| *** baojg has joined #openstack-swift | 01:34 | |
| *** baojg has quit IRC | 01:38 | |
| zaitcev_ | mattoliverau: I'm tempted to land https://review.opendev.org/659390 - py3 diskfile. But maybe I'm missing something. | 01:38 |
|---|---|---|
| patchbot | patch 659390 - swift - py3: port diskfile - 2 patch sets | 01:38 |
| *** tkajinam has quit IRC | 03:03 | |
| *** tkajinam has joined #openstack-swift | 03:04 | |
| *** tkajinam has quit IRC | 03:20 | |
| *** tkajinam has joined #openstack-swift | 03:28 | |
| *** psachin has joined #openstack-swift | 03:29 | |
| *** zaitcev_ has quit IRC | 03:33 | |
| *** zaitcev_ has joined #openstack-swift | 03:45 | |
| *** ChanServ sets mode: +v zaitcev_ | 03:45 | |
| *** zaitcev_ has quit IRC | 04:15 | |
| *** itlinux has quit IRC | 04:46 | |
| *** pcaruana has joined #openstack-swift | 05:02 | |
| *** pcaruana has quit IRC | 05:11 | |
| openstackgerrit | Tim Burke proposed openstack/swift master: s3api: Fix ETag when copying a MU part from another MU https://review.opendev.org/660249 | 05:12 |
| *** e0ne has joined #openstack-swift | 05:23 | |
| *** e0ne has quit IRC | 05:24 | |
| *** jistr is now known as jistr|mtg | 06:20 | |
| *** hseipp has joined #openstack-swift | 06:46 | |
| *** ccamacho has joined #openstack-swift | 06:56 | |
| *** tesseract has joined #openstack-swift | 07:06 | |
| *** rdejoux has joined #openstack-swift | 07:11 | |
| *** rcernin has quit IRC | 07:19 | |
| *** pcaruana has joined #openstack-swift | 07:21 | |
| *** pcaruana has quit IRC | 07:39 | |
| *** e0ne has joined #openstack-swift | 07:42 | |
| *** e0ne has quit IRC | 07:57 | |
| *** e0ne has joined #openstack-swift | 08:01 | |
| *** tkajinam has quit IRC | 08:12 | |
| *** diablo_rojo has joined #openstack-swift | 08:17 | |
| *** jistr|mtg is now known as jistr | 08:46 | |
| *** diablo_rojo has quit IRC | 09:03 | |
| *** openstackstatus has quit IRC | 09:13 | |
| *** openstackstatus has joined #openstack-swift | 09:15 | |
| *** ChanServ sets mode: +v openstackstatus | 09:15 | |
| *** diablo_rojo has joined #openstack-swift | 09:38 | |
| *** henriqueof has joined #openstack-swift | 11:44 | |
| *** henriqueof has quit IRC | 12:04 | |
| *** diablo_rojo has quit IRC | 12:33 | |
| *** psachin has quit IRC | 12:34 | |
| *** baojg has joined #openstack-swift | 12:46 | |
| openstackgerrit | Alexandre Lécuyer proposed openstack/swift feature/losf: Add tests for vfile_utils https://review.opendev.org/660381 | 13:12 |
| *** baojg has quit IRC | 13:22 | |
| *** m75abrams has joined #openstack-swift | 13:32 | |
| *** m75abrams has quit IRC | 13:36 | |
| *** itlinux has joined #openstack-swift | 13:42 | |
| *** henriqueof has joined #openstack-swift | 13:47 | |
| *** mrjk_ has joined #openstack-swift | 13:57 | |
| *** lifeless_ has joined #openstack-swift | 14:01 | |
| *** early` has joined #openstack-swift | 14:02 | |
| *** ktsuyuzaki has joined #openstack-swift | 14:04 | |
| *** ChanServ sets mode: +v ktsuyuzaki | 14:04 | |
| *** early has quit IRC | 14:06 | |
| *** kota_ has quit IRC | 14:06 | |
| *** lifeless has quit IRC | 14:06 | |
| *** dosaboy has quit IRC | 14:06 | |
| *** henriqueof has quit IRC | 14:06 | |
| *** mrjk has quit IRC | 14:06 | |
| *** itlinux has quit IRC | 14:13 | |
| notmyname | good morning | 15:13 |
| *** admin6 has joined #openstack-swift | 15:22 | |
| admin6 | Hi team, I’m trying to add a new storage policy. I updated the swift.conf file with : [storage-policy:2] name = SSD, create a ring file object-2.builder, add some disk into it and rebalance, update the conf dirs on all my servers and also restart the swift processe to be sure. But when I try to create a first container into this policy, I got an error : "Invalid X-Storage-Policy-Index 2". Did I miss something? | 15:29 |
| notmyname | sounds like you've done the right things. probably good to double check things. `swift-recon --md5` might be a good sanity check | 15:34 |
| *** mikecmpbll has joined #openstack-swift | 15:44 | |
| *** e0ne has quit IRC | 15:55 | |
| *** ccamacho has quit IRC | 15:57 | |
| timburke | fwiw, sounds like the proxy is correctly translating x-storage-policy: SSD to x-storage-policy-index: 2, but the container servers don't know what to do with that -- i'd look at the container servers first, make sure they really got restarted after the swift.conf update | 16:06 |
| *** gyee has joined #openstack-swift | 16:15 | |
| timburke | hmm... so we have tests that verify that you can do a PUT to an object with some blank sysmeta... and that the sysmeta will be in the headers on a subsequent GET | 16:22 |
| timburke | this seems... weird to me. if you try to PUT blank *user meta*, we drop it on the floor. why's sysmeta special in this regard? | 16:22 |
| timburke | oh... i stand corrected. we allow blank user meta, too... why'd i think we didn't? | 16:27 |
| timburke | still -- why do we save blank meta? more to my problem: how do i convince COPY to drop some sysmeta on the floor? | 16:28 |
| openstackgerrit | Tim Burke proposed openstack/swift master: Start transition to boto3 in s3api tests. https://review.opendev.org/657149 | 16:35 |
| *** mikecmpbll has quit IRC | 16:39 | |
| timburke | it's particularly problematic when trying to scrub a x-object-sysmeta-container-update-override... | 16:42 |
| *** itlinux has joined #openstack-swift | 16:57 | |
| *** hseipp has quit IRC | 17:05 | |
| *** itlinux has quit IRC | 17:22 | |
| *** ybunker has joined #openstack-swift | 17:43 | |
| *** itlinux has joined #openstack-swift | 17:44 | |
| ybunker | hi all, quick question.. i can't access from the regular copies to ~450 objects.. if I issue a `swift-get-nodes` command I can see (trying with the curl command) that the object exist (the three copies), but on the `handoff` nodes.. but..the user tries to issue a `swift download` command and of course its being truncated on the first 6 locations, and the user needs those objects asap, so the question is... how can I `script` something in or | 17:47 |
| ybunker | der to based on `container/object` to forcefully download those objects on the handoff locations? | 17:47 |
| notmyname | ybunker: check out this config option in the proxy server. https://github.com/openstack/swift/blob/master/etc/proxy-server.conf-sample#L216-L219 | 17:52 |
| notmyname | it defaults to 2* replica count, so you get 6 in a 3 replica policy | 17:53 |
| notmyname | but if you raise it, then objects that don't exist will take longer (that's the cost: missing objects are more expensive) | 17:53 |
| ybunker | notmyname: oh thanks :-), that's really cool, so changing the 2 for example for 3, would lead to try the first 9 entries.. considering replicas = 3, right? | 17:54 |
| timburke | out of curiosity, how far down the handoff list do you have to go to get to good data? | 17:54 |
| notmyname | right | 17:54 |
| notmyname | yes, timburke has a good question | 17:54 |
| notmyname | and you should also work on your replication processes to get the data on the primary nodes | 17:54 |
| ybunker | depends.. on some objects it's just two entries.. but for others maybe 15...16 entries | 17:55 |
| timburke | eep. yeah, kicking it up to 5 or 6 * replicas seems painful, but it should definitely work. notmyname's right as to next steps once you've fixed the availability problem: gotta get replication happy | 17:56 |
| ybunker | yes, once the guys download the ~450 files will change it back to the normal parameters | 17:57 |
| timburke | i wonder if there's a way we could prioritize replication for partitions where my disk is further out in the handoff list... | 17:58 |
| ybunker | thanks a lot for the quick response guys, i will use that parameter and see if we can download those files and then configure back with the original value. | 18:05 |
| *** rdejoux has quit IRC | 18:06 | |
| notmyname | ybunker: I'd *strongly* advise you make sure your replication processes are running | 18:07 |
| notmyname | we can help in here if you've got issues there; it's really important for the health of the data in the cluster | 18:07 |
| ybunker | notmyname: yeah, the fully disk scenarios from last weeks got almost resolved, lower down to 88% :-(, once we got to have all the nodes (olds + newly ones) balance, then will change back the parameters of handoff_ | 18:09 |
| ybunker | the problem with the replication processes is that latency goes really high once we started them, so we are still in a 4h window at night | 18:10 |
| *** itlinux has quit IRC | 18:40 | |
| *** gyee has quit IRC | 19:07 | |
| *** gyee has joined #openstack-swift | 19:19 | |
| *** ybunker has quit IRC | 19:28 | |
| *** mikecmpbll has joined #openstack-swift | 20:05 | |
| openstackgerrit | Tim Burke proposed openstack/swift master: Add operator tool to async-delete some or all objects in a container https://review.opendev.org/659162 | 20:21 |
| openstackgerrit | Tim Burke proposed openstack/swift master: WIP: s3api: Make multi-deletes async https://review.opendev.org/648263 | 20:21 |
| *** mikecmpbll has quit IRC | 20:30 | |
| *** mikecmpbll has joined #openstack-swift | 20:32 | |
| *** henriqueof has joined #openstack-swift | 20:35 | |
| *** e0ne has joined #openstack-swift | 20:39 | |
| openstackgerrit | Tim Burke proposed openstack/swift master: WIP: s3api: Make multi-deletes async https://review.opendev.org/648263 | 20:40 |
| *** e0ne has quit IRC | 20:56 | |
| openstackgerrit | Tim Burke proposed openstack/swift master: s3api: Fix ETag when copying a MU part from another MU https://review.opendev.org/660249 | 21:02 |
| *** djhankb has quit IRC | 21:02 | |
| openstackgerrit | Tim Burke proposed openstack/swift master: container: Unify storage-policy error paths https://review.opendev.org/660525 | 21:09 |
| *** mikecmpbll has quit IRC | 21:10 | |
| *** diablo_rojo has joined #openstack-swift | 21:19 | |
| openstackgerrit | Tim Burke proposed openstack/swift master: py3: port dlo func tests https://review.opendev.org/642920 | 21:19 |
| *** henriqueof has quit IRC | 21:21 | |
| *** zaitcev has joined #openstack-swift | 21:24 | |
| *** ChanServ sets mode: +v zaitcev | 21:24 | |
| openstackgerrit | Tim Burke proposed openstack/swift master: py3: port staticweb and domain_remap func tests https://review.opendev.org/647923 | 21:27 |
| openstackgerrit | Tim Burke proposed openstack/swift master: py3: Bring functional/test_object.py under test; add func-ec-py37 job https://review.opendev.org/645895 | 21:35 |
| timburke | fyi, i broke the func tests out to be independent patches now that we have the gate job landed | 21:36 |
| *** rcernin has joined #openstack-swift | 22:05 | |
| *** tesseract has quit IRC | 22:07 | |
| openstackgerrit | Tim Burke proposed openstack/swift master: Start transition to boto3 in s3api tests. https://review.opendev.org/657149 | 22:10 |
| openstackgerrit | John Dickinson proposed openstack/swift master: Fix quarantine when object path is not a directory https://review.opendev.org/653748 | 22:12 |
| *** itlinux has joined #openstack-swift | 22:26 | |
| *** rcernin has quit IRC | 22:40 | |
| *** rcernin has joined #openstack-swift | 22:41 | |
| *** itlinux has quit IRC | 22:43 | |
| *** zaitcev has quit IRC | 22:45 | |
| *** zaitcev has joined #openstack-swift | 22:57 | |
| *** ChanServ sets mode: +v zaitcev | 22:57 | |
| *** tkajinam has joined #openstack-swift | 22:59 | |
| openstackgerrit | Merged openstack/python-swiftclient master: Replace git.openstack.org URLs with opendev.org URLs https://review.opendev.org/654777 | 23:08 |
| openstackgerrit | Merged openstack/swift master: Make test_read_write_acl_listings more robust https://review.opendev.org/653655 | 23:08 |
| *** itlinux has joined #openstack-swift | 23:44 | |
| *** itlinux_ has joined #openstack-swift | 23:47 | |
| *** itlinux has quit IRC | 23:49 | |
| *** itlinux_ has quit IRC | 23:57 | |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!