ne has joined #openstack-swift | 11:12 | |
| *** e0ne has quit IRC | 11:15 | |
| *** e0ne has joined #openstack-swift | 11:19 | |
| *** e0ne has quit IRC | 11:20 | |
| *** e0ne has joined #openstack-swift | 11:24 | |
| *** ybunker has joined #openstack-swift | 11:55 | |
| *** e0ne has quit IRC | 12:09 | |
| *** e0ne has joined #openstack-swift | 12:12 | |
| *** e0ne has quit IRC | 12:13 | |
| *** e0ne has joined #openstack-swift | 12:28 | |
| openstackgerrit | Alexandre Lécuyer proposed openstack/swift feature/losf: Change the Makefile to work on vagrant swift all in one https://review.openstack.org/651520 | 12:45 |
| *** NM has joined #openstack-swift | 12:47 | |
| alecuyer | kota_: ^^ made some changes to the Makefile, I will look again tomorrow (the -rpath flag is not working and the resulting binary requires LD_LIBRARY_PATH) | 12:47 |
| *** irclogbot_3 has joined #openstack-swift | 13:04 | |
| *** altlogbot_0 has joined #openstack-swift | 13:07 | |
| *** mvkr has quit IRC | 13:19 | |
| *** e0ne has quit IRC | 13:20 | |
| *** e0ne has joined #openstack-swift | 13:25 | |
| *** mvkr has joined #openstack-swift | 13:53 | |
| *** e0ne has quit IRC | 13:54 | |
| *** e0ne has joined #openstack-swift | 13:57 | |
| openstackgerrit | Clay Gerrard proposed openstack/swift master: Add tests for unexpected names in part dir https://review.openstack.org/651359 | 14:13 |
| *** openstackgerrit has quit IRC | 14:14 | |
| *** openstackgerrit has joined #openstack-swift | 14:25 | |
| openstackgerrit | Thiago da Silva proposed openstack/liberasurecode master: Install QuadIron for unit tests https://review.openstack.org/650483 | 14:25 |
| *** tokuzfunpi has joined #openstack-swift | 14:30 | |
| *** e0ne has quit IRC | 14:50 | |
| *** e0ne has joined #openstack-swift | 14:54 | |
| *** ianychoi has quit IRC | 14:56 | |
| openstackgerrit | Thiago da Silva proposed openstack/liberasurecode master: Install QuadIron for unit tests https://review.openstack.org/650483 | 15:04 |
| *** gyee has joined #openstack-swift | 15:26 | |
| *** ccamacho has quit IRC | 15:28 | |
| *** tokuzfunpi has quit IRC | 15:47 | |
| *** tokuzfunpi has joined #openstack-swift | 15:47 | |
| openstackgerrit | Thiago da Silva proposed openstack/liberasurecode master: Install QuadIron for unit tests https://review.openstack.org/650483 | 15:54 |
| *** psachin has quit IRC | 16:14 | |
| timburke | tdasilva, huh. looks like we need to install gcc-c++ on centos, but i guess it's already there on ubuntu: http://logs.openstack.org/83/650483/8/check/liberasurecode-unittests-centos-7/ff36c4c/job-output.txt.gz#_2019-04-10_16_09_07_882105 | 16:17 |
| tdasilva | yeah, was about to test on my centos vm before making change | 16:17 |
| tdasilva | timburke: once I installed gcc-c++, cmake ran fine, but then the next step fails with: http://paste.openstack.org/show/749129/ | 16:22 |
| tdasilva | timburke: my guess is because quadiron is using c++14 | 16:25 |
| timburke | might be. looks like the fedora box i'd gotten it built and installed on has libstdc++/gcc-c++ 8.3.1, while the cent7 box i'm just now trying to make this work on is back on 4.8.5... | 16:35 |
| tdasilva | yep | 16:38 |
| *** joeljwright has quit IRC | 16:40 | |
| *** joeljwright has joined #openstack-swift | 16:40 | |
| *** ChanServ sets mode: +v joeljwright | 16:40 | |
| timburke | so... i guess we just test it on ubuntu? not super-happy about it... but better than not testing it at all | 16:42 |
| *** altlogbot_0 has quit IRC | 16:44 | |
| *** joeljwright has quit IRC | 16:47 | |
| *** e0ne has quit IRC | 16:55 | |
| *** mvkr has quit IRC | 17:09 | |
| *** tokuzfunpi has quit IRC | 17:32 | |
| *** tokuzfunpi has joined #openstack-swift | 17:34 | |
| *** tokuzfunpi has quit IRC | 17:38 | |
| *** joeljwright has joined #openstack-swift | 17:42 | |
| *** ChanServ sets mode: +v joeljwright | 17:42 | |
| rexor | Hello.asdasd | 17:57 |
| rexor | 1231asszzd | 17:57 |
| *** ChanServ sets mode: +o notmyname | 17:58 | |
| rexor | Oh sorry :) had lag on my internet. | 17:58 |
| notmyname | rexor: I hope that isn't your password :-) | 17:59 |
| rexor | notmyname: no :) | 17:59 |
| *** notmyname sets mode: -o notmyname | 17:59 | |
| rexor | notmyname: I prefer easier passwords :D | 17:59 |
| notmyname | hunter12? | 18:00 |
| rexor | I have question regarding to rebalance. I wan | 18:00 |
| notmyname | oh, it's hunter2, not hunter12 (http://bash.org/?244321=) | 18:00 |
| notmyname | rexor: what's your question? | 18:00 |
| rexor | I wanto add disk to my cluster, starting with low weight. When i can know that i can start another ring rebalance process? | 18:01 |
| rexor | Without losing data. | 18:01 |
| notmyname | when all the servers with the new ring have finished a replication cycle (or reconstructor cycle if it's EC) | 18:01 |
| notmyname | but you won't lose data, in a durability sense. if you do it more often, you may lose access to data, but it will eventually come back | 18:02 |
| notmyname | so let's say you have a few racks of gear and you roll in another rack and add it to the ring. the ring rebalance will only move one primary location per partition for each rebalance cycle, so you will (1) add the capacity (2) rebalance (3) push out new rings (4) wait for replication to finish (5) repeat to step 2 | 18:03 |
| notmyname | keep going until the ring rebuilder reports a good balance for the ring | 18:04 |
| rexor | Ok. Still only way to check replication state is check logs? | 18:04 |
| notmyname | if you add all the capacity at once, that's the basic plan. but it sounds like you're adding it gradually. that's good if you're adding a lot of capacity and you don't want to overwhelm the cluster with replication traffic (eg you're adding >20% capacity or something--that's not a hard threshold) | 18:05 |
| notmyname | yeah, check logs and statsd metrics for progress | 18:05 |
| rexor | I am adding like 25% of cluster capacity :) | 18:07 |
| rexor | 1,1PB cluster now | 18:07 |
| notmyname | nice! | 18:07 |
| rexor | Thank you very much for help and for devloping so great product :) | 18:09 |
| notmyname | rexor: it's always exciting to hear about people using it. is there anything you can share about how you're using swift? | 18:09 |
| rexor | I am woring in Oktawave, one of the biggest public cloud provider. We adopted to give our clients object storage. For now most of them using it as backup place, but more and more clients using it as web storage(video, jpgs, javascripts etc..) for their applications. | 18:14 |
| rexor | In peaks we have like ~700 req per sec on our swift gateways | 18:15 |
| notmyname | that's pretty cool! thanks for sharing :-) | 18:15 |
| mgagne | is there a script that can check consistency between object database and filesystem? for example: object was deleted on filesystem but it still marked as existing in database. | 18:16 |
| notmyname | mgagne: "database" meaning the container db? | 18:16 |
| mgagne | notmyname: I suppose =) | 18:16 |
| notmyname | ie its not on disk but is in the listing | 18:16 |
| mgagne | notmyname: yet, user is getting 404 on "existing" objects | 18:17 |
| rexor | notmyname: no problem :) | 18:17 |
| notmyname | mgagne: jus to check the important thing, is the object supposed to exist or has the user sent a delete for it? | 18:17 |
| mgagne | notmyname: the user sent a delete request and the object shouldn't exist anymore. | 18:17 |
| notmyname | mgagne: ok. that's the better situation. don't want some object to go missing without it having been deleted :-) | 18:18 |
| mgagne | notmyname: understandable | 18:18 |
| mgagne | so I'm more or less looking for a fsck for swift | 18:19 |
| mgagne | that will check database and report on objects that no longer exist on disk. | 18:19 |
| thyrst | mgagne: There is container updater running on your store. The listing should be eventually corrected | 18:19 |
| mgagne | thyrst: this service is running but inconsistencies have existed for months. So I don't think it's actually fixing stuff. | 18:21 |
| notmyname | mgagne: yeah. the listings will get updated "eventually", like thyrst said. from an ops perspective, check that the updater is running and check for async pendings in the cluster (eg the swift-recon tool can help with that) | 18:21 |
| notmyname | mgagne: but to your actual question... I'm checking. IIRC there's not anything like that in the repo | 18:22 |
| thyrst | Oh. Actually it's object updater. Container updater checks just account database. :) | 18:24 |
| timburke | notmyname, https://github.com/openstack/swift/blob/master/bin/swift-account-audit ? i don't know that i've ever used it though... | 18:25 |
| notmyname | yeah, that's the sort of thing I'm looking for! | 18:26 |
| mgagne | object updater is running too. there are no async tasks according to swit-recon. In fact, I'm getting: {u'async_pending': None} | 18:26 |
| timburke | has options to just check a single container, or even a single object | 18:26 |
| notmyname | mgagne: yeah, do what timburke says (that's just general good advice) | 18:26 |
| mgagne | will look into that script, thanks | 18:27 |
| mgagne | notmyname: this script looks to be what I was looking for. =) | 18:38 |
| notmyname | credit goes to timburke :-) | 18:41 |
| mgagne | timburke: thanks! | 18:43 |
| *** tokuzfunpi has joined #openstack-swift | 18:46 | |
| *** tokuzfunpi has quit IRC | 18:56 | |
| -openstackstatus- NOTICE: Restarting Gerrit on review.openstack.org to pick up new configuration for the replication plugin | 19:06 | |
| openstackgerrit | Thiago da Silva proposed openstack/liberasurecode master: Install QuadIron for unit tests https://review.openstack.org/650483 | 19:09 |
| tdasilva | timburke: ^^^ that worked in my env, hoping it works in the gate too | 19:09 |
| tdasilva | ah, i think i need to add that last task to .bashrc | 19:12 |
| timburke | thanks tdasilva! | 19:13 |
| openstackgerrit | Thiago da Silva proposed openstack/liberasurecode master: Install QuadIron for unit tests https://review.openstack.org/650483 | 19:14 |
| *** ybunker has quit IRC | 19:15 | |
| openstackgerrit | Thiago da Silva proposed openstack/liberasurecode master: Install QuadIron for unit tests https://review.openstack.org/650483 | 19:27 |
| tdasilva | gotta love the way we do gate job development | 19:27 |
| timburke | tdasilva, i can't help but feel like this is a large part of the reason most of our func tests are in-process -- we can all run tox easily enough, so it's way easier for us to hack together a gate job that's just "go run tox with this env" | 19:38 |
| openstackgerrit | Tim Burke proposed openstack/swift master: Remove our urlparse wrapper https://review.openstack.org/651004 | 19:39 |
| tdasilva | timburke: yeah, i'm glad we don't run into this issue with all of our jobs. still I think it's worthwhile in this particular case to test across both platforms... | 19:42 |
| clayg | timburke: since we let people create symlinks before they create the target object it's going to be difficult for versioned writes to create a hardlink to the symlink with an etag that won't 412 | 19:55 |
| timburke | :-/ hmm... | 19:56 |
| clayg | it's possible that hardlinks etags should just match the next hop? i.e. a hardlink to a symlink is always d41d8cd98f00b204e9800998ecf8427e | 19:56 |
| openstackgerrit | Thiago da Silva proposed openstack/liberasurecode master: Install QuadIron for unit tests https://review.openstack.org/650483 | 19:56 |
| timburke | maybe when versioning a symlink, we don't make if hard? idk... it's all a mess, isn't it... | 19:56 |
| clayg | but I'll leave the test failing for now and just keep thinking 👍 | 19:56 |
| clayg | s/keep thinking/wait for you to tell me what to do/ ;) | 19:57 |
| tdasilva | could we change that premise? no longer allow symlinks before target exists? | 20:29 |
| timburke | clayg, so looking at https://review.openstack.org/#/c/648263/4/swift/obj/expirer.py@425 -- what do you think about whether the bulk-delete task should be cleaning up the expiry queue? | 20:29 |
| patchbot | patch 648263 - swift - WIP: s3api: Make multi-deletes async - 4 patch sets | 20:30 |
| *** tokuzfunpi has joined #openstack-swift | 20:35 | |
| *** tokuzfunpi has quit IRC | 20:39 | |
| clayg | timburke: oh right, because we might be bulk deleting an object that was set to expire next month... sure! | 20:43 |
| *** mvkr has joined #openstack-swift | 20:52 | |
| kota_ | morning | 20:53 |
| timburke | kota_, o/ | 20:54 |
| kota_ | timburke: o/ | 20:55 |
| mattoliverau | Morning | 20:59 |
| openstackgerrit | Tim Burke proposed openstack/swift master: WIP: s3api: Make multi-deletes async https://review.openstack.org/648263 | 21:00 |
| timburke | clayg, tdasilva, zaitcev meeting ping | 21:01 |
| *** NM has quit IRC | 21:19 | |
| *** rcernin has joined #openstack-swift | 22:06 | |
| openstackgerrit | Tim Burke proposed openstack/swift master: Make staticweb return URL-encoded Location headers https://review.openstack.org/571907 | 22:22 |
| timburke | zaitcev, mattoliverau: in light of my realization about what went wrong on https://review.openstack.org/#/c/651004/, do you guys have a preference on whether we continue on https://review.openstack.org/#/c/651277/ ? | 22:33 |
| patchbot | patch 651004 - swift - Remove our urlparse wrapper - 3 patch sets | 22:33 |
| patchbot | patch 651277 - swift - Fix ostestr invocation - 2 patch sets | 22:33 |
| zaitcev | timburke: I have to say that 651277 looked a little fishy to me, because the error message in the logs didn't match the one in the linked commit. But I thought that if it fixes an issue you're experiencing, and it's not harming anything, then in it goes. If you realized that your tests weren't clean enough and it does not actually fix your problem, then abandon it. Sounds like a simple enough determinant...? | 22:46 |
| *** tkajinam has joined #openstack-swift | 22:53 | |
| *** tokuzfunpi has joined #openstack-swift | 23:00 | |
| *** tokuzfunpi has quit IRC | 23:04 | |
| *** tokuzfunpi has joined #openstack-swift | 23:22 | |
| openstackgerrit | Tim Burke proposed openstack/python-swiftclient master: Fix SLO re-upload https://review.openstack.org/651666 | 23:22 |
| *** tokuzfunpi has quit IRC | 23:27 | |
| *** ianychoi has joined #openstack-swift | 23:46 | |
| *** tokuzfunpi has joined #openstack-swift | 23:47 | |
| *** tokuzfunpi has quit IRC | 23:49 | |
| *** rcernin has quit IRC | 23:52 | |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!