| *** mmcardle1 has quit IRC | 00:01 | |
| zaitcev | Some of the stuff in SP kinda makes my blood pressure raise. Or maybe I ate too much chocolate today. | 00:21 |
|---|---|---|
| zaitcev | portante noticed too | 00:21 |
| zaitcev | where is_deleted() added into .merge_timestamps(), but there were others like that | 00:22 |
| portante | yes, agreed, seems like we are just opening up race conditions | 00:24 |
| *** matsuhashi has joined #openstack-swift | 00:24 | |
| portante | zaitcev: ^ | 00:25 |
| *** dmorita has joined #openstack-swift | 00:28 | |
| openstackgerrit | Matthew Oliver proposed a change to openstack/swift: When a filesystem does't support xattr return a 507 https://review.openstack.org/99883 | 00:30 |
| openstackgerrit | Matthew Oliver proposed a change to openstack/swift: When a filesystem does't support xattr return a 507 https://review.openstack.org/99883 | 00:31 |
| notmyname | looks like I'll be talking to some people about Swift on all SSDs tomorrow. should be interesting to listen and learn | 00:47 |
| *** scohen has joined #openstack-swift | 00:48 | |
| *** scohen has quit IRC | 00:50 | |
| *** haomaiwa_ has joined #openstack-swift | 00:54 | |
| *** kopparam has joined #openstack-swift | 00:55 | |
| *** kopparam has quit IRC | 01:01 | |
| portante | notmyname: do we really support floating point number of replicas? | 01:25 |
| mattoliverau | notmyname: cool | 01:26 |
| portante | looking at the documentation for set_replicas of swift-ring-builder command | 01:26 |
| *** nosnos has joined #openstack-swift | 01:26 | |
| *** kenhui has joined #openstack-swift | 01:29 | |
| portante | we apparently do at least that we can set a replica count of 3.25 | 01:29 |
| portante | on my | 01:29 |
| portante | oh, that is | 01:30 |
| *** kenhui1 has joined #openstack-swift | 01:30 | |
| mattoliverau | Is anyone here coming to pyconau this year? There will be a Openstack miniconf segment. I was thinking about proposing a short talk on storage policies in swift (which should be in master by then), to make sure swift gets some time on the day. That is if no one objects :) Unless someone else more qaulified then me is coming of course. | 01:32 |
| *** kenhui has quit IRC | 01:32 | |
| *** kenhui1 has quit IRC | 01:32 | |
| *** kenhui1 has joined #openstack-swift | 01:35 | |
| *** kenhui has joined #openstack-swift | 01:35 | |
| *** charz has joined #openstack-swift | 01:36 | |
| *** kenhui1 has quit IRC | 01:36 | |
| *** nsquare has quit IRC | 01:37 | |
| *** kenhui has quit IRC | 01:38 | |
| *** haomaiwa_ has quit IRC | 01:43 | |
| *** foexle_ has joined #openstack-swift | 01:48 | |
| *** foexle has quit IRC | 01:51 | |
| *** asdfsf has joined #openstack-swift | 01:55 | |
| *** lpabon has joined #openstack-swift | 02:02 | |
| *** kenhui has joined #openstack-swift | 02:02 | |
| portante | peluse_, notmyname, creiht: http://techreport.com/review/26523/the-ssd-endurance-experiment-casualties-on-the-way-to-a-petabyte | 02:09 |
| *** kenhui has quit IRC | 02:14 | |
| pandemicsyn | heh we did some wear leveling tests when we deployed Intel 320's at Rackspace. Wonder if I still have those anywhere. | 02:17 |
| pandemicsyn | kinda curious to see how older generations of SSD's hold up | 02:17 |
| notmyname | portante: yes. but it's not completely straightforward. 3.25 replicas means that everything has 3 replicas and 25% (.25) has 4 replicas | 02:19 |
| portante | whoa | 02:22 |
| portante | sounds complicated | 02:22 |
| notmyname | portante: think of it as the sane migration story that allows you to change the replication factor without rebuilding the whole cluster from scratch | 02:23 |
| portante | the code supports arbitrary floats, so 3.0001 is a valid replica count | 02:23 |
| notmyname | ya | 02:23 |
| portante | okay, I'll just ignore that part of swift for now. :) | 02:23 |
| notmyname | lol | 02:23 |
| notmyname | portante: the user story is when you have a one region cluster (3 replicas) and you want to move to two regions and 4 replicas | 02:24 |
| portante | I am gonna go back to my world of integer replicas so that I can think about all this storage policies without exploding | 02:24 |
| notmyname | ya, it's orthogonal to the policies. | 02:24 |
| portante | agreed | 02:24 |
| notmyname | portante: and, for your sanity, each partition in the ring will only have an integer number of replicas | 02:24 |
| notmyname | and also don't assume things :-) | 02:25 |
| portante | I was trying to update the in-process functional tests with one 3-replica ring and one 2-replica ring, but I am failing to do so for some reason that says I don't really understand what I am doing | 02:25 |
| portante | :) | 02:26 |
| notmyname | portante: good ssd article | 02:28 |
| notmyname | portante: seems to confirm my guess that one of the hardest part of SSDs in Swift is doing good management of them. eg figuring out more than just fullness, but current health and expected lifetime | 02:29 |
| zaitcev | So... Apparently the account maintains status_changed_at, but it's never used for anything. | 02:29 |
| notmyname | other important parts, based on my reading, is getting read buffers configured properly all the way through the stack so that performance is correct | 02:30 |
| notmyname | zaitcev: wasn't that used for direct account management? like with PUT/DELETE to an account? | 02:30 |
| notmyname | ie predates autocreated accounts | 02:30 |
| pandemicsyn | notmyname: re: ssd's...yeah....you wanna keep rebalancing to a minimum if you run object storage on ssd's too otherwise your churning data like crazy | 02:31 |
| zaitcev | I think it just goes into account_stat table to die. I don't see a select that fetches it out, ever (on both SP and current truck). But I may be mistaken. | 02:32 |
| notmyname | pandemicsyn: ya. we've also been looking at some things related to TRIM. turns out that's a good idea too :-) | 02:32 |
| zaitcev | Anyhow... | 02:32 |
| pandemicsyn | for the best performance when it comes to SSD's you'll wanna run Swift on Windows 7 Ultimate | 02:33 |
| notmyname | pandemicsyn: and it's that sort of thing (less rebalancing) that is something I'm very curious about and how much that can be set in Swift. ie can we have some config that keeps data less balanced because SSD? maybe in the ring-builder? | 02:33 |
| notmyname | pandemicsyn: naturally | 02:33 |
| notmyname | actually, the "less balanced but also less transfer" might be good in the non-SSD case with eg power management. peluse_, you're interested in that | 02:37 |
| notmyname | some sort of backblaze-style "fill up the jbod and turn it off" sort of storage. would be pretty distinct | 02:38 |
| * notmyname is imagining selecting storage policies based on durability schemes and power management schemes | 02:39 | |
| * pandemicsyn shudders | 02:39 | |
| pandemicsyn | notmyname: thats kinda how NAST worked | 02:40 |
| pandemicsyn | well as far as not using the drives that often that is | 02:41 |
| notmyname | pandemicsyn: ya. and there's a reason swift exists today :-) | 02:41 |
| *** foexle_ has quit IRC | 02:41 | |
| *** bkopilov_ has quit IRC | 02:41 | |
| notmyname | I'd much prefer to tackle any power concerns by first looking at overall efficiency and density | 02:42 |
| *** Midnightmyth has joined #openstack-swift | 02:45 | |
| *** kopparam has joined #openstack-swift | 03:00 | |
| *** kopparam has quit IRC | 03:05 | |
| peluse_ | portante: you there? | 03:18 |
| portante | yes | 03:19 |
| portante | well, at least I am typing here, not sure how much thought I can pull off at this point | 03:19 |
| peluse_ | wrt the failing probe test, did you see the 2 updates I put in there? | 03:20 |
| portante | not yet | 03:20 |
| peluse_ | ahh OK, they're after Sam's comment but before yours | 03:20 |
| portante | was just working with the tip of the sp chain, and did NOT see a failure on my two SAIOs, FWIW | 03:20 |
| portante | I don't have torgomatic's configuration, though | 03:20 |
| openstackgerrit | Matthew Oliver proposed a change to openstack/swift: When a filesystem does't support xattr return a 507 https://review.openstack.org/99883 | 03:21 |
| *** dmsimard_away is now known as dmsimard | 03:21 | |
| portante | peluse_: I see them now, yes, I started to debug it, but gave up after I could not keep focused on this longer than a minute or two | 03:22 |
| *** dmsimard is now known as dmsimard_away | 03:22 | |
| peluse_ | yeah, OK. Getting a bead on it sort of... and you do have to have the config he mentions in there. Non-0 default and rsync. I know what's happening but can't explain why ssync works and can't find what I think would be obvisouly wrong (using 0 insetad of the specified default)... | 03:23 |
| portante | I'd be happy to look at it tomorrow | 03:24 |
| portante | in the mean time, here is a simple patch to skip container-reconciler probe tests when no container-reconciler.conf file is present in a SAIO: https://gist.github.com/portante/6ab72faa6db065d32a8e | 03:25 |
| peluse_ | cool, thanks. doubt I'll resolve it here in the short amount of time I plan on spending for the rest of the eve :) | 03:25 |
| portante | One of my two SAIOs was in that state, and the errors from the probe tests were non-obvious what was happening | 03:25 |
| portante | peluse_: also, working on adding a storage-policy configuration for the in-process functional tests | 03:26 |
| peluse_ | cool | 03:26 |
| *** zhiyan_ is now known as zhiyan | 03:28 | |
| *** Midnightmyth has quit IRC | 03:30 | |
| *** matsuhashi has quit IRC | 03:30 | |
| peluse_ | portante: hmm, know off the top of your head if ssync's use of DiskFile would cause the creation of the policy specific tmp file? (self._tmpdir). Looking... if so that would explain it | 03:33 |
| *** matsuhashi has joined #openstack-swift | 03:37 | |
| openstackgerrit | Matthew Oliver proposed a change to openstack/swift: When a filesystem does't support xattr return a 507 https://review.openstack.org/99883 | 03:38 |
| *** nosnos has quit IRC | 03:41 | |
| *** matsuhashi has quit IRC | 03:44 | |
| *** matsuhashi has joined #openstack-swift | 03:45 | |
| *** matsuhashi has quit IRC | 03:49 | |
| *** Kbee has joined #openstack-swift | 03:53 | |
| portante | peluse_: I think so, because the tmpdir is created when the DiskFile object's .create() method is invoked | 03:54 |
| *** matsuhashi has joined #openstack-swift | 03:57 | |
| peluse_ | portante: yeah, that is probably part of it. Just updated gerrit... this is an annoying piss-ant little thing if nothing else... | 04:00 |
| portante | peluse_: but also the nature of having two sync methods. :) | 04:01 |
| portante | one that uses the DiskFile backend, and one that assumes it knows what the on-disk layout looks like. :) | 04:01 |
| *** matsuhashi has quit IRC | 04:03 | |
| *** matsuhashi has joined #openstack-swift | 04:04 | |
| *** bkopilov_ has joined #openstack-swift | 04:12 | |
| *** ppai has joined #openstack-swift | 04:21 | |
| *** nosnos has joined #openstack-swift | 04:30 | |
| *** matsuhashi has quit IRC | 04:36 | |
| *** matsuhas_ has joined #openstack-swift | 04:37 | |
| *** haomaiwa_ has joined #openstack-swift | 04:39 | |
| openstackgerrit | Zhang Hua proposed a change to openstack/swift: Add distributed tracing capablities in logging. https://review.openstack.org/93677 | 04:43 |
| *** gyee has quit IRC | 04:51 | |
| *** scohen has joined #openstack-swift | 04:53 | |
| *** haomaiw__ has joined #openstack-swift | 04:54 | |
| *** haomaiwa_ has quit IRC | 04:56 | |
| *** psharma has joined #openstack-swift | 05:01 | |
| *** nshaikh has joined #openstack-swift | 05:08 | |
| *** bkopilov_ has quit IRC | 05:16 | |
| *** bkopilov has joined #openstack-swift | 05:17 | |
| *** zhiyan is now known as zhiyan_ | 05:20 | |
| *** zaitcev has quit IRC | 05:23 | |
| *** mlipchuk has joined #openstack-swift | 05:24 | |
| *** mlipchuk has quit IRC | 05:28 | |
| *** kopparam has joined #openstack-swift | 05:39 | |
| *** mlipchuk has joined #openstack-swift | 05:42 | |
| openstackgerrit | Matthew Oliver proposed a change to openstack/swift: When a filesystem does't support xattr return a 507 https://review.openstack.org/99883 | 05:47 |
| *** scohen has joined #openstack-swift | 05:55 | |
| *** scohen has quit IRC | 05:55 | |
| *** zhiyan_ is now known as zhiyan | 06:02 | |
| *** kopparam has quit IRC | 06:05 | |
| *** kopparam has joined #openstack-swift | 06:05 | |
| *** kopparam has quit IRC | 06:10 | |
| Kbee | notmyname, Solved it.. there were conflicting ports assigned to account and object servers.. correcting that resolved the issue | 06:34 |
| *** asdfsf has quit IRC | 06:37 | |
| *** chandan_kumar has quit IRC | 06:38 | |
| *** scohen has joined #openstack-swift | 06:51 | |
| * hugokuo hurray | 06:51 | |
| mattoliverau | I'm calling it day, have a great night/day depending on where you are in the world :) | 06:53 |
| hugokuo | bye | 06:58 |
| *** foexle has joined #openstack-swift | 07:00 | |
| *** joeljwright has joined #openstack-swift | 07:05 | |
| *** kopparam has joined #openstack-swift | 07:11 | |
| *** kopparam_ has joined #openstack-swift | 07:15 | |
| *** kopparam has quit IRC | 07:16 | |
| *** ujjain has quit IRC | 07:20 | |
| *** itarchitectkev_ has joined #openstack-swift | 07:21 | |
| *** mmcardle has joined #openstack-swift | 07:39 | |
| *** ujjain2 has joined #openstack-swift | 07:40 | |
| *** ujjain2 has quit IRC | 07:41 | |
| *** ujjain has joined #openstack-swift | 07:41 | |
| *** ujjain has quit IRC | 07:43 | |
| *** ujjain has joined #openstack-swift | 07:43 | |
| *** matsuhas_ has quit IRC | 07:49 | |
| *** matsuhashi has joined #openstack-swift | 07:51 | |
| *** mkollaro has joined #openstack-swift | 07:53 | |
| *** nacim has joined #openstack-swift | 07:58 | |
| *** nacim has quit IRC | 07:59 | |
| *** mmcardle has quit IRC | 08:16 | |
| *** nosnos has quit IRC | 08:16 | |
| *** judd7 has quit IRC | 08:16 | |
| *** rieglflo has quit IRC | 08:16 | |
| *** ahale has quit IRC | 08:16 | |
| *** itarchitectkev_ has quit IRC | 08:16 | |
| *** haomaiw__ has quit IRC | 08:16 | |
| *** wer has quit IRC | 08:16 | |
| *** AbyssOne_ has quit IRC | 08:16 | |
| *** torgomatic has quit IRC | 08:16 | |
| *** d0ugal has quit IRC | 08:16 | |
| *** hurricanerix has quit IRC | 08:16 | |
| *** jd__ has quit IRC | 08:16 | |
| *** fbo has quit IRC | 08:16 | |
| *** portante has quit IRC | 08:16 | |
| *** russell_h has quit IRC | 08:16 | |
| *** briancline has quit IRC | 08:16 | |
| *** pandemicsyn has quit IRC | 08:16 | |
| *** sileht has quit IRC | 08:16 | |
| *** mlanner has quit IRC | 08:16 | |
| *** occupant has quit IRC | 08:16 | |
| *** zacksh has quit IRC | 08:16 | |
| *** mhu has quit IRC | 08:16 | |
| *** bgmccollum has quit IRC | 08:16 | |
| *** zigo has quit IRC | 08:16 | |
| *** notmyname has quit IRC | 08:16 | |
| *** zhiyan has quit IRC | 08:16 | |
| *** chmouel has quit IRC | 08:16 | |
| *** anderstj has quit IRC | 08:16 | |
| *** joearnold has quit IRC | 08:16 | |
| *** glange has quit IRC | 08:16 | |
| *** mkollaro has quit IRC | 08:16 | |
| *** ujjain has quit IRC | 08:16 | |
| *** foexle has quit IRC | 08:16 | |
| *** bkopilov has quit IRC | 08:16 | |
| *** psharma has quit IRC | 08:16 | |
| *** acoles has quit IRC | 08:16 | |
| *** miqui has quit IRC | 08:16 | |
| *** zul has quit IRC | 08:16 | |
| *** AbyssOne has quit IRC | 08:16 | |
| *** mattoliverau has quit IRC | 08:16 | |
| *** zz_wasmum has quit IRC | 08:16 | |
| *** nshaikh has quit IRC | 08:16 | |
| *** Kbee has quit IRC | 08:16 | |
| *** openstackgerrit has quit IRC | 08:16 | |
| *** fifieldt has quit IRC | 08:16 | |
| *** IRTermite has quit IRC | 08:16 | |
| *** omame has quit IRC | 08:16 | |
| *** matsuhashi has quit IRC | 08:16 | |
| *** joeljwright has quit IRC | 08:16 | |
| *** dmorita has quit IRC | 08:16 | |
| *** ondergetekende has quit IRC | 08:16 | |
| *** openfly has quit IRC | 08:16 | |
| *** ppai has quit IRC | 08:16 | |
| *** Dieterbe has quit IRC | 08:16 | |
| *** dosaboy has quit IRC | 08:16 | |
| *** DisneyRicky has quit IRC | 08:16 | |
| *** JoshNang has quit IRC | 08:16 | |
| *** scohen has quit IRC | 08:16 | |
| *** mlipchuk has quit IRC | 08:16 | |
| *** mkerrin has quit IRC | 08:16 | |
| *** jokke_ has quit IRC | 08:16 | |
| *** ekarlso has quit IRC | 08:16 | |
| *** charz has quit IRC | 08:16 | |
| *** madhuri has quit IRC | 08:16 | |
| *** StevenK has quit IRC | 08:16 | |
| *** byeager_away has quit IRC | 08:16 | |
| *** rpedde has quit IRC | 08:16 | |
| *** thurloat has quit IRC | 08:16 | |
| *** otherjon has quit IRC | 08:16 | |
| *** kragniz has quit IRC | 08:16 | |
| *** serverascode has quit IRC | 08:16 | |
| *** CybergeekDK has quit IRC | 08:16 | |
| *** ChanServ has quit IRC | 08:16 | |
| *** swat30 has quit IRC | 08:16 | |
| *** cschwede has quit IRC | 08:16 | |
| *** tanee has quit IRC | 08:16 | |
| *** ezml has quit IRC | 08:16 | |
| *** dfg has quit IRC | 08:16 | |
| *** sts has quit IRC | 08:16 | |
| *** jeblair has quit IRC | 08:16 | |
| *** wkelly has quit IRC | 08:16 | |
| *** anticw has quit IRC | 08:16 | |
| *** zackmdavis has quit IRC | 08:16 | |
| *** creiht has quit IRC | 08:16 | |
| *** amandap has quit IRC | 08:16 | |
| *** mtreinish has quit IRC | 08:17 | |
| *** redbo has quit IRC | 08:17 | |
| *** kopparam_ has quit IRC | 08:17 | |
| *** kevinc_ has quit IRC | 08:17 | |
| *** sandywalsh has quit IRC | 08:17 | |
| *** nthacker__ has quit IRC | 08:17 | |
| *** patchbot has quit IRC | 08:17 | |
| *** bobby_ has quit IRC | 08:17 | |
| *** flowouffff has quit IRC | 08:17 | |
| *** ctennis has quit IRC | 08:17 | |
| *** acorwin has quit IRC | 08:17 | |
| *** clayg has quit IRC | 08:17 | |
| *** goneri has quit IRC | 08:17 | |
| *** Anticimex has quit IRC | 08:17 | |
| *** grapsus_ has quit IRC | 08:17 | |
| *** Mikalv has quit IRC | 08:17 | |
| *** goodes has quit IRC | 08:17 | |
| *** MooingLemur has quit IRC | 08:17 | |
| *** esmute has quit IRC | 08:17 | |
| *** peluse_ has quit IRC | 08:17 | |
| *** joearnold has joined #openstack-swift | 08:22 | |
| *** glange has joined #openstack-swift | 08:22 | |
| *** anderstj has joined #openstack-swift | 08:22 | |
| *** chmouel has joined #openstack-swift | 08:22 | |
| *** notmyname has joined #openstack-swift | 08:22 | |
| *** zigo has joined #openstack-swift | 08:22 | |
| *** zhiyan has joined #openstack-swift | 08:22 | |
| *** bgmccollum has joined #openstack-swift | 08:22 | |
| *** mhu has joined #openstack-swift | 08:22 | |
| *** zacksh has joined #openstack-swift | 08:22 | |
| *** briancline has joined #openstack-swift | 08:22 | |
| *** russell_h has joined #openstack-swift | 08:22 | |
| *** occupant has joined #openstack-swift | 08:22 | |
| *** portante has joined #openstack-swift | 08:22 | |
| *** dickson.freenode.net sets mode: +vvvv glange chmouel notmyname portante | 08:22 | |
| *** jd__ has joined #openstack-swift | 08:22 | |
| *** fbo has joined #openstack-swift | 08:22 | |
| *** mlanner has joined #openstack-swift | 08:22 | |
| *** hurricanerix has joined #openstack-swift | 08:22 | |
| *** sileht has joined #openstack-swift | 08:22 | |
| *** ahale has joined #openstack-swift | 08:22 | |
| *** pandemicsyn has joined #openstack-swift | 08:22 | |
| *** d0ugal has joined #openstack-swift | 08:22 | |
| *** torgomatic has joined #openstack-swift | 08:22 | |
| *** rieglflo has joined #openstack-swift | 08:22 | |
| *** AbyssOne_ has joined #openstack-swift | 08:22 | |
| *** wer has joined #openstack-swift | 08:22 | |
| *** judd7 has joined #openstack-swift | 08:22 | |
| *** nosnos has joined #openstack-swift | 08:22 | |
| *** haomaiw__ has joined #openstack-swift | 08:22 | |
| *** itarchitectkev_ has joined #openstack-swift | 08:22 | |
| *** mmcardle has joined #openstack-swift | 08:22 | |
| *** mkollaro has joined #openstack-swift | 08:22 | |
| *** matsuhashi has joined #openstack-swift | 08:22 | |
| *** ujjain has joined #openstack-swift | 08:22 | |
| *** joeljwright has joined #openstack-swift | 08:22 | |
| *** foexle has joined #openstack-swift | 08:22 | |
| *** scohen has joined #openstack-swift | 08:22 | |
| *** mlipchuk has joined #openstack-swift | 08:22 | |
| *** bkopilov has joined #openstack-swift | 08:22 | |
| *** nshaikh has joined #openstack-swift | 08:22 | |
| *** psharma has joined #openstack-swift | 08:22 | |
| *** ppai has joined #openstack-swift | 08:22 | |
| *** Kbee has joined #openstack-swift | 08:22 | |
| *** dmorita has joined #openstack-swift | 08:22 | |
| *** acoles has joined #openstack-swift | 08:22 | |
| *** dickson.freenode.net sets mode: +vv torgomatic acoles | 08:22 | |
| *** kevinc_ has joined #openstack-swift | 08:22 | |
| *** miqui has joined #openstack-swift | 08:22 | |
| *** zul has joined #openstack-swift | 08:22 | |
| *** sts has joined #openstack-swift | 08:22 | |
| *** sandywalsh has joined #openstack-swift | 08:22 | |
| *** openstackgerrit has joined #openstack-swift | 08:22 | |
| *** ondergetekende has joined #openstack-swift | 08:22 | |
| *** madhuri has joined #openstack-swift | 08:22 | |
| *** AbyssOne has joined #openstack-swift | 08:22 | |
| *** fifieldt has joined #openstack-swift | 08:22 | |
| *** Dieterbe has joined #openstack-swift | 08:22 | |
| *** dosaboy has joined #openstack-swift | 08:22 | |
| *** mattoliverau has joined #openstack-swift | 08:22 | |
| *** DisneyRicky has joined #openstack-swift | 08:22 | |
| *** nthacker__ has joined #openstack-swift | 08:22 | |
| *** IRTermite has joined #openstack-swift | 08:22 | |
| *** omame has joined #openstack-swift | 08:22 | |
| *** mkerrin has joined #openstack-swift | 08:22 | |
| *** openfly has joined #openstack-swift | 08:22 | |
| *** zz_wasmum has joined #openstack-swift | 08:22 | |
| *** flowouffff has joined #openstack-swift | 08:22 | |
| *** StevenK has joined #openstack-swift | 08:22 | |
| *** goodes has joined #openstack-swift | 08:22 | |
| *** patchbot has joined #openstack-swift | 08:22 | |
| *** bobby_ has joined #openstack-swift | 08:22 | |
| *** JoshNang has joined #openstack-swift | 08:22 | |
| *** jokke_ has joined #openstack-swift | 08:22 | |
| *** swat30 has joined #openstack-swift | 08:22 | |
| *** thurloat has joined #openstack-swift | 08:22 | |
| *** cschwede has joined #openstack-swift | 08:22 | |
| *** MooingLemur has joined #openstack-swift | 08:22 | |
| *** esmute has joined #openstack-swift | 08:22 | |
| *** ekarlso has joined #openstack-swift | 08:22 | |
| *** tanee has joined #openstack-swift | 08:22 | |
| *** byeager_away has joined #openstack-swift | 08:22 | |
| *** ezml has joined #openstack-swift | 08:22 | |
| *** dfg has joined #openstack-swift | 08:22 | |
| *** rpedde has joined #openstack-swift | 08:22 | |
| *** goneri has joined #openstack-swift | 08:22 | |
| *** otherjon has joined #openstack-swift | 08:22 | |
| *** peluse_ has joined #openstack-swift | 08:22 | |
| *** kragniz has joined #openstack-swift | 08:22 | |
| *** jeblair has joined #openstack-swift | 08:22 | |
| *** Anticimex has joined #openstack-swift | 08:22 | |
| *** grapsus_ has joined #openstack-swift | 08:22 | |
| *** serverascode has joined #openstack-swift | 08:22 | |
| *** dickson.freenode.net sets mode: +v dfg | 08:22 | |
| *** wkelly has joined #openstack-swift | 08:22 | |
| *** Mikalv has joined #openstack-swift | 08:22 | |
| *** anticw has joined #openstack-swift | 08:22 | |
| *** zackmdavis has joined #openstack-swift | 08:22 | |
| *** CybergeekDK has joined #openstack-swift | 08:22 | |
| *** creiht has joined #openstack-swift | 08:22 | |
| *** amandap has joined #openstack-swift | 08:22 | |
| *** mtreinish has joined #openstack-swift | 08:22 | |
| *** ChanServ has joined #openstack-swift | 08:22 | |
| *** ctennis has joined #openstack-swift | 08:22 | |
| *** redbo has joined #openstack-swift | 08:22 | |
| *** acorwin has joined #openstack-swift | 08:22 | |
| *** clayg has joined #openstack-swift | 08:22 | |
| *** dickson.freenode.net sets mode: +vovv creiht ChanServ redbo clayg | 08:22 | |
| *** d0ugal has quit IRC | 08:24 | |
| *** d0ugal_ has joined #openstack-swift | 08:24 | |
| *** d0ugal_ is now known as d0ugal | 08:25 | |
| *** d0ugal has quit IRC | 08:25 | |
| *** d0ugal has joined #openstack-swift | 08:25 | |
| *** fifieldt has quit IRC | 08:27 | |
| *** fifieldt has joined #openstack-swift | 08:28 | |
| *** mmcardle has quit IRC | 08:33 | |
| *** nosnos has quit IRC | 08:33 | |
| *** judd7 has quit IRC | 08:33 | |
| *** rieglflo has quit IRC | 08:33 | |
| *** ahale has quit IRC | 08:33 | |
| *** itarchitectkev_ has quit IRC | 08:33 | |
| *** haomaiw__ has quit IRC | 08:33 | |
| *** wer has quit IRC | 08:33 | |
| *** AbyssOne_ has quit IRC | 08:33 | |
| *** torgomatic has quit IRC | 08:33 | |
| *** hurricanerix has quit IRC | 08:33 | |
| *** jd__ has quit IRC | 08:33 | |
| *** fbo has quit IRC | 08:33 | |
| *** portante has quit IRC | 08:33 | |
| *** russell_h has quit IRC | 08:33 | |
| *** briancline has quit IRC | 08:33 | |
| *** pandemicsyn has quit IRC | 08:33 | |
| *** sileht has quit IRC | 08:33 | |
| *** mlanner has quit IRC | 08:33 | |
| *** occupant has quit IRC | 08:33 | |
| *** zacksh has quit IRC | 08:33 | |
| *** mhu has quit IRC | 08:33 | |
| *** bgmccollum has quit IRC | 08:33 | |
| *** zigo has quit IRC | 08:33 | |
| *** notmyname has quit IRC | 08:33 | |
| *** zhiyan has quit IRC | 08:33 | |
| *** chmouel has quit IRC | 08:33 | |
| *** anderstj has quit IRC | 08:33 | |
| *** joearnold has quit IRC | 08:33 | |
| *** glange has quit IRC | 08:33 | |
| *** kopparam has joined #openstack-swift | 08:40 | |
| *** mmcardle has joined #openstack-swift | 08:43 | |
| *** itarchitectkev_ has joined #openstack-swift | 08:43 | |
| *** haomaiw__ has joined #openstack-swift | 08:43 | |
| *** nosnos has joined #openstack-swift | 08:43 | |
| *** judd7 has joined #openstack-swift | 08:43 | |
| *** wer has joined #openstack-swift | 08:43 | |
| *** AbyssOne_ has joined #openstack-swift | 08:43 | |
| *** rieglflo has joined #openstack-swift | 08:43 | |
| *** torgomatic has joined #openstack-swift | 08:43 | |
| *** pandemicsyn has joined #openstack-swift | 08:43 | |
| *** ahale has joined #openstack-swift | 08:43 | |
| *** sileht has joined #openstack-swift | 08:43 | |
| *** zacksh has joined #openstack-swift | 08:43 | |
| *** mhu has joined #openstack-swift | 08:43 | |
| *** bgmccollum has joined #openstack-swift | 08:43 | |
| *** zhiyan has joined #openstack-swift | 08:43 | |
| *** zigo has joined #openstack-swift | 08:43 | |
| *** notmyname has joined #openstack-swift | 08:43 | |
| *** chmouel has joined #openstack-swift | 08:43 | |
| *** anderstj has joined #openstack-swift | 08:43 | |
| *** glange has joined #openstack-swift | 08:43 | |
| *** dickson.freenode.net sets mode: +vvvv torgomatic notmyname chmouel glange | 08:43 | |
| *** joearnold has joined #openstack-swift | 08:43 | |
| *** briancline has joined #openstack-swift | 08:43 | |
| *** russell_h has joined #openstack-swift | 08:43 | |
| *** occupant has joined #openstack-swift | 08:43 | |
| *** portante has joined #openstack-swift | 08:43 | |
| *** jd__ has joined #openstack-swift | 08:43 | |
| *** fbo_away has joined #openstack-swift | 08:43 | |
| *** mlanner has joined #openstack-swift | 08:43 | |
| *** hurricanerix has joined #openstack-swift | 08:43 | |
| *** dickson.freenode.net sets mode: +v portante | 08:43 | |
| *** nosnos has quit IRC | 08:54 | |
| *** nosnos has joined #openstack-swift | 08:55 | |
| *** kopparam has quit IRC | 08:56 | |
| *** kopparam has joined #openstack-swift | 08:56 | |
| *** IRTermite has quit IRC | 08:59 | |
| *** Kbee has quit IRC | 09:01 | |
| *** kopparam has quit IRC | 09:01 | |
| *** mkollaro has quit IRC | 09:01 | |
| *** IRTermite has joined #openstack-swift | 09:03 | |
| *** kopparam has joined #openstack-swift | 09:03 | |
| *** kopparam has quit IRC | 09:04 | |
| *** Kbee has joined #openstack-swift | 09:09 | |
| *** jamiehannaford has joined #openstack-swift | 09:12 | |
| joeljwright | Hi everyone, the review process for the patch I proposed to add an additional high-level API for the python-swiftclient library (https://review.openstack.org/#/c/85453/) seems to have stalled | 09:18 |
| *** Kbee has quit IRC | 09:18 | |
| *** jamiehannaford has quit IRC | 09:19 | |
| joeljwright | This update aims to make a lot of the high-level, multithreaded logic in shell.py available to developers using python-swiftclient in their own projects | 09:19 |
| joeljwright | to try to help get the review process going again I have written a short design document that hopefully makes the motivation and changes more clear https://github.com/joel-wright/openstack-notes/blob/master/swift-service/design.md | 09:20 |
| *** pberis has quit IRC | 09:21 | |
| *** matsuhashi has quit IRC | 09:22 | |
| *** pberis has joined #openstack-swift | 09:23 | |
| *** matsuhashi has joined #openstack-swift | 09:24 | |
| *** matsuhashi has quit IRC | 09:30 | |
| *** matsuhashi has joined #openstack-swift | 09:30 | |
| *** pberis has quit IRC | 09:30 | |
| *** Kbee has joined #openstack-swift | 09:31 | |
| *** pberis has joined #openstack-swift | 09:33 | |
| *** matsuhashi has quit IRC | 09:34 | |
| acoles | joeljwright: thanks. suggest you link to that doc from the blueprint if you haven't already. | 09:36 |
| joeljwright | I added it to the review, will add a link from the blueprint now | 09:36 |
| acoles | joeljwright: ah, yes, just saw it on review. thx | 09:37 |
| *** matsuhashi has joined #openstack-swift | 09:37 | |
| *** goneri has quit IRC | 09:39 | |
| *** nshaikh has quit IRC | 09:41 | |
| *** nshaikh has joined #openstack-swift | 09:41 | |
| *** pberis has quit IRC | 09:41 | |
| *** openstackgerrit_ has joined #openstack-swift | 09:43 | |
| *** matsuhashi has quit IRC | 09:48 | |
| *** matsuhashi has joined #openstack-swift | 09:50 | |
| *** kopparam has joined #openstack-swift | 09:53 | |
| *** kopparam_ has joined #openstack-swift | 09:53 | |
| *** kopparam has quit IRC | 09:57 | |
| *** jamiehannaford has joined #openstack-swift | 10:01 | |
| *** jamiehannaford has quit IRC | 10:02 | |
| *** matsuhashi has quit IRC | 10:03 | |
| *** matsuhashi has joined #openstack-swift | 10:03 | |
| *** matsuhashi has quit IRC | 10:04 | |
| *** matsuhashi has joined #openstack-swift | 10:04 | |
| *** dmorita has quit IRC | 10:13 | |
| *** matsuhashi has quit IRC | 10:14 | |
| *** matsuhashi has joined #openstack-swift | 10:21 | |
| *** matsuhashi has quit IRC | 10:35 | |
| *** matsuhashi has joined #openstack-swift | 10:36 | |
| *** matsuhashi has quit IRC | 10:40 | |
| rieglflo | hi everyone. i'm authenticated via keystone as admin. i create a user + tenant + root container (within swift). i set the read and write acl for the new user on the root container. why does a following list container request, authenticated as the newly created user, not show the container? because the owner is the admin user? can i change that? | 10:43 |
| *** Kbee has quit IRC | 10:52 | |
| rieglflo | maybe i'm an impasse here. what is the default role i have to assign a user + tenant so he is able to list his containers, create new ones and create objects? swiftoperator works fine, but is too much i think? | 11:00 |
| *** sombrafam has joined #openstack-swift | 11:24 | |
| acoles | rieglflo: the default role required for a user to create containers & objects is admin or swiftoperator | 11:31 |
| *** joeljwright1 has joined #openstack-swift | 11:32 | |
| acoles | rieglflo: i.e. in keystone the user must have role admin or swiftoperator on the tenant | 11:32 |
| *** joeljwright2 has joined #openstack-swift | 11:35 | |
| *** joeljwright has quit IRC | 11:36 | |
| *** joeljwright1 has quit IRC | 11:37 | |
| *** mkollaro has joined #openstack-swift | 11:42 | |
| *** kopparam_ has quit IRC | 12:11 | |
| *** kopparam has joined #openstack-swift | 12:12 | |
| rieglflo | acoles: is there no way to create a container as admin, set quotas and acl read/write for another user and set the new user as new owner? | 12:13 |
| *** kopparam has quit IRC | 12:16 | |
| acoles | rieglflo: a user with admin role on the tenant can create a container and set ACLs on that container to grant access to another user | 12:18 |
| acoles | rieglflo: where 'access'=list container and create/delete objects | 12:19 |
| *** joeljwright has joined #openstack-swift | 12:19 | |
| acoles | rieglflo: but ACL does not grant right to delete the container or change the ACL, if that is what you mean by set other user as 'owner' | 12:19 |
| acoles | rieglflo: looks like the container-quotas middleware takes care of setting a quota | 12:20 |
| *** ahale has quit IRC | 12:20 | |
| *** joeljwright1 has joined #openstack-swift | 12:21 | |
| *** joeljwright2 has quit IRC | 12:22 | |
| *** joeljwright has quit IRC | 12:23 | |
| rieglflo | acoles: but the container will never show up in the list of the containers the user owns a.k. via GET /v1/AUTH_b7f1cad131474e01a7305c5122d1bc72?delimiter=%2F&limit=100&format=json if the user-id is b7f1cad131474e01a7305c5122d1bc72 when the container has been created as another user(admin) | 12:23 |
| rieglflo | acoles: thats an issue because already existing clients, work that way on listing a users containers | 12:28 |
| *** ppai has quit IRC | 12:28 | |
| *** annegentle has joined #openstack-swift | 12:30 | |
| acoles | rieglflo: the piece b7f1cad131474e01a7305c5122d1bc72 after AUTH_ is usually the tenant id , not the user id. It defines the account in swift within which containers are created. So containers in one tenant/account AUTH_X will never list under another account AUTH_Y. | 12:30 |
| *** zul has quit IRC | 12:31 | |
| *** zul has joined #openstack-swift | 12:31 | |
| acoles | rieglflo: but more than one user can access the same tenant/account AUTH_X is each user has admin role on that tenant, or via an ACL. | 12:31 |
| acoles | rieglflo: /is/if/ | 12:32 |
| acoles | rieglflo: sorry i have to got to a meeting. | 12:33 |
| rieglflo | acoles: thanks for your clarifcations and time | 12:33 |
| *** mmcardle has quit IRC | 12:36 | |
| *** nacim has joined #openstack-swift | 12:41 | |
| *** mmcardle has joined #openstack-swift | 12:42 | |
| *** haomaiw__ has quit IRC | 12:43 | |
| *** haomaiwang has joined #openstack-swift | 12:43 | |
| *** haomaiwang has quit IRC | 12:46 | |
| *** haomaiwang has joined #openstack-swift | 12:47 | |
| *** haomaiw__ has joined #openstack-swift | 12:52 | |
| *** haomaiwang has quit IRC | 12:53 | |
| *** nosnos has quit IRC | 12:53 | |
| *** bkopilov has quit IRC | 12:54 | |
| *** annegentle has quit IRC | 12:58 | |
| peluse_ | portante: FYI just updated the probe test failure thingy we were talking about that Sam reported, root cuased it finally | 13:01 |
| rieglflo | is there a way to restrict the max expire seconds on the tempurl plugin on the server side, without patching the plugin? | 13:05 |
| portante | peluse_: cool I'll take a look when I get into the office in a bit | 13:07 |
| *** zul has quit IRC | 13:08 | |
| *** zul has joined #openstack-swift | 13:08 | |
| *** psharma has quit IRC | 13:12 | |
| acoles | peluse_: i was just looking at that. reached the same conclusion i think... tmp dir is created by the db replicator, tmp-N dirs shouldn't be expected. | 13:15 |
| *** zhiyan is now known as zhiyan_ | 13:17 | |
| *** uksysadmin has joined #openstack-swift | 13:25 | |
| *** itarchitectkev_ has quit IRC | 13:28 | |
| *** itarchitectkev_ has joined #openstack-swift | 13:33 | |
| peluse_ | acoles: yup, thanks | 13:34 |
| *** uksysadmin has quit IRC | 13:36 | |
| peluse_ | acoles: and that tmp creation in db repl feels like a bit of a hack... | 13:36 |
| *** pberis has joined #openstack-swift | 13:42 | |
| *** pberis has quit IRC | 13:42 | |
| *** pberis has joined #openstack-swift | 13:42 | |
| *** nacim has quit IRC | 13:51 | |
| *** kopparam has joined #openstack-swift | 13:55 | |
| *** 77CAADKRD has joined #openstack-swift | 13:56 | |
| openstackgerrit | OpenStack Proposal Bot proposed a change to openstack/swift: Updated from global requirements https://review.openstack.org/88736 | 13:57 |
| *** scohen has quit IRC | 13:59 | |
| *** scohen_ has joined #openstack-swift | 13:59 | |
| *** kopparam has quit IRC | 14:00 | |
| *** lpabon has joined #openstack-swift | 14:01 | |
| *** kopparam has joined #openstack-swift | 14:04 | |
| *** nshaikh has quit IRC | 14:16 | |
| *** zhiyan_ is now known as zhiyan | 14:16 | |
| *** mwstorer has joined #openstack-swift | 14:21 | |
| *** uksysadmin has joined #openstack-swift | 14:23 | |
| *** agentleone has joined #openstack-swift | 14:24 | |
| *** itarchitectkev_ has quit IRC | 14:26 | |
| *** haomaiw__ has quit IRC | 14:27 | |
| *** haomaiwa_ has joined #openstack-swift | 14:28 | |
| *** zhiyan is now known as zhiyan_ | 14:35 | |
| *** zhiyan_ is now known as zhiyan | 14:36 | |
| *** zhiyan is now known as zhiyan_ | 14:37 | |
| *** zhiyan_ is now known as zhiyan | 14:37 | |
| *** tdasilva has joined #openstack-swift | 14:40 | |
| *** agentleone has quit IRC | 14:42 | |
| *** haomai___ has joined #openstack-swift | 14:43 | |
| *** haomaiwa_ has quit IRC | 14:46 | |
| *** thiago has joined #openstack-swift | 14:51 | |
| *** thiago is now known as Guest67341 | 14:51 | |
| *** agentleone has joined #openstack-swift | 14:53 | |
| *** Guest67341 has quit IRC | 14:55 | |
| openstackgerrit | Alistair Coles proposed a change to openstack/swift: Restrict keystone cross-tenant ACLs to IDs https://review.openstack.org/86430 | 14:59 |
| *** Krazy has joined #openstack-swift | 15:01 | |
| Krazy | what is the recommended backend for storing keystone token? mysql or memcached? | 15:01 |
| *** tdasilva has left #openstack-swift | 15:04 | |
| *** chandan_kumar has joined #openstack-swift | 15:04 | |
| *** tdasilva_ has joined #openstack-swift | 15:04 | |
| *** nacim has joined #openstack-swift | 15:05 | |
| *** uksysadmin has quit IRC | 15:05 | |
| *** uksysadmin has joined #openstack-swift | 15:06 | |
| *** kopparam has quit IRC | 15:08 | |
| *** kopparam has joined #openstack-swift | 15:09 | |
| *** kopparam has quit IRC | 15:13 | |
| *** tdasilva_ is now known as tdasilva | 15:15 | |
| *** zhiyan is now known as zhiyan_ | 15:32 | |
| *** uksysadmin has quit IRC | 15:32 | |
| *** mlipchuk has quit IRC | 15:43 | |
| *** nacim has quit IRC | 15:49 | |
| *** judd7 has quit IRC | 15:52 | |
| *** agentleone has quit IRC | 15:53 | |
| acoles | Krazy: you may be best off asking in #openstack-keystone | 15:56 |
| clayg | finished sifting through review comments from yesterday | 15:57 |
| Krazy | cool thanks | 16:03 |
| *** Krazy has left #openstack-swift | 16:03 | |
| *** scohen_ has quit IRC | 16:03 | |
| peluse_ | clayg: awesome, there's a few more you up there as I'm sure you've noticed :) | 16:04 |
| clayg | peluse_: so the final patch you came up with for the db/rsync tmpdir thing was on gerrit or email? | 16:09 |
| clayg | peluse_: i don't get it, containers don't use per-policy tmp dirs... | 16:14 |
| *** joeljwright1 has left #openstack-swift | 16:14 | |
| clayg | oh well, i'll look at it when i get in, bbiab | 16:15 |
| peluse_ | clayg: gerrit comment but I also sent patch via email. You'll get it when you look at it... | 16:16 |
| *** foexle has quit IRC | 16:19 | |
| *** gyee has joined #openstack-swift | 16:20 | |
| *** bkopilov has joined #openstack-swift | 16:29 | |
| *** Midnightmyth has joined #openstack-swift | 16:31 | |
| *** agentleone has joined #openstack-swift | 16:38 | |
| *** pberis has quit IRC | 16:42 | |
| *** zaitcev has joined #openstack-swift | 16:47 | |
| *** ChanServ sets mode: +v zaitcev | 16:47 | |
| *** pberis has joined #openstack-swift | 16:48 | |
| zaitcev | clayg, peluse_: What's the best way to deliver fix-ups for SP? I thought I better show something rather than carping in comments. | 16:56 |
| zaitcev | I mean re-upload a patch in the middle of the chain or tuck a new one at the back? | 16:57 |
| peluse_ | zaitcev: clayg has preferred sending him stuff or comments (but not pushing directly) to make management of the chain, well, manageable. | 16:58 |
| zaitcev | peluse_: got it | 16:58 |
| *** agentleone has quit IRC | 16:59 | |
| *** agentleone has joined #openstack-swift | 17:02 | |
| *** agentleone has quit IRC | 17:07 | |
| *** shri has joined #openstack-swift | 17:08 | |
| *** nsquare has joined #openstack-swift | 17:13 | |
| *** mmcardle has quit IRC | 17:22 | |
| *** kopparam has joined #openstack-swift | 17:27 | |
| *** kevinc_ has quit IRC | 17:27 | |
| *** agentleone has joined #openstack-swift | 17:30 | |
| *** kevinc_ has joined #openstack-swift | 17:31 | |
| *** kopparam has quit IRC | 17:31 | |
| *** agentleone has quit IRC | 17:31 | |
| *** agentleone has joined #openstack-swift | 17:32 | |
| *** agentleone has quit IRC | 17:33 | |
| *** agentleone has joined #openstack-swift | 17:33 | |
| *** agentleone has quit IRC | 17:38 | |
| *** foexle has joined #openstack-swift | 17:44 | |
| clayg | peluse_: i only really see that the patch goes out of it's way not to delete some dirs on the receiving end - but that doesn't seem like something that should be required | 17:55 |
| clayg | peluse_: if you stuff an empty disk back into a container server replication should do what it needs to do receive data from the other container nodes | 17:55 |
| clayg | zaitcev: something that applies on 96026 would be ideal | 17:56 |
| zaitcev | clayg: will do | 17:56 |
| *** chuck__ has joined #openstack-swift | 18:25 | |
| *** zul has quit IRC | 18:25 | |
| *** chuck__ has quit IRC | 18:26 | |
| *** zul has joined #openstack-swift | 18:26 | |
| *** sandywalsh has quit IRC | 18:30 | |
| *** grapsus_ has quit IRC | 18:30 | |
| *** gholt has joined #openstack-swift | 18:32 | |
| *** ChanServ sets mode: +v gholt | 18:32 | |
| *** sandywalsh has joined #openstack-swift | 18:32 | |
| *** agentleone has joined #openstack-swift | 18:32 | |
| *** agentleone has quit IRC | 18:37 | |
| *** foexle has quit IRC | 18:37 | |
| *** agentleone has joined #openstack-swift | 18:39 | |
| *** agentleone has quit IRC | 18:40 | |
| *** mkollaro has quit IRC | 18:46 | |
| peluse_ | clayg: the problem is that test deletes the entire tree from suffix dir on down and then compares it to what it was after replciation runs. Any dir other than accounts, conainters, objects-n or tmp that doesn't exist after repl runs will cause the assert. The only reason tmp dir doesn't is because its created by the db repl | 18:46 |
| peluse_ | clayg: so yo could eiher create tmp-N in the repl every time, which makes little sense as its only needed to pass a probe test, or you can modify the test not to epxect dirs that have no reason to get created after wiping the tree out to somehow get created again... | 18:47 |
| *** CaioBrentano has quit IRC | 18:48 | |
| *** mkollaro has joined #openstack-swift | 18:48 | |
| clayg | peluse_: ok, I think i'm getting closer, but I like the part where the test delete's the entire tree from suffix on down - and the part where it validates the directories are what it expects | 18:50 |
| clayg | so why doesn't it just expect the tmp-N dir? Any why do only probetests need the tmp-N dir? | 18:50 |
| peluse_ | clayg: oops, not suffix dir, node level dir or whatever we call the level where accounts/containers/object--n live :) Also the repl isn't recreating those others either, the test blindlyh waits 60 secs so the other services are re-creating their dirs respectively... | 18:57 |
| peluse_ | clayg: it doesn't expect tmp-N because rsync doesn't create it becuase it doesn't use diskFile, ssync creates it via DiskFile | 18:57 |
| *** annegentle has joined #openstack-swift | 18:57 | |
| clayg | so the test only fails with ssync? ssync always fails probetests for me because of the x-static-large-object thing... | 18:58 |
| peluse_ | hmmm, the specific test that failed for me only failed for rsync, passes for ssync | 18:58 |
| clayg | peluse_: i'm sure it will all be clear once I manged to duplicate it - i'm only half way paying attention to the other window thats running probetests (first it failed because of ssync, then passed because default was 0, maybe this time...) | 18:58 |
| clayg | peluse_: yeah ok, i don't think that test uses object metadata | 18:59 |
| peluse_ | :) cool. Here's the test that fails for me: test_replication_servers_working.py:TestReplicatorFunctions.test_main | 18:59 |
| clayg | k | 18:59 |
| peluse_ | driving into work so will be offline for 45 min or so... | 18:59 |
| * peluse_ is getting off to a late start today :) | 18:59 | |
| *** annegentle has quit IRC | 18:59 | |
| *** tdasilva_ has joined #openstack-swift | 19:04 | |
| *** tdasilva has quit IRC | 19:05 | |
| *** foexle has joined #openstack-swift | 19:06 | |
| *** tdasilva_ is now known as tdasilva | 19:09 | |
| *** foexle has quit IRC | 19:13 | |
| *** DisneyRicky has quit IRC | 19:14 | |
| *** otoolee has quit IRC | 19:15 | |
| *** gholt has quit IRC | 19:23 | |
| *** gholt has joined #openstack-swift | 19:24 | |
| *** ChanServ sets mode: +v gholt | 19:24 | |
| *** kopparam has joined #openstack-swift | 19:30 | |
| notmyname | good morning, everyone. I was in an interesting meeting all morning, but I'm online now | 19:33 |
| notmyname | clayg: welcome back | 19:34 |
| *** kopparam has quit IRC | 19:34 | |
| *** DisneyRicky has joined #openstack-swift | 19:36 | |
| *** ahale has joined #openstack-swift | 19:41 | |
| *** Midnightmyth has quit IRC | 19:43 | |
| *** gyee has quit IRC | 19:43 | |
| peluse_ | notmyname: morning, anything interesting you can share? | 19:44 |
| clayg | notmyname: hi | 19:47 |
| notmyname | peluse_: in general, it's all about getting very dense and efficient storage. there's a lot of parts to that, of course, but some of the maybe-not-so-crazy crazy ideas to explore have to do with exploring RDMA | 19:48 |
| ahale | i was hoping you had typo'd rDNA, as that would be really awesome but I guess you didn't :( | 19:52 |
| notmyname | ahale: now I'm imagining creating synthetic organisms with a swift cluster (hey! it's already used to store genenome sequences) | 19:53 |
| portante | notmyname: cool | 19:54 |
| notmyname | ahale: imagine if you had a high-powered CPU that could deal with replication for a whole zone of object servers. especially if it could use RDMA to get the NIC to talk directly to the drive without needing to do even copy-to-kernel memory management or spend CPU cycles on it | 19:55 |
| notmyname | just some long-term stuff to thing about | 19:55 |
| portante | clayg: have you seen the comment I made about the container code using "is_deleted, then a sql statement, then another is_deleted call, then the update_change_at thingy" | 19:55 |
| portante | thinging about it now ... | 19:55 |
| ahale | notmyname: pure, awesome, scifi stuff | 19:56 |
| notmyname | ahale: or to rephrase, Swift being awesomer :-) | 19:56 |
| portante | when swift is that awesome we won't need spelling anymore | 19:57 |
| ahale | you say that but imagine the headache multi galactic replication would be over multi-region | 19:57 |
| portante | ahale: all we need is that flux capacitor ... | 19:58 |
| notmyname | ahale: ya, there may also need to be a few improvements there :-) | 19:58 |
| ahale | heheh | 19:58 |
| *** annegentle has joined #openstack-swift | 19:58 | |
| notmyname | but kota has all that figured out already :-) | 19:58 |
| ahale | i look forward to that! :) | 19:58 |
| peluse_ | notmyname: that meeting wasn't with some of our guys was it?? | 20:00 |
| portante | I think he had his people talk to their people | 20:01 |
| notmyname | peluse_: no, but I wouldn't be surprised if y'all are tangentally involved | 20:01 |
| peluse_ | notmyname: heh, its something we need to talk about for sure | 20:02 |
| peluse_ | portante: check your IRC pm | 20:02 |
| portante | IRC pm? | 20:03 |
| peluse_ | private messsage / chat whateever its called | 20:03 |
| notmyname | portante: I think he means check it tomorrow afternoon | 20:03 |
| peluse_ | huh? no, trying to ping portante now and not sure its working | 20:04 |
| notmyname | lol | 20:04 |
| portante | :) | 20:04 |
| notmyname | peluse_: maybe he doesn't want to talk to you ;-) | 20:04 |
| portante | I use emacs' "erc" client, and that appears to be a bit challenged in this area | 20:04 |
| peluse_ | story of my life :) | 20:05 |
| portante | try using "/query portante" and see if that works | 20:05 |
| portante | everybody is well to do so as well, don't want to exclude anyone. :) | 20:05 |
| notmyname | so what's up with storage policies today? will we be able to merge it this evening? | 20:08 |
| portante | I am not comfortable with the container sql query stuff | 20:08 |
| clayg | portante: sure i saw that | 20:10 |
| zaitcev | portante: in theory I'm working on that is_deleted and there's related thing I want to fix TODAY | 20:10 |
| portante | k | 20:10 |
| *** elmiko has joined #openstack-swift | 20:17 | |
| elmiko | hey all, i've got a question or two about the TempAuth middleware. is that just for testing? | 20:17 |
| *** wer has quit IRC | 20:19 | |
| *** AbyssOne_ has quit IRC | 20:19 | |
| *** torgomatic has quit IRC | 20:19 | |
| *** hurricanerix has quit IRC | 20:19 | |
| *** jd__ has quit IRC | 20:19 | |
| *** fbo_away has quit IRC | 20:19 | |
| *** portante has quit IRC | 20:19 | |
| *** russell_h has quit IRC | 20:19 | |
| *** briancline has quit IRC | 20:19 | |
| *** zaitcev has quit IRC | 20:19 | |
| *** pandemicsyn has quit IRC | 20:19 | |
| *** sileht has quit IRC | 20:19 | |
| *** mlanner has quit IRC | 20:19 | |
| *** occupant has quit IRC | 20:19 | |
| *** zacksh has quit IRC | 20:19 | |
| *** mhu has quit IRC | 20:19 | |
| *** bgmccollum has quit IRC | 20:19 | |
| *** zigo has quit IRC | 20:19 | |
| *** notmyname has quit IRC | 20:19 | |
| *** zhiyan_ has quit IRC | 20:19 | |
| *** chmouel has quit IRC | 20:19 | |
| *** anderstj has quit IRC | 20:19 | |
| *** joearnold has quit IRC | 20:19 | |
| *** glange has quit IRC | 20:19 | |
| *** sombrafam has quit IRC | 20:19 | |
| *** zaitcev has joined #openstack-swift | 20:20 | |
| *** wer has joined #openstack-swift | 20:20 | |
| *** AbyssOne_ has joined #openstack-swift | 20:20 | |
| *** torgomatic has joined #openstack-swift | 20:20 | |
| *** pandemicsyn has joined #openstack-swift | 20:20 | |
| *** sileht has joined #openstack-swift | 20:20 | |
| *** hurricanerix has joined #openstack-swift | 20:20 | |
| *** mlanner has joined #openstack-swift | 20:20 | |
| *** fbo_away has joined #openstack-swift | 20:20 | |
| *** jd__ has joined #openstack-swift | 20:20 | |
| *** portante has joined #openstack-swift | 20:20 | |
| *** occupant has joined #openstack-swift | 20:20 | |
| *** russell_h has joined #openstack-swift | 20:20 | |
| *** briancline has joined #openstack-swift | 20:20 | |
| *** joearnold has joined #openstack-swift | 20:20 | |
| *** glange has joined #openstack-swift | 20:20 | |
| *** dickson.freenode.net sets mode: +vvvv zaitcev torgomatic portante glange | 20:20 | |
| *** anderstj has joined #openstack-swift | 20:20 | |
| *** chmouel has joined #openstack-swift | 20:20 | |
| *** notmyname has joined #openstack-swift | 20:20 | |
| *** zigo has joined #openstack-swift | 20:20 | |
| *** zhiyan_ has joined #openstack-swift | 20:20 | |
| *** bgmccollum has joined #openstack-swift | 20:20 | |
| *** mhu has joined #openstack-swift | 20:20 | |
| *** zacksh has joined #openstack-swift | 20:20 | |
| *** dickson.freenode.net sets mode: +vv chmouel notmyname | 20:20 | |
| portante | elmiko: yes | 20:21 |
| portante | there are brave souls that use it in production | 20:21 |
| portante | at least, I have heard of a few in the past | 20:21 |
| elmiko | portante: ok, thanks. i was having a little trouble understanding the documentation but that clears it up :) | 20:21 |
| *** Midnightmyth has joined #openstack-swift | 20:23 | |
| creiht | yes please only use it for testing :) | 20:24 |
| *** pberis has quit IRC | 20:24 | |
| *** chandan_kumar has quit IRC | 20:24 | |
| *** rieglflo has quit IRC | 20:24 | |
| elmiko | creiht: i had a basic flaw in my understanding of how it worked. i thought it might be akin to a TempURL for an object. | 20:25 |
| clayg | creiht: well even then really only for development testing - it suck to bechmark your storage cluster then turn on auth for realzy and find out token validation is eating 10% of your request rate (or worse) | 20:25 |
| creiht | clayg: indeed | 20:26 |
| *** chandan_kumar has joined #openstack-swift | 20:33 | |
| *** pberis has joined #openstack-swift | 20:36 | |
| *** rieglflo has joined #openstack-swift | 20:36 | |
| *** Midnightmyth has quit IRC | 20:54 | |
| *** tdasilva has quit IRC | 21:02 | |
| *** gyee has joined #openstack-swift | 21:04 | |
| *** lpabon has quit IRC | 21:05 | |
| openstackgerrit | OpenStack Proposal Bot proposed a change to openstack/python-swiftclient: Updated from global requirements https://review.openstack.org/89250 | 21:06 |
| *** scohen has joined #openstack-swift | 21:08 | |
| *** scohen has quit IRC | 21:09 | |
| *** annegentle has quit IRC | 21:13 | |
| *** foexle has joined #openstack-swift | 21:23 | |
| *** elmiko is now known as _elmiko | 21:27 | |
| clayg | torgomatic: the line in the release notes to "turn on v2 endpoints responses" is sorta opposed to the comment "make v2 the default" | 21:28 |
| torgomatic | clayg: yeah, that's cause I'm stupid. I meant "turn off v2 endpoints responses" | 21:28 |
| torgomatic | so on by default with a note that says "if you're actually using this thing, set this variable to keep it working like before" | 21:29 |
| clayg | torgomatic: i guess what I sorta think I should do is make /endpoints/v1 and /endpoints/v2 both work then the config option is which one you get if you just say /endpoints | 21:29 |
| *** _elmiko has left #openstack-swift | 21:29 | |
| clayg | torgomatic: where "keep it working like before" ~= "doesn't provide all needed info on to work with storage policies" | 21:30 |
| torgomatic | clayg: yup, so it'll only work (i.e. provide all you need) if you haven't defined new storage policies yet | 21:30 |
| clayg | torgomatic: but that sounds like an argument to not go changing the default response unless we have to? | 21:32 |
| *** kopparam has joined #openstack-swift | 21:33 | |
| torgomatic | clayg: it's a tradeoff; if we set v1 by default, then upgrades are painless | 21:33 |
| torgomatic | if we set v2 by default, then new installs are painless | 21:33 |
| *** kevinc_ has quit IRC | 21:33 | |
| torgomatic | and there's a lot more Swift installs in the future than there are today | 21:36 |
| Alex_Gaynor | So teh swift docs say, about DLO: "It is best to upload all the segments first and then create or update the manifest. In this way, the full object won’t be available for downloading until the upload is complete." | 21:36 |
| Alex_Gaynor | If I /want/ the object to be downloadable when only some of the segments are uploaded, is ti fine to create the manifest and then start uploading segments? | 21:36 |
| *** mkollaro has quit IRC | 21:36 | |
| torgomatic | Alex_Gaynor: yeah, it won't break anything | 21:36 |
| Alex_Gaynor | torgomatic: perfect, thanks | 21:36 |
| notmyname | Alex_Gaynor: yes. but in both cases you are subject to the container listings | 21:36 |
| Alex_Gaynor | notmyname: meaning what? | 21:37 |
| torgomatic | it's not like Swift can tell when you're done uploading segments unless you've got it hooked up to a crystal ball ;) | 21:37 |
| Alex_Gaynor | Oh. Yeah, that's fine, that'll be the applications responsibility in my case | 21:38 |
| notmyname | Alex_Gaynor: if the container listings are taking a long time to be updated (eg in large or busy containers) the DLO manifest won't "work" until the listing is up to date. and that can take a while | 21:38 |
| *** kopparam has quit IRC | 21:38 | |
| clayg | torgomatic: well I think either default is fine if there's a way for the client to decide when it wants the storage_policy_index key - but I couldn't think of an obviously correct way to decide if /endpoints/v2.0 was asking for a v2 response or the account named "v2.0" | 21:38 |
| Alex_Gaynor | meaning, if I'm building a multi-tenanted system I probably want to give each tenant it's own container? | 21:38 |
| torgomatic | clayg: oh yeah, it's not just objects is it... | 21:39 |
| notmyname | Alex_Gaynor: is there a reason you wouldn't give each tenant it's own swift account? | 21:39 |
| Alex_Gaynor | notmyname: because I'm deploying this on top of an existing swift deployment in a large public cloud | 21:39 |
| notmyname | clayg: torgomatic: v1_endpoints/ and /endpoints | 21:39 |
| notmyname | ? | 21:39 |
| torgomatic | notmyname: maybe? now it's even more namespace devoted to this little niche middleware | 21:40 |
| notmyname | Alex_Gaynor: it's always a good idea to use many containers when storing application data in swift | 21:40 |
| notmyname | torgomatic: I'm jsut throwing stuff on the wall :-) | 21:40 |
| torgomatic | :) | 21:40 |
| notmyname | AFAIK zerovm are the only ones to use the endpoints middleware. I haven't seen anyone else talking about it. of course, somebody probably is somewhere | 21:41 |
| torgomatic | notmyname: good point; any zerovm guys in here? | 21:43 |
| notmyname | not that are speaking, and the #zerovm channel is very quiet | 21:43 |
| *** kevinc_ has joined #openstack-swift | 21:43 | |
| torgomatic | well, so much for just asking them | 21:43 |
| *** nsquare has quit IRC | 21:44 | |
| notmyname | but also looking at a swift release being with SP and being a major version bump, I'm much more ok with doing something backwards incompatible than eg on a 2.0->2.1 release or 1.12->1.13 release | 21:45 |
| *** nsquare has joined #openstack-swift | 21:46 | |
| clayg | notmyname: torgomatic: I don't think it needs to be backwards incompatible if I can come up with a good path parsing schema - would be trivial if it was versioned from the start... | 21:46 |
| clayg | maybe the option is just path_has_version and if you say no then you get v1 responses, if you say yes then you clients can either add /v1/ or parse 'endpoints' out of the dict | 21:47 |
| clayg | still no good way to update cluster and clients except at the same time | 21:48 |
| clayg | bah | 21:48 |
| clayg | I think some sort of goofy account_part_looks_like_a_version_if_you_squint method would probably good enough | 21:49 |
| notmyname | clayg: what about checking if the first split part has a reseller. or more simply, an underscore? | 21:50 |
| *** pberis has quit IRC | 21:54 | |
| peluse_ | notmyname: I'm still learning to speak clayg but I kinda thought that's what he meant :) | 21:54 |
| notmyname | peluse_: ya, he's a pretty smart guy | 21:54 |
| clayg | notmyname: well you'd either have to make list endpoints take a configurable list of reseller bits but underscore might work | 21:56 |
| notmyname | clayg: I think that's where the _if_you_squint part of your method comes in ;-) | 21:57 |
| clayg | notmyname: alright, that'll probably work | 21:59 |
| portante | notmyname, creiht: your favorite topic, pbr: https://bugs.launchpad.net/keystone/+bug/1330771 | 22:25 |
| notmyname | portante: oh yes. that was a fun thing to read this morning. (turns out I'm subscribed to all things PBR in LP) | 22:26 |
| torgomatic | portante: notmyname: does this mean swift devs are Thought Leaders(tm)? | 22:30 |
| clayg | torgomatic: so the object server was already using per-policy tmpdirs in the rsync commands? | 22:31 |
| torgomatic | clayg: I think so? | 22:31 |
| clayg | heh | 22:32 |
| torgomatic | like I said, perhaps in the vicinity of a fix :) | 22:32 |
| torgomatic | clayg: although now that I look at it more, it seems like rsync drops temp files in the same dir as the destination, at least by default | 22:36 |
| clayg | torgomatic: idk, why would that be there in the replicator instead of the obj server? | 22:36 |
| torgomatic | and I don't see us setting -T or --temp-dir anywhere in obj/replicator.py | 22:36 |
| torgomatic | or the obj server | 22:36 |
| clayg | torgomatic: yeah i'm applying the patch and seeing if it fixes the failure OMM too | 22:37 |
| torgomatic | this patch may be entirely the wrong way to go; it might be the case that enforcing the presence of a tmp-1 directory is the wrong thing for the probe test to do | 22:38 |
| torgomatic | and I'm just making a bogus directory to satisfy a bogus check | 22:39 |
| clayg | torgomatic: is that all peluse_'s change did? was make the test not give a shit if rsync needs a tmp dir to sync objects or not? | 22:39 |
| torgomatic | clayg: I think so; let me go read it again | 22:39 |
| torgomatic | clayg: no, his patch changes the setup from nuking /srv/node/sdblah/* to nuking only /srv/node/sdblah/(accounts|containers|objects) | 22:41 |
| torgomatic | which would have the effect of leaving tmp-1 around so the later assertion is satisfied | 22:41 |
| clayg | torgomatic: looks like there's a bit before the delete that gets a list of shit on the existing node, and it explicitly excludes .pending files cause it don't care if the recieving end needs those - probably do the same for tmpdirs and call it a day | 22:43 |
| torgomatic | clayg: yeah, that sounds like a better fix | 22:43 |
| *** nsquare has quit IRC | 22:44 | |
| *** nsquare has joined #openstack-swift | 22:45 | |
| clayg | torgomatic: can you test this one -> https://gist.github.com/clayg/858efb1f472d72a472cc | 22:46 |
| clayg | torgomatic: WOMM | 22:46 |
| torgomatic | clayg: heh, I had almost that exact same diff written out over here, only I had a call to os.path.basename that you don't :) | 22:48 |
| * torgomatic goes to try clayg's patch | 22:48 | |
| torgomatic | clayg: works for me. Ship it! | 22:50 |
| clayg | torgomatic: oh i think i like the basename bit | 22:51 |
| clayg | torgomatic: oh well i guess that's from os.walk so they're not really full paths | 22:51 |
| torgomatic | clayg: yeah, exactly. I wasn't sure, so I put it in there out of paranoia with a note to see if it was really necessary | 22:52 |
| torgomatic | turns out it's not | 22:52 |
| * clayg is really glad this wasn't a 4th bug related to the per-policy tmpdirs | 22:52 | |
| torgomatic | seriously | 22:52 |
| *** kevinc_ has quit IRC | 22:56 | |
| *** foexle has quit IRC | 23:03 | |
| *** kevinc_ has joined #openstack-swift | 23:05 | |
| mattoliverau | Morning all | 23:07 |
| *** scohen has joined #openstack-swift | 23:10 | |
| *** scohen has quit IRC | 23:11 | |
| shri | hey all… have a quick question. Is there any such thing as Swift regions *without* keystone? | 23:13 |
| torgomatic | shri: Swift regions and Keystone regions have nothing to do with one another, so yes | 23:15 |
| peluse_ | clayg: just looked at the tmp dir patch above, looks good to me too! | 23:15 |
| shri | torgomatic : Thanks! | 23:15 |
| torgomatic | np | 23:15 |
| *** kopparam has joined #openstack-swift | 23:37 | |
| torgomatic | is there a way to ask Gerrit for patches that I haven't reviewed the latest change set on? | 23:40 |
| torgomatic | I found reviewer:self, but that seems to apply to any change set | 23:41 |
| *** kopparam has quit IRC | 23:43 | |
| *** openfly has quit IRC | 23:44 | |
| mattoliverau | torgomatic: I use a search string: status:open project:openstack/swift NOT label:Code-Review>=0 label:Verified>=1,jenkins NOT label:Code-Review<=-1 NOT label:Workflow<=-1 | 23:48 |
| mattoliverau | You can see if that helps, I have another more detailed for infra reviews, if you want that. :) | 23:49 |
| torgomatic | mattoliverau: thanks, I'll play with those search params | 23:49 |
| mattoliverau | torgomatic: this might be useful, its a blog entry written by Sean: http://is.gd/vN5k1q | 23:57 |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!