| *** tonanhngo has quit IRC | 00:08 | |
| *** tonanhngo has joined #openstack-swift | 00:26 | |
| *** tonanhngo has quit IRC | 00:30 | |
| *** tonanhngo has joined #openstack-swift | 00:33 | |
| *** tonanhngo_ has joined #openstack-swift | 00:35 | |
| *** tonanhngo has quit IRC | 00:37 | |
| *** tonanhngo_ has quit IRC | 00:39 | |
| *** jerrygb has joined #openstack-swift | 00:50 | |
| *** catintheroof has quit IRC | 00:52 | |
| *** jerrygb has quit IRC | 00:56 | |
| *** gyee has quit IRC | 01:03 | |
| *** klrmn has quit IRC | 01:13 | |
| *** Sukhdev has quit IRC | 01:25 | |
| *** m_kazuhiro has joined #openstack-swift | 01:30 | |
| m_kazuhiro | good morning | 01:31 |
|---|---|---|
| *** JimCheung has quit IRC | 01:37 | |
| *** jerrygb has joined #openstack-swift | 01:43 | |
| *** chsc has joined #openstack-swift | 01:49 | |
| *** chsc has joined #openstack-swift | 01:49 | |
| charz | m_kazuhiro: good morning | 01:54 |
| m_kazuhiro | charz: good morning! | 01:54 |
| *** zhurong has joined #openstack-swift | 02:11 | |
| *** bkopilov has quit IRC | 02:13 | |
| *** klrmn has joined #openstack-swift | 02:35 | |
| *** gkadam has joined #openstack-swift | 02:53 | |
| *** jerrygb has quit IRC | 02:55 | |
| *** zhurong has quit IRC | 03:21 | |
| mattoliverau | m_kazuhiro, charz morning | 03:25 |
| *** zhurong has joined #openstack-swift | 03:27 | |
| *** zhurong has quit IRC | 03:31 | |
| *** Sukhdev has joined #openstack-swift | 03:33 | |
| m_kazuhiro | mattoliverau: morning! | 03:37 |
| *** winggundamth has joined #openstack-swift | 03:39 | |
| *** Sukhdev has quit IRC | 03:40 | |
| *** bkopilov has joined #openstack-swift | 03:44 | |
| *** jerrygb has joined #openstack-swift | 04:15 | |
| *** m_kazuhiro_ has joined #openstack-swift | 04:21 | |
| *** m_kazuhiro has quit IRC | 04:22 | |
| *** zhurong has joined #openstack-swift | 04:31 | |
| *** jerrygb has quit IRC | 04:32 | |
| *** jerrygb has joined #openstack-swift | 04:32 | |
| *** JimCheung has joined #openstack-swift | 04:38 | |
| *** JimCheung has quit IRC | 04:42 | |
| *** psachin has joined #openstack-swift | 04:44 | |
| *** jerrygb has quit IRC | 04:47 | |
| mahatic | mattoliverau: m_kazuhiro_ kota_ charz o/ | 04:54 |
| m_kazuhiro_ | mahatic: o/ | 04:54 |
| mattoliverau | mahatic: morning | 04:57 |
| kota_ | mahatic: morning | 04:57 |
| *** chsc has quit IRC | 05:22 | |
| *** ivru has joined #openstack-swift | 05:26 | |
| *** ChubYann has quit IRC | 05:32 | |
| *** cshastri has joined #openstack-swift | 05:37 | |
| *** adriant has quit IRC | 05:50 | |
| ivru | https://pbs.twimg.com/media/C9xpjGpXcAAzNvH.jpg:large | 06:10 |
| *** tonanhngo has joined #openstack-swift | 06:10 | |
| *** zhurong has quit IRC | 06:12 | |
| *** rcernin has joined #openstack-swift | 06:28 | |
| *** klrmn has quit IRC | 06:28 | |
| ivru | wrong channel, sorry guys | 06:34 |
| *** hseipp has joined #openstack-swift | 06:40 | |
| *** pcaruana has joined #openstack-swift | 06:40 | |
| *** tesseract has joined #openstack-swift | 06:56 | |
| *** d0ugal has quit IRC | 06:56 | |
| *** tonanhngo_ has joined #openstack-swift | 07:03 | |
| *** tonanhngo has quit IRC | 07:04 | |
| *** jaosorior_away is now known as jaosorior | 07:11 | |
| *** zhurong has joined #openstack-swift | 07:11 | |
| *** m_kazuhiro_ has quit IRC | 07:22 | |
| *** jamielennox has quit IRC | 07:28 | |
| *** gkadam has quit IRC | 07:29 | |
| *** gkadam has joined #openstack-swift | 07:29 | |
| *** jamielennox has joined #openstack-swift | 07:42 | |
| *** geaaru has joined #openstack-swift | 07:48 | |
| *** foutatoro has joined #openstack-swift | 07:55 | |
| *** geaaru has quit IRC | 07:56 | |
| *** cbartz has joined #openstack-swift | 08:18 | |
| acoles | good morning | 08:20 |
| *** vinsh_ has quit IRC | 08:22 | |
| acoles | kota_: ping | 08:24 |
| *** vinsh has joined #openstack-swift | 08:27 | |
| *** JimCheung has joined #openstack-swift | 08:39 | |
| *** JimCheung has quit IRC | 08:43 | |
| acoles | kota_: I wanted to catch up with you re composite ring patches - I am wondering if we are ready to squash patch 449310 and patch 441921? | 08:46 |
| patchbot | https://review.openstack.org/#/c/449310/ - swift - Add id to RingBuilder to differentiate rings in co... | 08:46 |
| patchbot | https://review.openstack.org/#/c/441921/ - swift - Add Composite Ring Functionality | 08:46 |
| acoles | Also I think there is a change still needed that clayg requested to how (if?) metadata is added to the ring (for it to not be completely unstructured but under a fixed key) | 08:47 |
| kota_ | good morning, acoles | 08:48 |
| kota_ | acoles: sorry, I was not able to look back again that but my first look for patch 449310 patch set 8 (the latest), i think it can be squashed into the parent | 08:50 |
| patchbot | https://review.openstack.org/#/c/449310/ - swift - Add id to RingBuilder to differentiate rings in co... | 08:50 |
| kota_ | i'm feeling | 08:51 |
| kota_ | re: structured vs unstructured for the metadata. I don't have strong opinion that we should have structured. sometimes unstructured could be useful so I need to catch up the fear case we should prevent with unstructured metadata | 08:52 |
| kota_ | from clayg | 08:52 |
| kota_ | acoles: ^^ | 08:52 |
| acoles | kota_: ok. thanks for looking. | 08:53 |
| acoles | kota_: I will do the squash, then we can continue the discussion with clayg re metadata | 08:53 |
| acoles | kota_: maybe i can put together a diff for the metadata so we understand what the change would be | 08:53 |
| kota_ | acoles: diff at somewhere like gist/paste/another patch? | 08:56 |
| acoles | kota_: either a git or maybe easier to put a follow on patch on gerrit? then squash again if we agree? | 08:57 |
| acoles | what do you prefer? | 08:57 |
| kota_ | acoles: just a confirmation, anywhere if i can reach is fine | 08:58 |
| acoles | kota_: ok | 09:01 |
| kota_ | acoles: thanks | 09:08 |
| kota_ | acoles: and sorry, i have to leave here today, will look at the logs tommorow morning to catch up what progressed at that time. | 09:08 |
| acoles | kota_: NP I will be away tomorrow back Monday. thanks! | 09:09 |
| *** gkadam is now known as gkadam-afk | 09:26 | |
| openstackgerrit | Romain LE DISEZ proposed openstack/swift master: Fix SSYNC failing to replicate unexpired object https://review.openstack.org/456921 | 09:30 |
| *** zhurong has quit IRC | 10:19 | |
| *** hosseini has joined #openstack-swift | 10:23 | |
| *** cshastri has quit IRC | 10:33 | |
| openstackgerrit | Merged openstack/swift master: Update swift version in storage-install-rdo.rst. https://review.openstack.org/456806 | 10:39 |
| openstackgerrit | Merged openstack/swift master: Update swift version in storage-install-ubuntu. https://review.openstack.org/456808 | 10:43 |
| *** cshastri has joined #openstack-swift | 10:45 | |
| *** tonanhngo has joined #openstack-swift | 10:53 | |
| *** tonanhngo_ has quit IRC | 10:54 | |
| *** geaaru has joined #openstack-swift | 10:54 | |
| *** mvk has quit IRC | 11:06 | |
| *** bkopilov has quit IRC | 11:07 | |
| *** jaosorior has quit IRC | 11:09 | |
| openstackgerrit | Merged openstack/swift master: Remove redundant assignments of self._datadir https://review.openstack.org/457807 | 11:10 |
| *** zhurong has joined #openstack-swift | 11:17 | |
| *** jaosorior has joined #openstack-swift | 11:20 | |
| *** cshastri has quit IRC | 11:21 | |
| *** cshastri has joined #openstack-swift | 11:22 | |
| *** openstackgerrit has quit IRC | 11:32 | |
| *** winggundamth has quit IRC | 11:47 | |
| *** jerrygb has joined #openstack-swift | 11:49 | |
| *** jerrygb has quit IRC | 11:54 | |
| *** jerrygb has joined #openstack-swift | 12:09 | |
| *** klamath has joined #openstack-swift | 12:12 | |
| *** klamath has quit IRC | 12:12 | |
| *** klamath has joined #openstack-swift | 12:13 | |
| *** jerrygb has quit IRC | 12:14 | |
| *** mvk has joined #openstack-swift | 12:15 | |
| *** catintheroof has joined #openstack-swift | 12:27 | |
| *** catintheroof has quit IRC | 12:30 | |
| *** catintheroof has joined #openstack-swift | 12:30 | |
| *** catintheroof has quit IRC | 12:40 | |
| *** catintheroof has joined #openstack-swift | 12:41 | |
| *** catintheroof has quit IRC | 12:41 | |
| *** catintheroof has joined #openstack-swift | 12:41 | |
| *** catintheroof has quit IRC | 12:43 | |
| *** catintheroof has joined #openstack-swift | 12:43 | |
| *** bkopilov has joined #openstack-swift | 12:50 | |
| *** gkadam-afk has quit IRC | 12:51 | |
| *** jerrygb has joined #openstack-swift | 12:56 | |
| dewanee_ | hello guys | 13:09 |
| *** ig0r_ has joined #openstack-swift | 13:35 | |
| *** chlong has joined #openstack-swift | 13:51 | |
| *** vint_bra has joined #openstack-swift | 13:52 | |
| dewanee_ | an update on the bulk delete operation | 14:11 |
| dewanee_ | it just works | 14:11 |
| dewanee_ | it was a configuration mistake on the proxy pipeline | 14:12 |
| dewanee_ | we can currently delete files and folders in a container even as regular users provided the correct ACL are set | 14:13 |
| *** xinli has joined #openstack-swift | 14:16 | |
| hugokuo | morning guys | 14:18 |
| hugokuo | Two questions here : | 14:18 |
| *** Jeffrey4l has quit IRC | 14:19 | |
| hugokuo | 1. How Swift worker determine which TZ for logging ? We seen nodes have same TZ in CEST from date output. But nodes have different TZ in Swift logs. I wonder how Swift pick TZ for logging in CentOS 6 | 14:20 |
| hugokuo | 2. Do we have existing topic about the improvement for single token with high concurrency which causes timeout while proxy attempts to validate the token from the memcached host | 14:21 |
| *** cshastri has quit IRC | 14:38 | |
| *** openstackgerrit has joined #openstack-swift | 14:39 | |
| openstackgerrit | Pavel Kvasnička proposed openstack/swift master: WIP: DB replicator cleanup https://review.openstack.org/448480 | 14:39 |
| openstackgerrit | Pavel Kvasnička proposed openstack/swift master: DB replicator cleanup https://review.openstack.org/448480 | 14:39 |
| *** psachin has quit IRC | 14:42 | |
| *** stephen_m has joined #openstack-swift | 14:49 | |
| notmyname | hugokuo: perhaps related, https://review.openstack.org/#/c/331369/ | 14:51 |
| patchbot | patch 331369 - swift - Always set swift processes to use UTC (MERGED) | 14:51 |
| *** ivru has quit IRC | 14:52 | |
| hugokuo | @notmyname thx. is this patch will be delivered in next Swift release ? | 14:53 |
| notmyname | hugokuo: correct. it's not yet released in any upstream tag, but it's going to be in the next release (Real Soon Now) | 14:53 |
| hugokuo | @notmyname got it. thx | 14:53 |
| *** bkopilov has quit IRC | 14:54 | |
| *** PavelK has quit IRC | 14:55 | |
| openstackgerrit | Ondřej Nový proposed openstack/swift master: Fix unit tests on i386 and other archs https://review.openstack.org/458539 | 14:58 |
| *** rcernin has quit IRC | 15:01 | |
| *** xinli has quit IRC | 15:07 | |
| *** xinli has joined #openstack-swift | 15:14 | |
| *** klrmn has joined #openstack-swift | 15:14 | |
| *** pcaruana has quit IRC | 15:14 | |
| *** oshritf_ has quit IRC | 15:15 | |
| notmyname | onovy: my first thought when reading that commit message was "is onovy running on 32bit machines?!" | 15:19 |
| notmyname | but I see that it's only about IO priority | 15:20 |
| *** chsc has joined #openstack-swift | 15:26 | |
| *** Jeffrey4l has joined #openstack-swift | 15:31 | |
| *** ig0r_ has quit IRC | 16:01 | |
| *** newmember has joined #openstack-swift | 16:03 | |
| *** tonanhngo has quit IRC | 16:08 | |
| *** zhurong has quit IRC | 16:11 | |
| *** cbartz has quit IRC | 16:13 | |
| timburke | dewanee_: glad to hear it! out of curiosity, what was the misconfiguration? | 16:22 |
| *** newmember has quit IRC | 16:22 | |
| *** newmember has joined #openstack-swift | 16:23 | |
| *** foutatoro has quit IRC | 16:27 | |
| openstackgerrit | OpenStack Proposal Bot proposed openstack/swift master: Updated from global requirements https://review.openstack.org/88736 | 16:27 |
| dewanee_ | timburke, was the lack of bulk support in the pipeline. And apparently clients are not that good in checking that capability | 16:29 |
| dewanee_ | I still need to do some digging because a few of the explanation I gave myself in the past weeks proved to be wrong | 16:29 |
| timburke | :-/ that's unfortunate -- was this python-swiftclient? i know that one looks at /info, and bulk should only show up there if it's actually enabled... | 16:30 |
| openstackgerrit | Alistair Coles proposed openstack/swift master: Add Composite Ring Functionality https://review.openstack.org/441921 | 16:33 |
| openstackgerrit | Alistair Coles proposed openstack/swift master: Add CompositeRingBuilder class https://review.openstack.org/453827 | 16:33 |
| openstackgerrit | Alistair Coles proposed openstack/swift master: Make composite_metadata an attribute of RingData https://review.openstack.org/458585 | 16:33 |
| *** tesseract has quit IRC | 16:34 | |
| dewanee_ | I need to check with my collegue but afaik we were troubles also with that not accepting multiple files. I'll give you more details tomorrow. | 16:35 |
| dewanee_ | *were having troubles | 16:36 |
| timburke | oh! maybe something where https://review.openstack.org/#/c/454849/ would help? | 16:38 |
| patchbot | patch 454849 - python-swiftclient - respect bulk delete page size and fix logic error | 16:38 |
| *** JimCheung has joined #openstack-swift | 16:38 | |
| notmyname | timburke: oh! I didn't see your last review comment there | 16:39 |
| acoles | kota_: clayg I made some more notes on etherpad about the interface to a possible CompositeRingBuilder class (under heading Do we pass builders or builder_files to a CompositeRingBuilder class?) - value your thoughts - https://etherpad.openstack.org/p/composite_rings | 16:39 |
| *** geaaru has quit IRC | 16:40 | |
| timburke | notmyname: then i'm glad i brought it up :-) | 16:40 |
| notmyname | timburke: if you've still got that diff locally, please feel free to push it over. lgtm (at a quick glance) | 16:40 |
| notmyname | yeah, I was kinda busy the last few days and didn't get a chance to do much upstream. | 16:42 |
| *** SkyRocknRoll has joined #openstack-swift | 16:43 | |
| *** _JZ_ has joined #openstack-swift | 16:44 | |
| openstackgerrit | Tim Burke proposed openstack/python-swiftclient master: respect bulk delete page size and fix logic error https://review.openstack.org/454849 | 16:47 |
| notmyname | timburke: thanks | 16:47 |
| *** jaosorior has quit IRC | 16:49 | |
| *** SkyRocknRoll has quit IRC | 16:50 | |
| timburke | man, i really should have looked at https://review.openstack.org/#/c/302494/ sooner :-( that patch should not have stood unreviewed for a year... | 16:54 |
| patchbot | patch 302494 - swift - Sync metadata in 'rsync_then_merge' in db_replicator | 16:54 |
| *** newmember has quit IRC | 16:56 | |
| *** vinsh has quit IRC | 17:00 | |
| *** vinsh has joined #openstack-swift | 17:02 | |
| *** mvk has quit IRC | 17:03 | |
| *** tonanhngo has joined #openstack-swift | 17:13 | |
| *** tonanhngo has quit IRC | 17:17 | |
| *** tonanhngo has joined #openstack-swift | 17:17 | |
| *** mvk has joined #openstack-swift | 17:25 | |
| *** ig0r_ has joined #openstack-swift | 17:26 | |
| *** hseipp has quit IRC | 17:34 | |
| openstackgerrit | Tim Burke proposed openstack/swift master: fixup! Sync metadata in 'rsync_then_merge' in db_replicator https://review.openstack.org/457808 | 17:38 |
| *** xinli has quit IRC | 18:08 | |
| clayg | fixup! I like it for reviews, but not so much for the commit log? Is it just a place holder commit message? | 18:10 |
| timburke | clayg: i use it whenever i fully intend that it should just get squashed into the parent | 18:10 |
| clayg | timburke: full ack - love it - can I steal!? | 18:11 |
| timburke | absolutely! | 18:11 |
| * clayg appropriates more brillance from timburke - nom nom nom | 18:12 | |
| *** Renich___ has joined #openstack-swift | 18:26 | |
| *** ChubYann has joined #openstack-swift | 18:26 | |
| timburke | clayg: i'm looking through the open backports... what are your thoughts on https://github.com/openstack/swift/commit/69f7be9 ? that UpgradeImpact makes me think pavel was probably right to not include that in the backports he proposed... but i figure i oughta get your feel for it, too | 18:27 |
| *** SkyRocknRoll has joined #openstack-swift | 18:33 | |
| *** Sukhdev has joined #openstack-swift | 18:34 | |
| *** jerrygb has quit IRC | 18:39 | |
| *** Sukhdev has quit IRC | 18:40 | |
| *** Sukhdev has joined #openstack-swift | 18:40 | |
| clayg | timburke: i think that is nbd | 18:42 |
| *** Sukhdev has quit IRC | 18:43 | |
| clayg | timburke: i always considered it mostly a doc bug - I think I sorta realized reclaim-age needed to be in DEFAULT section the first time I went to change the value | 18:43 |
| clayg | Only later I realized we had it doc'd wrong | 18:43 |
| clayg | then during doc fix people pointed out how insane the config plumbing was | 18:43 |
| clayg | ... and the stupid bug where one place we like... ignored the configured value or some bs? | 18:43 |
| clayg | that's not "upgrade impact" - that's ftfy | 18:44 |
| clayg | i just don't think it's a big enough deal to go backporting IMHO - unless someone is asking for it? | 18:44 |
| *** PavelK has joined #openstack-swift | 18:45 | |
| timburke | for context, i'm looking at https://review.openstack.org/#/c/458485/ -- there were some minor differences as a result of that commit | 18:45 |
| patchbot | patch 458485 - swift (stable/newton) - Better optimistic lock in get_hashes | 18:45 |
| clayg | oh right - and the ftfy is like "if you run with a 1 second reclaim age it's weird" - but OMG dude - why are you running with a 1 second reclaim age!? The actual impact on "sane" configurations was nominal - and fixing the plumbing doesn't stop you from putting one reclaime in reconstructor and another in replicator - or different values on different configs - which are both worse/operator-error | 18:46 |
| PavelK | timburke: thanks. in newton is not merged "nice to have" fixes - so little changes were necessary | 18:46 |
| clayg | PavelK: ohai! | 18:46 |
| timburke | but now i'm starting to wonder how much *that* one really belongs on newton... i think ocata would make sense, given the stable policy, but it seems like https://review.openstack.org/#/c/433565/ was the one that really fixed the bug? | 18:46 |
| patchbot | patch 433565 - swift (stable/newton) - Fix race in new partitions detecting new/invalid s... (MERGED) | 18:46 |
| PavelK | clayg: :-) | 18:47 |
| clayg | ugh, everytime with the _get_hashes *_*_get_hashes I hate myself so much :'( | 18:48 |
| clayg | I don't really see what changed with patch 458485 from the backport | 18:48 |
| patchbot | https://review.openstack.org/#/c/458485/ - swift (stable/newton) - Better optimistic lock in get_hashes | 18:48 |
| timburke | PavelK: yup, and i think clayg and i agree with your assessment that it was better to have a slightly-changed diff rather than include a not-really-necessary backport | 18:48 |
| clayg | but I'm sure PavelK did it right | 18:48 |
| PavelK | clayg: I'm back in the evening because another performance issue don't give me to have a rest... | 18:49 |
| clayg | OH the reclaim_age - timburke you move too fast | 18:49 |
| clayg | timburke: is it weird you're constantly looking around where is everyone and we all a mile behind staring at our belly buttons | 18:49 |
| *** Sukhdev has joined #openstack-swift | 18:50 | |
| clayg | PavelK: sorry to hear that, but always happy to get to chat with you :\ | 18:50 |
| *** SkyRocknRoll has quit IRC | 18:51 | |
| *** jerrygb has joined #openstack-swift | 18:51 | |
| timburke | clayg: PavelK definitely did it right. patch looks great, and i'll include some notes for notmyname to call out the differences and why they're fine. right now i'm mainly looking for why that patch *definitely* belongs on the 2.10.x series | 18:52 |
| clayg | timburke: probably the perforamnce thing were we rehash twice? it's pretty crappy we've done that since the hash invalidations file hit. | 18:53 |
| clayg | and the race/correctness w/e | 18:53 |
| PavelK | timburke: and there was posibility of race contition when force_rewrite was used | 18:56 |
| *** silor has joined #openstack-swift | 18:56 | |
| clayg | typo in commit msg - maybe I should have left it? I don't really know how to backport | 18:57 |
| *** tonanhngo has quit IRC | 19:01 | |
| PavelK | clayg: timburke: a race condition is for new hashes.pkl fiels - and there is next patch #458486 that solves another situation with new pkl file - and because it removes lines witch I don't like I will be happy merged it too and have it in production soon | 19:01 |
| patchbot | https://review.openstack.org/#/c/458486/ - swift (stable/newton) - Fix race when consolidating new partition | 19:01 |
| PavelK | in that next patch was removed change of obj.test_reconstructor unit test because in stable/newton is another check that fails on unexpected file. I don't know what change made the difference but it seems correct that there is a difference in reconstructor behaviour. | 19:07 |
| *** jerrygb has quit IRC | 19:07 | |
| *** silor has quit IRC | 19:08 | |
| timburke | i'm a little sad that neither of those newton backports link to a bug in their commit messages :-/ | 19:08 |
| clayg | timburke: yeah fix that? lp bug #1651530 | 19:09 |
| openstack | Launchpad bug 1651530 in OpenStack Object Storage (swift) "suffix hash invalidation may be lost" [Critical,Fix released] https://launchpad.net/bugs/1651530 | 19:09 |
| timburke | clayg: related, or closes? | 19:10 |
| clayg | idk? | 19:10 |
| clayg | oh... wow, yeah idk! :P | 19:10 |
| timburke | yeah.... | 19:10 |
| clayg | I mean if it is *the* fix for *that* bug in the backport - the bug is alreayd closed | 19:10 |
| clayg | ask in #openstack-dev? | 19:10 |
| clayg | there's probably different processes that would work - not sure if there's good reasons to do one thing over something else | 19:11 |
| timburke | no no, it's fine to put closes-bug in a backport -- that's to mark the bug as closed in that branch. but it seems like https://review.openstack.org/#/c/433565/ already claims to close the bug for newton? | 19:12 |
| patchbot | patch 433565 - swift (stable/newton) - Fix race in new partitions detecting new/invalid s... (MERGED) | 19:12 |
| timburke | so do we need all three patches to properly close the bug? or is the situation already better? is there some other bug that would better describe what these new backorts fix? | 19:13 |
| clayg | GD, I can't keep all of them straight - it's possible some of the fixes we landed to master didn't get proper bug reports because we were under water trying to get it all fixed - so the book keeping at this point could be jacked - or maybe it just needs someone to do a table | 19:14 |
| clayg | <change on master> <bug> <backport> - and double check the p's and q's | 19:15 |
| clayg | PavelK: you wouldn't happen to already have such a spreadsheet do you? | 19:15 |
| PavelK | timburke: I do not recommend to delete any hashes.pkl manually before these two patches are in your production... | 19:17 |
| timburke | duly noted :-) | 19:17 |
| PavelK | clayg: I'm not sure to understant your question. but I have a list of fixes that must be merged in production. that's reason why I would like to have these merged in next newton release | 19:22 |
| *** newmember has joined #openstack-swift | 19:23 | |
| *** newmember has quit IRC | 19:27 | |
| *** newmember has joined #openstack-swift | 19:28 | |
| clayg | PavelK: i'm not even sure I can *remember* the mess that lead to this one https://gist.github.com/clayg/33f057fb1d9a9d5cc2fd84812e06a9fd - but I would say removing hashes.pkl in production is *fun*! | 19:28 |
| PavelK | clayg: off topic question: do you have unxpected data and ts files replication in your cluster? Unexpected means repeated data rsync and pushing .ts older than reclaim age? | 19:30 |
| clayg | pushing .ts older than reclaim age... what's your configured reclaim age? | 19:31 |
| clayg | before mahatic fixed that one thing I would have some clusters with inactive suffixes that could keep a .ts around after a reclaim because no one noticed it? | 19:32 |
| clayg | but I'm not aware of any rsync "noise" for old .ts getting shipped around except maybe ... from a old-primary? | 19:32 |
| clayg | I'll keep an eye out for it | 19:33 |
| PavelK | clayg: There was an idea to delete hashes.pkl from some members when lp bug #1651530 was found... | 19:33 |
| openstack | Launchpad bug 1651530 in OpenStack Object Storage (swift) "suffix hash invalidation may be lost" [Critical,Fix released] https://launchpad.net/bugs/1651530 | 19:33 |
| *** newmember has quit IRC | 19:34 | |
| *** newmember has joined #openstack-swift | 19:35 | |
| PavelK | clayg: I don't want to describe reason from you... I just want to know if you did notice it or not. | 19:35 |
| PavelK | I would like to know whether it is common problem or we just have a broken drive | 19:36 |
| *** jerrygb has joined #openstack-swift | 19:36 | |
| clayg | are we talking about two different things 1) fallout from the suffix hashing bug (non trivial, we had rolled invalidations file to most deployments by the time you found it and the more we looked and did manual cleanups - the more we were glad it was fixed!) 2) something about rsyncing tombstones with timestamps older than recliam (which to my expereince and understanding so far if it's happening | 19:38 |
| clayg | is unrelated to #1) | 19:39 |
| PavelK | timburke: summary - the patch is in master and I think that it is not neccessary to find a bug for a commit message. A performance regression and a low likely race condition is described in patches and you can find it not neccessary. We will have it in prod. and I think that it will help others to. | 19:42 |
| *** cebruns has quit IRC | 19:46 | |
| PavelK | clayg: yes, tombstones few weeks old, few months and few years were replicated before existing patches were applied. The #2 think is not related with them. | 19:47 |
| PavelK | clayg: I'm happy that you don't have more problems with replicator and hope that we will not need a new patch :-) | 19:48 |
| *** cebruns has joined #openstack-swift | 19:48 | |
| PavelK | I have to leave, have a noce day :-) | 19:56 |
| * notmyname back online | 19:58 | |
| *** PavelK has quit IRC | 20:03 | |
| *** spotz is now known as spotz_zzz | 20:15 | |
| *** spotz_zzz is now known as spotz | 20:17 | |
| *** Sukhdev has quit IRC | 20:28 | |
| *** stephen_m has quit IRC | 20:55 | |
| *** tonanhngo has joined #openstack-swift | 20:55 | |
| *** tonanhngo has quit IRC | 20:59 | |
| mattoliverau | Morning | 21:04 |
| notmyname | hello mattoliverau | 21:06 |
| onovy | notmyname: https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=860638 :] | 21:16 |
| openstack | Debian bug 860638 in src:swift "swift: FTBFS on i386: Test failures" [Serious,Open] | 21:16 |
| notmyname | onovy: yeah, I saw that. your patch to swift's test will resolve that right? | 21:17 |
| onovy | yep | 21:17 |
| onovy | https://review.openstack.org/#/c/458653/ // preparing fixed package | 21:17 |
| notmyname | great | 21:17 |
| patchbot | patch 458653 - deb-swift (debian/newton) - d/patches/FTBFS_i386.patch: Fixies FTBFS on i386 (... | 21:17 |
| notmyname | seems like your patch is ok. a nice follow-up would be to mock out the architecture and catch that we are doing the right thing on both ia64 and other platforms | 21:18 |
| notmyname | ie a new test | 21:18 |
| onovy | sounds good | 21:20 |
| openstackgerrit | Pete Zaitcev proposed openstack/swift master: PUT+POST and its development test https://review.openstack.org/427911 | 21:30 |
| *** _JZ_ has quit IRC | 21:31 | |
| zaitcev | okay, this should address Sam's critique at PTG | 21:31 |
| zaitcev | in time for Boston :-) | 21:31 |
| zaitcev | although I'm not coming | 21:31 |
| notmyname | "Swift’s object storage showed greater demand in this survey, with 16% of users provisioning more than 1 petabyte of storage, compared to just 4% reporting that amount provisioned one year ago. | 21:33 |
| notmyname | Last year, nearly half of users required more than 10 terabytes of Swift object storage. This year, nearly two-thirds of users require that amount." | 21:33 |
| notmyname | from https://www.openstack.org/assets/survey/April2017SurveyReport.pdf | 21:33 |
| zaitcev | Hrm. Hopefuly it's not just a petabyte of useless VM snapshots | 21:34 |
| notmyname | I'm not here to judge your bytes. just to make sure they are stored safely :-) | 21:34 |
| *** Renich___ is now known as Renich | 21:34 | |
| *** catintheroof has quit IRC | 21:34 | |
| zaitcev | swift upload imas '[Leopard-Raws] Idolmaster Cinderella Girls Gekijou - 01 RAW (BS11 1280x720 x264 AAC).mp4' | 21:36 |
| onovy | for us it's petabyte of useless video and image | 21:38 |
| onovy | better than VM snapshosts? | 21:39 |
| notmyname | it's just a zero or one. (or 2**50 of them) | 21:41 |
| pdardeau_ | 5 years from now someone might look back and say this was the genesis of carbon credits by container (or storage policy) | 21:45 |
| clayg | notmyname: onovy: one time I think I wrote a diff to fix those damn tests - my comment on patch set 3 - https://review.openstack.org/#/c/423783/ | 21:45 |
| patchbot | patch 423783 - swift - IO Priority support on the AArch64 architecture (MERGED) | 21:45 |
| clayg | the point is the bytes are useless - eat arbys | 21:46 |
| onovy | clayg: right | 21:48 |
| onovy | working too | 21:48 |
| onovy | btw: we really want Swift to work on ARM | 21:49 |
| onovy | because we are building our hardware for storage | 21:49 |
| notmyname | onovy: that's really cool | 21:51 |
| onovy | thanks for fixing endianity of ring :) | 21:51 |
| onovy | but to be honest, we are just experimenting with it now. we have standard amd64 servers today | 21:52 |
| clarkb | I know the linaro folks had openstack in general running on arm | 21:55 |
| clarkb | they may have useful infos | 21:55 |
| onovy | yep, guys which are working on this project knows it | 21:56 |
| onovy | but thanks :) | 21:56 |
| zaitcev | i'd love to see an ARM server board with a real I/O. Everyone I found thus far always hooks SATA off a USB-to-SATA adapter, and it's awfully slow. | 21:58 |
| onovy | aa, we will be better :) | 21:59 |
| onovy | no USB-to-SATA adapter | 21:59 |
| onovy | but SATA SSD disk | 21:59 |
| clarkb | zaitcev: system76 starling pro is a neat looking arm system. But seems more compute density focused. But appears to have proper IO too | 22:00 |
| clarkb | pcie even | 22:00 |
| *** Sukhdev has joined #openstack-swift | 22:06 | |
| *** tonanhngo has joined #openstack-swift | 22:08 | |
| *** tonanhngo has quit IRC | 22:10 | |
| *** klamath has quit IRC | 22:10 | |
| mattoliverau | Btw sorry if I'm a bit distracted this week, will be trying to get back to some community stuff today. | 22:17 |
| mattoliverau | While I still have an army of cloud saio's | 22:17 |
| notmyname | definitely understandable | 22:17 |
| notmyname | mattoliverau: assuming swift-related job offers don't come raining down next week, one of the most helpful things you can do is make sure all your data/code/ideas are written down someplace we can read them | 22:19 |
| mattoliverau | Yeah will do, I'll make sure the sharding Trello is updated, and wrangle the sharding patches into shape | 22:20 |
| mattoliverau | As well ^ | 22:21 |
| *** jamielennox is now known as jamielennox|away | 22:25 | |
| timburke | mattoliverau: fwiw, i'ma try to come back around to that early next week, take on the heart of it in container/sharder | 22:26 |
| *** jamielennox|away is now known as jamielennox | 22:29 | |
| mattoliverau | timburke: next week will be my last week at Rackspace, so how about we use that time to make sure you up to speed. Tho I'm not going to just disappear, unless the I get a job that doesn't let me work on Swift. But I'd make sure it isn't sudden. | 22:34 |
| timburke | mattoliverau: sounds like a plan | 22:36 |
| *** peluse has joined #openstack-swift | 22:46 | |
| peluse | what up people!? | 22:46 |
| notmyname | peluse: !!!!! | 22:47 |
| peluse | I'm a little amazed my IRC client still works :) | 22:47 |
| notmyname | :-) | 22:47 |
| notmyname | peluse: you're here to tell us you're back, right? ;-) | 22:47 |
| peluse | LOL, I wish! I miss the Swift community for sure! I was seriously just checking to see if this thing stil worked and wanted to say Hi to all you guys | 22:48 |
| peluse | notmyname: did U see my email yet? No hurry but maybe by Mon whatever you got would be great! | 22:49 |
| notmyname | peluse: yes I did | 22:49 |
| peluse | rock n roll | 22:49 |
| notmyname | peluse: actually, interesting timing. I'll be working on something new for the boston summit | 22:49 |
| notmyname | but likely won't be done by monday | 22:50 |
| peluse | where's everyone else? off drinking maybe? | 22:50 |
| peluse | notmyname: ahh, OK. I'll take anything in anyform. Just informal use for my internal team here, that's all | 22:50 |
| clayg | peluse: good to hear from you my friend :) | 22:50 |
| peluse | I'm having the discussion Tue afternoon | 22:50 |
| peluse | clayg!!! | 22:50 |
| notmyname | actually, it's been a little more quiet lately | 22:50 |
| notmyname | good to have a monday deadline. i'll get a response by then | 22:51 |
| peluse | excellent... | 22:51 |
| peluse | clayg: they keeping *you* busy I hope :) | 22:51 |
| clayg | i just sorta piddle with this and that - you know me | 22:51 |
| peluse | rock n roll | 22:51 |
| peluse | allrighty... tell the rest of the gang I said "what up?" and next time I'm in the bay area I'll swing by for sure!! | 22:52 |
| clayg | have you invented any new technologies at Intel that are going to revolutionalize storage ... or are you taking this week off? | 22:52 |
| notmyname | lol | 22:52 |
| clayg | peluse: you're always welcome here | 22:52 |
| peluse | clayg: haha! Yes of course!! http://spdk.io | 22:53 |
| clayg | peluse: rock n roll ;) | 22:53 |
| peluse | but I didn't invent any of that... its just what I'm about to start working on. I took some vacation and sabbatical combined so last Thu was my first day back to work since the new year started if you can believe it!! | 22:53 |
| openstackgerrit | Tim Burke proposed openstack/swift master: Add support to increase object ring partition power https://review.openstack.org/337297 | 22:53 |
| notmyname | nice | 22:53 |
| peluse | in fact, its getting to be quitting time here soon. time to crack a cold one :) | 22:54 |
| openstackgerrit | Tim Burke proposed openstack/swift master: Improve test_get_valid_utf8_str coverage https://review.openstack.org/457776 | 22:55 |
| clayg | peluse: I'll send you a cheers when I get off today! thanks for stopping by | 22:55 |
| peluse | clayg: sound about right... take care man! | 22:56 |
| *** peluse has quit IRC | 23:01 | |
| *** vinsh has quit IRC | 23:07 | |
| *** Sukhdev has quit IRC | 23:08 | |
| mattoliverau | Sigh, I step away, and when I get back I miss peluse :( | 23:15 |
| clayg | awwww | 23:26 |
| *** kei_yama has joined #openstack-swift | 23:33 | |
| mattoliverau | sounds like we need to find a way to integrate SPDK and then we may get peluse back ;) | 23:35 |
| jrichli | zaitcev: sorry you wont be at Boston (I wont either actually), and mattoliverau: sorry again to hear about the job, and peluse: sorry I missed you! | 23:36 |
| pdardeau_ | mattoliverau: the new version of SPDK does open up some possibilities. there's now a blobstore api and even a lightweight filesystem (non-posix) | 23:38 |
| *** jerrygb has quit IRC | 23:45 | |
| zaitcev | I keep all my patches in depths of Jenkins just in case, although the amount and complexity are notwhere near container sharding, of course. | 23:55 |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!
