| *** ccamel has joined #openstack-swift | 00:02 | |
| *** camelCaser has quit IRC | 00:04 | |
| *** diablo_rojo has joined #openstack-swift | 00:15 | |
| *** ccamacho has quit IRC | 00:51 | |
| *** nanzha has joined #openstack-swift | 01:22 | |
| *** tomha has joined #openstack-swift | 01:49 | |
| *** tomha has quit IRC | 01:54 | |
| *** Guest14611 has quit IRC | 03:50 | |
| *** benj has joined #openstack-swift | 03:51 | |
| *** benj is now known as Guest85345 | 03:51 | |
| *** diablo_rojo has quit IRC | 06:11 | |
| *** diablo_rojo has joined #openstack-swift | 06:26 | |
| *** tkajinam has quit IRC | 06:50 | |
| *** tkajinam has joined #openstack-swift | 06:51 | |
| *** diablo_rojo has quit IRC | 07:03 | |
| *** nanzha has quit IRC | 07:04 | |
| *** nanzha has joined #openstack-swift | 07:08 | |
| *** rcernin has quit IRC | 07:28 | |
| *** ccamacho has joined #openstack-swift | 07:33 | |
| *** pcaruana has joined #openstack-swift | 07:37 | |
| *** tesseract has joined #openstack-swift | 07:49 | |
| *** tkajinam has quit IRC | 07:54 | |
| openstackgerrit | Thiago da Silva proposed openstack/swift master: Enable s3api and staticweb tests across all func tests https://review.opendev.org/691747 | 08:25 |
|---|---|---|
| *** rdejoux has joined #openstack-swift | 08:33 | |
| *** mikecmpbll has joined #openstack-swift | 08:48 | |
| openstackgerrit | Gilles Biannic proposed openstack/swift master: Stop retrying every deletes in container-sync https://review.opendev.org/691729 | 08:50 |
| *** rpittau|afk is now known as rpittau | 09:08 | |
| *** tkajinam has joined #openstack-swift | 09:15 | |
| *** tkajinam has quit IRC | 09:16 | |
| *** pcaruana has quit IRC | 10:03 | |
| *** pcaruana has joined #openstack-swift | 10:24 | |
| *** ianychoi__ is now known as ianychoi | 11:24 | |
| openstackgerrit | Thiago da Silva proposed openstack/swift master: Enable s3api and staticweb tests across all func tests https://review.opendev.org/691747 | 12:38 |
| *** pcaruana has quit IRC | 12:45 | |
| *** pcaruana has joined #openstack-swift | 12:58 | |
| openstackgerrit | Thiago da Silva proposed openstack/swift master: WIP: Allow internal clients to use reserved namespace https://review.opendev.org/682138 | 13:18 |
| clayg | tdasilva: that test should probably patch policies 😬 | 13:36 |
| tdasilva | oh, good point, let me fix it up | 13:37 |
| openstackgerrit | Thiago da Silva proposed openstack/swift master: WIP: Allow internal clients to use reserved namespace https://review.opendev.org/682138 | 13:44 |
| tdasilva | clayg: ^^^ let me know if you were thinking something different | 13:44 |
| *** pcaruana has quit IRC | 13:52 | |
| *** pcaruana has joined #openstack-swift | 13:55 | |
| clayg | no, that looks perfect! | 14:02 |
| clayg | 😍 | 14:02 |
| *** diablo_rojo has joined #openstack-swift | 14:02 | |
| *** renich_ has joined #openstack-swift | 14:09 | |
| *** renich has quit IRC | 14:09 | |
| *** gyee has joined #openstack-swift | 14:57 | |
| openstackgerrit | Clay Gerrard proposed openstack/python-swiftclient master: WIP: object versioning features https://review.opendev.org/691877 | 14:59 |
| clayg | ^ I'm acctually still figuring out the API - so it'd be great to have some help with this fella | 15:00 |
| clayg | tdasilva: do you have an etherpad going on object versioning? | 15:09 |
| tdasilva | clayg: https://etherpad.openstack.org/p/swift-object-versioning | 15:10 |
| tdasilva | clayg: the proposed api has probably changed a bit since I first wrote it, so i need to update it | 15:18 |
| clayg | no worries, if I spot something that seems off I'll just update - thanks | 15:18 |
| *** renich has joined #openstack-swift | 15:40 | |
| *** renich_ has quit IRC | 15:41 | |
| *** renich has quit IRC | 15:49 | |
| *** nanzha has quit IRC | 15:58 | |
| *** nanzha has joined #openstack-swift | 15:59 | |
| *** nanzha has quit IRC | 16:31 | |
| *** tesseract has quit IRC | 16:38 | |
| *** nanzha has joined #openstack-swift | 16:59 | |
| *** nanzha has quit IRC | 17:05 | |
| *** rpittau is now known as rpittau|afk | 17:25 | |
| *** gmann is now known as gmann_afk | 17:29 | |
| *** gmann_afk is now known as gmann | 17:35 | |
| timburke | \o/ there's gonna be a cschwede! https://etherpad.openstack.org/p/swift-ptg-shanghai | 17:39 |
| *** rdejoux has quit IRC | 17:42 | |
| *** mikecmpbll has quit IRC | 17:47 | |
| clayg | 🙌 | 18:19 |
| binwiederhier | timburke: You said a while ago that people have thought about getting rid of the 3-syncs-per-PUT, and I have been thinking about the same thing. With 3 replicas in 3 different failure domains (potentially even physical locations), the chances of losing all writes is pretty slim, don't you agree? Have you thought about making the sync calls optional or otherwise configurable? | 18:48 |
| zaitcev | You can't decide which one to sync before you know which one succeeded. So, I suppose, the best we can do is to mandate a form of PUT+POST even for the replicated policies. Then, proxy first lays down non-durable copies, then issues POSTs until a quorum is accomplished. | 18:52 |
| binwiederhier | I think you are talking about still ensuring durable writes, while I'm talking about accepting the risk of 3 failures and just straight up not syncing (i.e. let xfs do the sync) | 18:55 |
| DHE | even if one machine is up in smoke between OK'ing the object and its death because sync writes were disabled, you still have 2 other copies that are far enough away that any failure would be contained | 18:56 |
| DHE | while I like the idea conceptually, I think I would want to make it a per-storage-policy option which is... uhh... | 18:56 |
| binwiederhier | Yeah I was thinking of a object-node wide setting `sync_behavior = nosync|always|every-now-and-then`, but per storage policy is probably even better | 18:57 |
| zaitcev | Meh, that's shallow thinking. What you really need are John Corbin's RAM arrays. There's no way to lose power on 3 continents simultaneously, right? | 18:59 |
| DHE | I have plans that involve some storage policies spanning multiple locations, but also some that are going to be in a much smaller footprint. hence my preference that it be a storage policy option. | 19:02 |
| *** Guest85345 is now known as benj_ | 19:08 | |
| binwiederhier | Why does it have to be this hard though? | 19:08 |
| binwiederhier | Maybe I'm not quite following when you say "spanning multiple locations"; how is that related to the sync behavior? | 19:09 |
| DHE | <binwiederhier> timburke: You said a while ago that people have thought about getting rid of the 3-syncs-per-PUT, and I have been thinking about the same thing. With 3 replicas in 3 different failure domains (potentially even physical locations) # this bit here at the end | 19:09 |
| binwiederhier | ok I see, so you'd like to make the sync behavior dependend on the failure domains. | 19:10 |
| binwiederhier | Interesting. | 19:11 |
| DHE | there's some situations where the failure domains are sufficiently apart that it would be worth doing, but other storage policies not so much... | 19:11 |
| binwiederhier | It would be significantly easier to leave that up to the administrator though; leave the default at `sync = always`, but give the option to change the behavior | 19:12 |
| DHE | yes, but put the option inside the storage policy section rather than making it global. that's my suggestion | 19:13 |
| binwiederhier | :+1: | 19:14 |
| binwiederhier | Ship it. | 19:14 |
| clayg | zaitcev: I have to assume on p 691729 the assumption is that 409 typically means the remote side has a newer timestamp | 19:25 |
| patchbot | https://review.opendev.org/#/c/691729/ - swift - Stop retrying every deletes in container-sync - 2 patch sets | 19:25 |
| zaitcev | So, fortunately we're unable to delete it because someone else wrote it, right? | 19:26 |
| zaitcev | Meaning, our deletion listing is out of date. Okay. | 19:26 |
| gmann | timburke: clayg : sent the requirement of swift on py2 on ML - http://lists.openstack.org/pipermail/openstack-discuss/2019-October/010430.html | 19:54 |
| gmann | feel free to explain or add more if needed | 19:54 |
| clayg | 🤗 | 19:55 |
| openstackgerrit | Tim Burke proposed openstack/swift master: exclude utf8 tests under py3 https://review.opendev.org/689225 | 20:12 |
| *** renich has joined #openstack-swift | 20:27 | |
| *** renich has quit IRC | 20:35 | |
| *** renich has joined #openstack-swift | 21:12 | |
| *** pcaruana has quit IRC | 21:15 | |
| openstackgerrit | Thiago da Silva proposed openstack/swift master: WIP: New Object Versioning mode https://review.opendev.org/682382 | 21:27 |
| openstackgerrit | Thiago da Silva proposed openstack/swift master: Enable s3api and staticweb tests across all func tests https://review.opendev.org/691747 | 21:43 |
| *** renich has quit IRC | 21:59 | |
| *** rcernin has joined #openstack-swift | 22:02 | |
| *** gyee has quit IRC | 22:27 | |
| *** gyee has joined #openstack-swift | 22:52 | |
| *** tkajinam has joined #openstack-swift | 23:08 | |
| *** diablo_rojo has quit IRC | 23:10 | |
| *** diablo_rojo has joined #openstack-swift | 23:11 | |
| *** tkajinam has quit IRC | 23:48 | |
| *** gyee has quit IRC | 23:54 | |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!