| *** jamielennox is now known as jamielennox|away | 00:30 | |
| *** jamielennox|away is now known as jamielennox | 00:48 | |
| kota_ | good morning | 01:03 |
|---|---|---|
| * kota_ is back from long weekend due to Japanese holiday. | 01:04 | |
| mattoliverau | kota_: morning, and welcome back :) | 01:13 |
| kota_ | mattolivaeru: o/ | 01:14 |
| *** klrmn has quit IRC | 01:15 | |
| *** takashi has joined #openstack-swift | 01:18 | |
| *** Jeffrey4l has joined #openstack-swift | 01:21 | |
| *** Jeffrey4l has quit IRC | 01:22 | |
| *** Jeffrey4l has joined #openstack-swift | 01:23 | |
| *** Jeffrey4l has quit IRC | 01:24 | |
| *** Jeffrey4l has joined #openstack-swift | 01:25 | |
| *** Jeffrey4l has quit IRC | 01:26 | |
| *** Jeffrey4l has joined #openstack-swift | 01:26 | |
| *** Jeffrey4l_ has joined #openstack-swift | 01:54 | |
| *** Jeffrey4l has quit IRC | 01:57 | |
| *** tqtran has quit IRC | 02:01 | |
| *** yarkot has joined #openstack-swift | 02:03 | |
| *** takashi has quit IRC | 02:06 | |
| *** jcook has quit IRC | 02:08 | |
| *** jcook has joined #openstack-swift | 02:13 | |
| openstackgerrit | Hanxi Liu proposed openstack/swift: Add links to more detailed overview in overview_architecture https://review.openstack.org/381446 | 02:20 |
| *** klrmn has joined #openstack-swift | 02:30 | |
| -openstackstatus- NOTICE: Jobs running on osic nodes are failing due to network issues with the mirror. We are temporarily disabling the cloud. | 02:49 | |
| *** lhx_ has joined #openstack-swift | 03:13 | |
| openstackgerrit | Hanxi Liu proposed openstack/swift: Add links to more detailed overview in overview_architecture https://review.openstack.org/381446 | 03:23 |
| *** psachin has joined #openstack-swift | 03:40 | |
| *** vern has quit IRC | 03:57 | |
| clayg | timburke: Do you know Jamie Lennox? Is he helping us with keystone sessions work? Do you have any experience with requests-mock? Could you splash a little water on patch 360298 - i'm worried I may have un-intentionally been a little thorny | 04:00 |
| patchbot | https://review.openstack.org/#/c/360298/ - python-swiftclient - Use requests_mock and oslotest | 04:00 |
| *** links has joined #openstack-swift | 04:01 | |
| *** vern has joined #openstack-swift | 04:17 | |
| *** vern has quit IRC | 04:24 | |
| openstackgerrit | Merged openstack/swift: add assert error information https://review.openstack.org/379428 | 04:45 |
| *** tqtran has joined #openstack-swift | 04:57 | |
| *** tqtran has quit IRC | 05:03 | |
| *** lhx_ has quit IRC | 05:05 | |
| *** lhx_ has joined #openstack-swift | 05:11 | |
| *** d0ugal has quit IRC | 05:40 | |
| *** d0ugal has joined #openstack-swift | 05:41 | |
| *** takashi has joined #openstack-swift | 05:51 | |
| *** qwertyco has joined #openstack-swift | 06:00 | |
| *** qwertyco has quit IRC | 06:01 | |
| *** qwertyco has joined #openstack-swift | 06:02 | |
| *** lhx_ has quit IRC | 06:06 | |
| *** rcernin has joined #openstack-swift | 06:07 | |
| *** vern has joined #openstack-swift | 06:17 | |
| *** klrmn has quit IRC | 06:18 | |
| *** bjolo_ has joined #openstack-swift | 06:21 | |
| *** lhx_ has joined #openstack-swift | 06:26 | |
| *** vern has quit IRC | 06:26 | |
| *** pcaruana has joined #openstack-swift | 06:36 | |
| *** vern has joined #openstack-swift | 06:52 | |
| *** Vinsh has quit IRC | 07:02 | |
| *** Vinsh has joined #openstack-swift | 07:03 | |
| *** rledisez has joined #openstack-swift | 07:28 | |
| *** takashi has quit IRC | 07:38 | |
| *** amoralej|off is now known as amoralej | 07:39 | |
| *** qwertyco has quit IRC | 07:39 | |
| *** takashi has joined #openstack-swift | 07:40 | |
| *** hogepodge has quit IRC | 07:51 | |
| *** Vinsh has quit IRC | 07:52 | |
| *** acoles_ is now known as acoles | 07:54 | |
| acoles | goo dmorning | 07:56 |
| *** geaaru has joined #openstack-swift | 07:57 | |
| *** hogepodge has joined #openstack-swift | 07:58 | |
| *** zaitcev has quit IRC | 07:58 | |
| *** acoles is now known as acoles_ | 08:00 | |
| *** zaitcev has joined #openstack-swift | 08:16 | |
| *** ChanServ sets mode: +v zaitcev | 08:16 | |
| kota_ | good morning acoles | 08:20 |
| kota_ | acoles: thanks for pinging me sometimes in the back log. I was in long weekend (and actually down wiht cold). I'm trying to open it step by step. | 08:22 |
| *** jordanP has joined #openstack-swift | 08:22 | |
| *** admin6_ has joined #openstack-swift | 08:27 | |
| *** natarej has joined #openstack-swift | 08:31 | |
| *** acoles_ is now known as acoles | 08:35 | |
| *** jeblair_ has joined #openstack-swift | 08:36 | |
| *** mahatic_ has joined #openstack-swift | 08:36 | |
| *** clayg_ has joined #openstack-swift | 08:36 | |
| *** ChanServ sets mode: +v clayg_ | 08:36 | |
| *** dosaboy_ has joined #openstack-swift | 08:37 | |
| *** early has joined #openstack-swift | 08:38 | |
| *** mmotiani has joined #openstack-swift | 08:38 | |
| *** swifterdarrell_ has joined #openstack-swift | 08:39 | |
| *** amoralej_ has joined #openstack-swift | 08:40 | |
| *** jraju has joined #openstack-swift | 08:40 | |
| *** vern has quit IRC | 08:41 | |
| *** links has quit IRC | 08:41 | |
| *** yarkot has quit IRC | 08:41 | |
| *** hoonetorg has quit IRC | 08:41 | |
| *** amoralej has quit IRC | 08:41 | |
| *** peterlisak has quit IRC | 08:41 | |
| *** clayg has quit IRC | 08:41 | |
| *** early` has quit IRC | 08:41 | |
| *** zackmdavis has quit IRC | 08:41 | |
| *** amit213 has quit IRC | 08:41 | |
| *** ogelbukh has quit IRC | 08:41 | |
| *** ediardo has quit IRC | 08:41 | |
| *** nottrobin has quit IRC | 08:41 | |
| *** wasmum has quit IRC | 08:41 | |
| *** fungi has quit IRC | 08:41 | |
| *** zacksh has quit IRC | 08:41 | |
| *** sudorandom has quit IRC | 08:41 | |
| *** charz has quit IRC | 08:41 | |
| *** swifterdarrell has quit IRC | 08:41 | |
| *** a1|away has quit IRC | 08:41 | |
| *** mahatic has quit IRC | 08:41 | |
| *** jeblair has quit IRC | 08:41 | |
| *** dosaboy has quit IRC | 08:41 | |
| *** onovy has quit IRC | 08:41 | |
| *** swifterdarrell_ is now known as swifterdarrell | 08:41 | |
| *** amoralej_ is now known as amoralej | 08:41 | |
| *** charz has joined #openstack-swift | 08:41 | |
| *** zackmdavis has joined #openstack-swift | 08:42 | |
| *** yarkot has joined #openstack-swift | 08:42 | |
| *** a1|away has joined #openstack-swift | 08:44 | |
| *** hoonetorg has joined #openstack-swift | 08:45 | |
| *** vern has joined #openstack-swift | 08:49 | |
| *** peterlisak has joined #openstack-swift | 08:49 | |
| *** amit213 has joined #openstack-swift | 08:49 | |
| *** ogelbukh has joined #openstack-swift | 08:49 | |
| *** ediardo has joined #openstack-swift | 08:49 | |
| *** nottrobin has joined #openstack-swift | 08:49 | |
| *** wasmum has joined #openstack-swift | 08:49 | |
| *** fungi has joined #openstack-swift | 08:49 | |
| *** zacksh has joined #openstack-swift | 08:49 | |
| *** sudorandom has joined #openstack-swift | 08:49 | |
| *** onovy has joined #openstack-swift | 08:49 | |
| *** ogelbukh has quit IRC | 08:52 | |
| *** ogelbukh has joined #openstack-swift | 08:56 | |
| *** takashi has quit IRC | 09:01 | |
| *** admin6 has quit IRC | 09:08 | |
| *** admin6_ is now known as admin6 | 09:08 | |
| *** admin6_ has joined #openstack-swift | 09:13 | |
| *** admin6_ has left #openstack-swift | 09:14 | |
| *** admin6 has quit IRC | 09:15 | |
| *** admin6_ has joined #openstack-swift | 09:15 | |
| *** vern has quit IRC | 09:24 | |
| *** tanee_ is now known as tanee | 09:26 | |
| *** openstack has joined #openstack-swift | 09:31 | |
| *** onovy has joined #openstack-swift | 09:38 | |
| *** peterlisak has joined #openstack-swift | 09:38 | |
| *** VonGoofy has joined #openstack-swift | 09:54 | |
| VonGoofy | Hello | 09:54 |
| VonGoofy | Problem | 09:54 |
| VonGoofy | I have a problem with swift. I have added x-object-account metadata to my swift account, deleted it because it had space in it, so it was invalid, and now I can't add the metadata to account anymore. Doing this via object storage API. | 09:54 |
| VonGoofy | I get 204 response on my metadata POST request, issue show metadata, metadata is not there | 09:55 |
| VonGoofy | help? anyone? | 10:00 |
| *** VonGoofy has quit IRC | 10:02 | |
| *** kodokuu has joined #openstack-swift | 10:03 | |
| *** Vinsh has joined #openstack-swift | 10:04 | |
| kodokuu | Hi, anyone have a script for GET all DATA in swift ?? I create a new platform and we need to transfer all data to the new swift . | 10:04 |
| *** VonGoofy has joined #openstack-swift | 10:09 | |
| VonGoofy | if anyone knows why I cant inject metadata definitions to the account that would be helpful. I'm trying to add X-Account-Temp-URL-Key | 10:10 |
| *** jraju has quit IRC | 10:15 | |
| *** tanee is now known as tanee_away | 10:17 | |
| *** bjolo_ has quit IRC | 10:18 | |
| *** tqtran has joined #openstack-swift | 10:20 | |
| *** tqtran has quit IRC | 10:24 | |
| *** mvk has quit IRC | 10:30 | |
| VonGoofy | Anyone? | 10:32 |
| *** dfflanders has quit IRC | 10:37 | |
| acoles | VonGoofy: temp url key metadata starts with X-Account-Meta- see http://docs.openstack.org/developer/swift/api/temporary_url_middleware.html#secret-keys | 10:44 |
| VonGoofy | yes I know | 10:50 |
| VonGoofy | I cannot set this X-Account-Meta-Temp-URL-Key, nor this X-Container-Meta-Temp-URL-Key | 10:50 |
| VonGoofy | I had a typo in my previous post | 10:51 |
| VonGoofy | Any other metadata definition I can set | 10:51 |
| *** mvk has joined #openstack-swift | 11:01 | |
| *** links has joined #openstack-swift | 11:09 | |
| *** qwertyco has joined #openstack-swift | 11:12 | |
| *** thebloggu has joined #openstack-swift | 11:15 | |
| *** qwertyco has quit IRC | 11:15 | |
| *** qwertyco has joined #openstack-swift | 11:16 | |
| *** etienneme has quit IRC | 11:32 | |
| *** jamielennox is now known as jamielennox|away | 11:36 | |
| *** nikivi has joined #openstack-swift | 11:43 | |
| acoles | VonGoofy: sorry, been away. so using the swiftclient, I can do 'swift post -H "x-account-meta-tempurl-key:foo"' and then 'swift stat' will show me the Tempurl-Key | 11:45 |
| VonGoofy | I got it why it's not showing through api. Because those are secret metadata keys and won't show for other users | 11:47 |
| VonGoofy | only to owners of swift and if it's set in the prox-server.conf | 11:47 |
| *** openstackgerrit has quit IRC | 11:48 | |
| *** openstackgerrit has joined #openstack-swift | 11:49 | |
| VonGoofy | acoles: Now I'm fighting with a post form and how to upload files from remote website | 11:50 |
| openstackgerrit | OpenStack Proposal Bot proposed openstack/swift: Updated from global requirements https://review.openstack.org/88736 | 12:07 |
| *** thebloggu has quit IRC | 12:16 | |
| *** openstackgerrit has quit IRC | 12:18 | |
| *** openstackgerrit has joined #openstack-swift | 12:18 | |
| *** amoralej is now known as amoralej|lunch | 12:23 | |
| *** admin6_ has quit IRC | 12:25 | |
| *** admin6 has joined #openstack-swift | 12:25 | |
| *** bjolo_ has joined #openstack-swift | 12:27 | |
| *** qwertyco has quit IRC | 12:50 | |
| VonGoofy | Is there any example on how to upload files through browser remotely? using signed forms not username/password/token. Form post doesn't work for me for some reason. It only changes metadata | 12:51 |
| *** kei_yama has quit IRC | 12:52 | |
| *** lhx_ has quit IRC | 13:11 | |
| *** kodokuu has quit IRC | 13:17 | |
| *** StraubTW has joined #openstack-swift | 13:24 | |
| *** amoralej|lunch is now known as amoralej | 13:33 | |
| *** lhx_ has joined #openstack-swift | 13:36 | |
| *** VonGoofy has quit IRC | 13:42 | |
| *** StraubTW_ has joined #openstack-swift | 13:46 | |
| *** StraubTW has quit IRC | 13:50 | |
| *** rcernin has quit IRC | 14:07 | |
| acoles | tdasilva: hi. re your comment here https://review.openstack.org/#/c/355958/5/test/probe/test_reconstructor_durable.py, I had same thought - I think the test isn't doing what the comment suggests. But I thought it could be fixed separately to keep this patch simpler. what do you think? | 14:23 |
| patchbot | patch 355958 - swift - EC - eliminate .durable files | 14:23 |
| *** bjolo_ has quit IRC | 14:28 | |
| *** sgundur has joined #openstack-swift | 14:34 | |
| *** _JZ_ has joined #openstack-swift | 14:54 | |
| *** klrmn has joined #openstack-swift | 15:09 | |
| *** psachin has quit IRC | 15:17 | |
| *** sgundur has quit IRC | 15:19 | |
| *** jistr is now known as jistr|biab | 15:25 | |
| *** sgundur has joined #openstack-swift | 15:30 | |
| *** pcaruana has quit IRC | 15:31 | |
| tdasilva | acoles: agree, I can submit that as a follow-up patch, what do you think? | 15:37 |
| tdasilva | acoles: I was also going to take a look at the other probe tests to see if actual data reconstruction is being tested the same way | 15:38 |
| tdasilva | acoles: looks like it is | 15:39 |
| tdasilva | acoles: https://github.com/openstack/swift/blob/master/test/probe/test_reconstructor_rebuild.py#L92 | 15:39 |
| acoles | tdasilva: that would be great since you have already started a patch (your paste) | 15:46 |
| acoles | tdasilva: that test ^^ you link to looks ok, I think? | 15:47 |
| acoles | tdasilva: although it does look like there is some repeated code wrt the one I fixed | 15:47 |
| acoles | tdasilva: yesterday I searched for all occurrences of '.durable' in the code paste and stepped through them all to be sure they were appropriate (basically, legacy testing) | 15:48 |
| tdasilva | acoles: sorry, the link above is not really about the .durable issue, it's more about the issue I brought up about how the test is written | 15:49 |
| *** sgundur has quit IRC | 15:49 | |
| tdasilva | acoles: it's following the same pattern as the .durable tests, but for the actual data file | 15:49 |
| *** geaaru has quit IRC | 15:50 | |
| tdasilva | you can see in lines 155-157, it builds the lists of nodes to delete data from, but then it deletes and reconstructs each node individually | 15:50 |
| *** nikivi has quit IRC | 15:50 | |
| *** sgundur has joined #openstack-swift | 15:51 | |
| acoles | tdasilva: oic sorry, yes that issue, how it just iterates over the list of lists but only ever breaks one node at a time. | 15:52 |
| tdasilva | acoles: right | 15:53 |
| tdasilva | acoles: so in the new patch i will fix both and look for more | 15:53 |
| acoles | tdasilva: feels like there could be a common pattern between the two tests, break the *list* of nodes, run reconstructor, validate, where definition of 'break' differs between tests | 15:53 |
| acoles | tda great, thanks! | 15:54 |
| acoles | tdasilva: ^^ | 15:54 |
| *** links has quit IRC | 15:54 | |
| tdasilva | acoles: right :) | 15:54 |
| *** chsc has joined #openstack-swift | 15:58 | |
| *** chsc has joined #openstack-swift | 15:58 | |
| *** donagh has quit IRC | 16:03 | |
| *** jistr|biab is now known as jistr | 16:07 | |
| *** chsc has quit IRC | 16:12 | |
| openstackgerrit | Cheng Li proposed openstack/swift: add check of min_part_hours https://review.openstack.org/345524 | 16:14 |
| *** acoles is now known as acoles_ | 16:15 | |
| *** acoles_ is now known as acoles | 16:17 | |
| *** acoles is now known as acoles_ | 16:17 | |
| *** acoles_ is now known as acoles | 16:18 | |
| *** sgundur has quit IRC | 16:24 | |
| briancline | what's the history behind the object controller's x-container-{partition,host,device} headers being more authoritative than an object server doing an in-memory lookup of the same info? | 16:24 |
| *** sgundur has joined #openstack-swift | 16:24 | |
| briancline | traced it back to the initial commit | 16:25 |
| *** tongli has joined #openstack-swift | 16:27 | |
| *** hoonetorg has quit IRC | 16:35 | |
| *** lhx_ has quit IRC | 16:37 | |
| *** rledisez has quit IRC | 16:47 | |
| *** hoonetorg has joined #openstack-swift | 16:48 | |
| *** sgundur has quit IRC | 16:48 | |
| *** sgundur has joined #openstack-swift | 16:58 | |
| notmyname | good morning | 17:00 |
| *** chsc has joined #openstack-swift | 17:01 | |
| tdasilva | acoles: just realized you already wrote some of the tests I was thinking about here: https://review.openstack.org/#/c/376630/ | 17:03 |
| patchbot | patch 376630 - swift - EC: reconstruct using non-durable fragments | 17:03 |
| tdasilva | acoles: nice!!!! | 17:03 |
| *** tongli has quit IRC | 17:04 | |
| *** tqtran has joined #openstack-swift | 17:11 | |
| acoles | tdasilva: oh yeah, I had forgotten that! so break_nodes there is a multipurpose tool to set up some broken state across the nodes. | 17:27 |
| acoles | huh. | 17:27 |
| tdasilva | acoles: yep! we could probably even get rid of the check_nodes method, it seems to be trying to do similar work | 17:28 |
| *** sgundur has quit IRC | 17:36 | |
| *** sgundur has joined #openstack-swift | 17:37 | |
| *** vint_bra has joined #openstack-swift | 17:42 | |
| *** vint_bra has quit IRC | 17:42 | |
| *** vint_bra has joined #openstack-swift | 17:44 | |
| *** amoralej is now known as amoralej|off | 17:59 | |
| *** jordanP has quit IRC | 17:59 | |
| timur | zaitcev: no, that's a different timur | 18:04 |
| zaitcev | timur: sorry | 18:04 |
| timur | no worries | 18:04 |
| *** mvk has quit IRC | 18:04 | |
| *** jamielennox|away is now known as jamielennox | 18:10 | |
| *** bjolo_ has joined #openstack-swift | 18:11 | |
| *** sgundur has quit IRC | 18:12 | |
| clayg_ | tdasilva: is https://github.com/dtantsur/ironic-bug-dashboard the dashboard thing you were adapting to run on swit? | 18:19 |
| *** clayg_ is now known as clayg | 18:19 | |
| clayg | did you see the ironic subteam update - it had some bug stats - week deltas - pretty interesting! | 18:19 |
| tdasilva | clayg: no, i forked off a project from the fuel team: https://github.com/thiagodasilva/launchpad-reporting | 18:20 |
| tdasilva | but will take a look | 18:20 |
| tdasilva | clayg: just found the update email, yeah, they seem to keep a very updated delta | 18:25 |
| tdasilva | clayg: it looks like they have a process for staying on top of (initial) bug triaging: https://bugs.launchpad.net/ironic\ | 18:26 |
| tdasilva | you can see they have very few "Undecided" | 18:26 |
| *** vint_bra has quit IRC | 18:26 | |
| *** vint_bra has joined #openstack-swift | 18:26 | |
| clayg | tdasilva: I like that | 18:29 |
| *** hoonetorg has quit IRC | 18:35 | |
| *** sgundur has joined #openstack-swift | 18:36 | |
| tdasilva | notmyname: there's been a lot of emails on the ML about the PTG (e.g., 'Sahara PTG plans') and I was wondering where that is coming from? just wondering if we are supposed to be telling somebody else we are planning to be there... | 18:37 |
| tdasilva | notmyname: if you already took care of that sorry for the nosie | 18:37 |
| notmyname | tdasilva: I've got some notes I'm working on similar to "here's what we've done that works, and we'll do something similar at the PTG" | 18:37 |
| tdasilva | notmyname: cool, it's a bit weird that this conversation is coming up before barca, but it is probably due to the very short release cycle | 18:38 |
| *** sgundur has quit IRC | 18:43 | |
| *** sgundur has joined #openstack-swift | 18:48 | |
| *** hoonetorg has joined #openstack-swift | 18:50 | |
| *** ppalacios has joined #openstack-swift | 18:52 | |
| ppalacios | does anyone know if it is possible to change etag header on swift object operations to use sha256 instead of md5? | 18:54 |
| clayg | i know that is is not | 18:54 |
| notmyname | no. but you can add your own hash to your own key name. like x-object-meta-better-etag | 18:54 |
| clayg | notmyname: oh that's a good idea | 18:54 |
| tdasilva | but it won't be "verified" by the swift cluster, like it is for the etag | 18:55 |
| zaitcev | Yeah, that. But you can't know if it's stored correctly. | 18:55 |
| zaitcev | Well, I meant the value that PUT returns | 18:55 |
| zaitcev | ppalacios: So, was there ever a problem with using MD5 or you're just curious? | 18:56 |
| clayg | zaitcev: custom middleware to avlidate x-object-meta-better-etag? | 18:56 |
| zaitcev | clayg: maybe | 18:57 |
| ppalacios | tdasilva: so is there no option to change how swift verifies it? | 18:57 |
| clayg | i honestly would *love* to see a discussion about ehancing the content verification system to support other algorithms besides md5 - specifically i'm interested in crc instead of expensive stuff like md5 and sha - but if other pople have cpu cycles to burn I'm done | 18:57 |
| pdardeau | ^ +1 | 18:58 |
| clayg | ... http doesn't care at all about the content of etag - just that it does a good job at indicating same-ness from one request to the next | 18:58 |
| notmyname | clayg: yeah, md5 is like the worst possible thing. not good enough for any crypto checks and there are so many much faster ones out there for bit flip checks | 18:58 |
| tdasilva | ppalacios: i was saying the same thing as zaitcev said and was thinking that the same solution would be what clayg said ;) | 18:58 |
| clayg | lol | 18:58 |
| clayg | so we're all thinking about and no one is working on it | 18:59 |
| ppalacios | zaitcev: not really, it is just we have a project where we use sha256 to make this verification, so we would like to not waste it. | 18:59 |
| clayg | i've seen apis that say something like 'x-etag-value-algorithm' and there's some set of supported values | 19:00 |
| tdasilva | clayg: any idea why the "industry" settled as md5 and sha as the default? i'm thinking of all the websites that provide downloads and checksums for people to verify | 19:01 |
| notmyname | torgomatic worked on a patch a long time ago that let the algo be selectable | 19:01 |
| openstackgerrit | Merged openstack/swift: updated output for swift stat https://review.openstack.org/371944 | 19:01 |
| clayg | notmyname: I think that was for ring hash | 19:01 |
| notmyname | tdasilva: widely available + decent crypto (at the time) + fast enough | 19:01 |
| clayg | tdasilva: swift used md5 cause we were already using it other places - crc is far less suitable for spreading things out evenly - which is why md5 was choosen and also why it's slower than crc | 19:02 |
| notmyname | also, built into python (no no new dependencies for stuff like murmur) | 19:03 |
| * tdasilva nods | 19:03 | |
| notmyname | also, remember swift came from an existing system which used md5 and was written circa 2008. so here we are | 19:04 |
| tdasilva | ppalacios: so you already have the shas calculated when you are uploading the objects to swift? | 19:04 |
| notmyname | we should have used a Content-Md5 header and then been free to change etag to be whatever we wanted (as long as it was stable) | 19:05 |
| notmyname | that's what 6-8 years of experience gets you, i guess ;-) | 19:05 |
| ppalacios | tdasilva: in fact, we already have an infrastructure that uses sha256. We are migrating this infrastructure to use swift as storage and it will be very strange to calculate MD5 and sha256 to feed swift and this structure. | 19:10 |
| * tdasilva just realized that hummingbird is also pretty hardcoded to use md5 for etag | 19:12 | |
| hurricanerix | notmyname you could add Content-Md5 to the v2 docs, and maybe some day there will be enough stuff there for people to want a v2. | 19:13 |
| hurricanerix | or maybe it is there already, can't remember... | 19:13 |
| openstackgerrit | OpenStack Proposal Bot proposed openstack/swift: Updated from global requirements https://review.openstack.org/88736 | 19:16 |
| clayg | hurricanerix: yeah I don't see much other option beyond v2 | 19:19 |
| *** bjolo_ has quit IRC | 19:22 | |
| hurricanerix | clayg yeah, i still think v2 could be a good thing allowing stuff to be simplified and made more consistent. but i don't think many people have shared that opinion (at least to the extent where they thought it was worth developing it). | 19:22 |
| openstackgerrit | Ondřej Nový proposed openstack/swift: Apply bash error handling consistently in all bash scripts https://review.openstack.org/357025 | 19:42 |
| onovy | clayg: hi, maybe we can continue discussion here? ^^ | 19:43 |
| *** sgundur has quit IRC | 19:45 | |
| onovy | notmyname: hi, can anyone look to https://review.openstack.org/#/c/240726/ pls? ~5 weeks without reply :/ | 19:47 |
| patchbot | patch 240726 - swift - Respect server type for --md5 check in swift-recon | 19:47 |
| *** sgundur has joined #openstack-swift | 19:50 | |
| clayg | onovy: yeah! | 19:53 |
| clayg | onovy: so honestly on resetswift I'm not sure I disagree - thank you for offering your counter point | 19:53 |
| clayg | in truth it won't effect my workflow - i use vsaio and it has it own method for doing a reset -> https://github.com/swiftstack/vagrant-swift-all-in-one/blob/master/bin/resetswift | 19:54 |
| onovy | hmhmm | 19:54 |
| clayg | onovy: my assumption is that `resetswift && resetswift` would exit 0 because the point should be - it reset - if the second one fails because swift is already stopped or something.... | 19:55 |
| onovy | it swift is already stopped, resetswift should return 0 | 19:55 |
| clayg | onovy: cool, i'm not sure if any of the other reset behaviors would have the property of failing if it's not already cleared up | 19:56 |
| clayg | I could also imagine that my data file system got unmounted for some point - and I'm not sure if unmount would exit because... it's already unmounted? | 19:56 |
| onovy | hmm. if swift is stopped, swift-init all stop returns 0 | 19:56 |
| clayg | that sort of thing | 19:56 |
| onovy | sry returns 1 | 19:56 |
| onovy | which is wrong | 19:56 |
| clayg | yeah - i think that's not uncommon - oh rly? | 19:56 |
| onovy | swift-init all kill | 19:57 |
| onovy | kill: stop a server (no error if not running) | 19:57 |
| onovy | so i will change it to "kill" | 19:57 |
| clayg | onovy: what amount unmount? | 19:57 |
| clayg | should it be unmount || true? | 19:57 |
| onovy | good question | 19:57 |
| clayg | the ideal would say unmount if mounted but the point is - make it be not mounted | 19:57 |
| onovy | 32 | 19:57 |
| *** vint_bra has quit IRC | 19:57 | |
| clayg | :\ | 19:58 |
| onovy | so wrong too | 19:58 |
| onovy | so i will change it to kill + unmount if mounted | 19:58 |
| onovy | so unmount of unmounted = return 0 | 19:58 |
| clayg | maybe this is a systemd thing? I would have *swore* i've had the experience with provisioning automation that if you try to call stop and the service is already stopped it exits 1 (not just swift-init; i'm sure it's behavior was modeled after some /etc/init.d script) - same for start when already running | 19:59 |
| clayg | but on my current xenial/systemd build it's all stop stop stop stop = 0 0 0 0 | 20:00 |
| clayg | ^ does that sound right to everyone? | 20:00 |
| onovy | sysv init scripts returns 0, if you try to stop already stopped service | 20:00 |
| clayg | hrm... so i'm misremembering | 20:00 |
| onovy | mmnt i will check | 20:00 |
| ahale | i just remember swift-init having weird unexpected return codes for some stuff, years ago | 20:01 |
| clayg | ahale: and you filed the bug.... where? :P | 20:01 |
| *** vint_bra has joined #openstack-swift | 20:01 | |
| ahale | i was naive and tried to submit a patch, then realised the horrors inside swift-init and ran away | 20:02 |
| clayg | ahale: good call there | 20:02 |
| ahale | https://bugs.launchpad.net/swift/+bug/920520 | 20:02 |
| openstack | Launchpad bug 920520 in OpenStack Object Storage (swift) "swift-init reload/restart inconsistencies" [Undecided,Invalid] | 20:02 |
| ahale | ! | 20:02 |
| * clayg hugs ahale | 20:05 | |
| clayg | what... why's it invalid? | 20:05 |
| onovy | what about: if grep -qs '/mnt/foo' /proc/mounts; then umount | 20:05 |
| ahale | no idea really, maybe something changed | 20:06 |
| clayg | onovy: seems reasonable - i'll try to make a cursory validation that `resetswift && resetswift` works once OMM and call it good enough | 20:08 |
| clayg | because of the way we build saio's it's not very likely to break someone unless they actively recopy the new scirpts into the $PATH | 20:08 |
| *** sgundur has quit IRC | 20:10 | |
| onovy | clayg: in progress | 20:11 |
| clayg | onovy: regarding .functests and .probetests and consistency -> http://redsymbol.net/articles/unofficial-bash-strict-mode/#expect-nonzero-exit-status | 20:12 |
| clayg | the unofficial bash strict mode is a good reminder that a) bash is trying to make your life difficult at every possible turn b) if you're trying to write bash you are either lazy or doing it wrong | 20:13 |
| clayg | ... lazy isn't so bad tho really - it's just lazy - lazy is good sometimes | 20:13 |
| clayg | unless the effort you have to put into defeats the point of being lazy in the first place | 20:13 |
| *** sgundur has joined #openstack-swift | 20:14 | |
| onovy | yep. so that's reason why i'm not using set -e in .functests :] | 20:14 |
| patchbot | Error: Spurious "]". You may want to quote your arguments with double quotes in order to prevent extra brackets from being evaluated as nested commands. | 20:14 |
| openstackgerrit | Ondřej Nový proposed openstack/swift: Apply bash error handling consistently in all bash scripts https://review.openstack.org/357025 | 20:14 |
| clayg | onovy: well except I think it'd be more reasonable to just wrap the command you expect to generate the useful non-zero exit code and let the set -e do it's job? | 20:15 |
| clayg | or "just as reasonable" | 20:15 |
| clayg | ... maybe more - see I've already thought about it to much ;) | 20:15 |
| onovy | so i will enable "set -e", and only command which can fail i will override with || | 20:16 |
| clayg | onovy: i'm not sure - don't put to much faith into me - i'm actively anti bash | 20:16 |
| onovy | overengineering? :) | 20:17 |
| clayg | i know... see that's my point :'( i'm so sorry - use your best judgement | 20:17 |
| onovy | so i think i fixed resetswift now - never fail if already reseted and i will keep .functests this way. we can change it later :) | 20:19 |
| clayg | is patch 3 the latest one you'd like me to opine on? | 20:19 |
| *** gyee has joined #openstack-swift | 20:19 | |
| onovy | yep | 20:19 |
| clayg | onovy: awesome - thanks for pining me in here! | 20:19 |
| onovy | thanks too | 20:19 |
| timburke | so i was looking at the pyeclib memory leak that kota pointed out (as a bit of an aside) in bug 1631144 | 20:25 |
| openstack | bug 1631144 in OpenStack Object Storage (swift) "Swift Erasure Coding : Error when decoding fragment" [High,Confirmed] https://launchpad.net/bugs/1631144 | 20:25 |
| timburke | and trying to wrap my head around reference ownership, the nuances of the various Py*_SetItem functions and how they relate to ownership, and what all needs a Py_XDECREF in https://github.com/openstack/pyeclib/blob/1.3.1/src/c/pyeclib_c/pyeclib_c.c#L1030-L1094 | 20:25 |
| timburke | when i came across this snippet from https://docs.python.org/2/c-api/intro.html#reference-count-details : "However, in practice, you will rarely use these ways of creating and populating a tuple or list. There’s a generic function, Py_BuildValue(), that can create most common objects from C values, directed by a format string." | 20:25 |
| timburke | so... does anyone know why we don't just use that? | 20:25 |
| clayg | onovy: find: ‘/var/log/swift’: No such file or directory | 20:25 |
| clayg | there's a bit about "remove the following line if you didn't setup rsyslog ...." something? | 20:26 |
| onovy | # Remove the following line if you did not set up rsyslog for individual logging | 20:26 |
| onovy | so remove that line, if you did not setup resyslog for logging in this directory :) | 20:26 |
| clayg | so but ... i mean if everyone is just writing/editing this script/template to suit their own needs? | 20:27 |
| onovy | yep, you should edit this script if you don't follow SAIO docs completly | 20:28 |
| clayg | timburke: you could try to email tushar and kevin - but I'm guessing if works and simplifies things it would be acceptable? | 20:28 |
| briancline | does the use of x-container-{partition,host,device} maybe have to do with race conditions between ring changes and object-updater? | 20:31 |
| clayg | briancline: I think it's mostly because object servers's don't have container rings loaded in memory | 20:31 |
| briancline | orly | 20:32 |
| clayg | briancline: I think they could do the hash calculation and figure the partition from the container name - but they don't even validate the partition in the object request matches the name - they do whatever the proxy tells it | 20:32 |
| clayg | if the proxy didn't send the host & device - it'd have to look it up in the ring - and then also figure out some deterministic pattern to decide *which* of thoe nodes int he ring *this* object server should update | 20:33 |
| clayg | ... i'm not sure if there's an expectation that the proxy service will reload new rings faster that other services on the object nodes | 20:33 |
| briancline | oh, right. the coordination aspect of which nodes. that makes sense | 20:33 |
| clayg | the object-updater (pushes async pendings to container dbs) *does* reload the it's container rings and recalc the container host/device form the partition | 20:34 |
| briancline | ahh, so those headers are only really used when there's no need to write out an async | 20:35 |
| briancline | (which hopefully is almost always) | 20:35 |
| onovy | "always" | 20:36 |
| onovy | [async_pending] low: 24, high: 367, avg: 205.1, total: 6975, Failed: 0.0%, no_result: 0, reported: 34 | 20:36 |
| briancline | after 4 years with swift, i still learn all sorts of new nuanced things often | 20:36 |
| onovy | don't know if this number is normal but it's normal for our production cluster :) | 20:37 |
| briancline | yeah, that seems pretty normal. i usually see similar numbers for our busiest cluster | 20:39 |
| *** acoles is now known as acoles_ | 20:46 | |
| *** sgundur has quit IRC | 20:50 | |
| onovy | clayg: hey! if i change pwd in bash script, it's not changed in parent? that's new for me | 20:51 |
| onovy | so that cd - is completly useless | 20:51 |
| onovy | so i think i should change my opinion and add set -e to .*tests, and remove that cd - | 20:52 |
| clayg | onovy: no idea | 20:53 |
| onovy | ah, in .unittests is rm -rf of .coverage | 20:53 |
| * clayg taps out | 20:53 | |
| onovy | fok it :) | 20:54 |
| *** sgundur has joined #openstack-swift | 20:55 | |
| clayg | timburke: i pinged you last night on patch 360298 - any opinion on requests-mock? I think you and Jamie have been working on some sessions/keystone stuff with swiftclient - i may be incorrectly infering the use of requests-mock is related to that effort? | 20:55 |
| patchbot | https://review.openstack.org/#/c/360298/ - python-swiftclient - Use requests_mock and oslotest | 20:55 |
| onovy | thanks for review | 20:56 |
| clayg | honestly the *only* think I like about the current handling of httplib mocking is that if i'm not clear what it's doing I can read it in the same tree i'm currently working and if it doesn't do what I want in my test I can change it w/o any fuss | 20:56 |
| clayg | I do but a high value on both of those - but if using requests-mock means I don't have to go read source to have a good expectation of what it does or have to make any changes because it already does what I want - there's not really any downside ;) | 20:57 |
| *** chsc has quit IRC | 21:00 | |
| *** jordanP has joined #openstack-swift | 21:01 | |
| mattoliverau | Morning | 21:09 |
| *** mvk has joined #openstack-swift | 21:22 | |
| *** jordanP has quit IRC | 21:29 | |
| *** tdasilva has quit IRC | 21:30 | |
| *** ppalacios has quit IRC | 21:31 | |
| *** tdasilva has joined #openstack-swift | 21:33 | |
| *** sgundur has quit IRC | 21:34 | |
| *** sgundur has joined #openstack-swift | 21:34 | |
| *** sgundur has quit IRC | 21:51 | |
| *** zigo has quit IRC | 22:02 | |
| *** zigo has joined #openstack-swift | 22:05 | |
| *** zigo is now known as Guest50630 | 22:05 | |
| *** StraubTW_ has quit IRC | 22:09 | |
| *** vint_bra has quit IRC | 22:16 | |
| *** _JZ_ has quit IRC | 22:39 | |
| *** vint_bra has joined #openstack-swift | 22:44 | |
| *** vint_bra has quit IRC | 22:53 | |
| openstackgerrit | Tim Burke proposed openstack/pyeclib: Fix formatted get_metadata memory leak https://review.openstack.org/385185 | 22:57 |
| openstackgerrit | Tim Burke proposed openstack/pyeclib: Avoid segfault when raising exceptions https://review.openstack.org/385186 | 23:01 |
| *** nikivi has joined #openstack-swift | 23:27 | |
| *** kei_yama has joined #openstack-swift | 23:27 | |
| *** chsc has joined #openstack-swift | 23:39 | |
| *** nikivi has quit IRC | 23:53 | |
| *** nikivi has joined #openstack-swift | 23:53 | |
| *** nottrobin has quit IRC | 23:54 | |
| *** nottrobin has joined #openstack-swift | 23:55 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!