| *** jamielennox is now known as jamielennox|away | 00:00 | |
| timburke | clayg: well, transparent in the sense that post-as-copy/fast-post is transparent. you'll definitely know which mode you're in, but with sufficient timeouts, you shouldn't really care | 00:01 |
|---|---|---|
| timburke | x-versions-location & x-better-versions-location | 00:01 |
| *** tongli has quit IRC | 00:02 | |
| clayg | timburke: oh interesting - i tought something about POSTing somewhere would make something a difference? | 00:02 |
| clayg | but versions + metadata updates makes my head hurt anyway | 00:03 |
| *** Suyi_ has quit IRC | 00:03 | |
| timburke | clayg: ok, jrichli pointed out earlier today that one of them sends back a body on POST while the other doesn't | 00:03 |
| timburke | metadata updates aren't versioned! it's great! | 00:04 |
| clayg | timburke: oh yeah that's right - someone was pointing that body/content difference - i thought that was you? | 00:08 |
| *** jamielennox|away is now known as jamielennox | 00:14 | |
| openstackgerrit | Darrell Bishop proposed openstack/swift: Add support for PROXY protocol v1 (only) https://review.openstack.org/373563 | 00:22 |
| *** tqtran has quit IRC | 00:25 | |
| tdasilva | hi clayg, timburke i'm back | 00:38 |
| timburke | yay, more opinions! | 00:38 |
| tdasilva | i just read the scrollback really quickly and my 2 cents on x-stack-container is that i don't see the point of introducing thew new header. the problem is not with the current header name per se, but the functionality it provides. so (IMO) when we talk of deprecating x-versions-location | 00:40 |
| tdasilva | we are talking about deprecating the functionality, not just the header | 00:40 |
| tdasilva | so if we ever were to deprecate x-version-location, we would also deprecate x-stack-container... | 00:40 |
| clayg | tdasilva: oh that's a good point too - the only thing i hate more than the name of the header is the stack based functionality | 00:42 |
| clayg | but if the name was better I might forget how much I hate it? | 00:43 |
| tdasilva | clayg: the whole point of introducing history mode was that it was better functionality | 00:44 |
| clayg | tdasilva: yeah it is | 00:45 |
| kota_ | good morning | 01:02 |
| mattoliverau | kota_: o/ | 01:03 |
| kota_ | mattoliverau: o/ | 01:03 |
| timburke | tdasilva: k. so x-versions-location / x-history-container, or x-versions-location / x-history-location? | 01:04 |
| tdasilva | x-history-location??? | 01:07 |
| tdasilva | don't have a really strong opinion on that thou | 01:07 |
| timburke | for consistency with x-versions-location. should i take that as a vote to stick with "container"? | 01:08 |
| tdasilva | timburke: not sure i understand your question :( | 01:09 |
| * mattoliverau has been in a meeting and now has to step out.. bbs | 01:10 | |
| timburke | you seemed incredulous at x-history-location. or more likely, subtlety is lost in irc ;-) | 01:10 |
| tdasilva | timburke: what's the chance we can ever get to a point where users don't know about the container where archives are stored? | 01:10 |
| tdasilva | sorry...the question marks was more like "i think x-history-location is better, but really don't have a strong opinion" | 01:11 |
| timburke | yep, got it. thanks | 01:11 |
| timburke | on never knowing/thinking about the archive container, it'll probably make for some unhappy users who aren't sure where their usage went... | 01:12 |
| tdasilva | well, they would have a way of listing versions | 01:12 |
| tdasilva | i'm thinking something more like s3 | 01:12 |
| tdasilva | where you just enable versioning on a bucket | 01:12 |
| timburke | i know you are :-) | 01:13 |
| tdasilva | but don't specify where archives get stored | 01:13 |
| tdasilva | heh | 01:13 |
| timburke | and oh man am i gonna have fun with it in swift3 | 01:13 |
| *** McMurlock1 has quit IRC | 01:13 | |
| tdasilva | right, so if you are planning on support that in swift3, would it make sense to have the same in swift? | 01:14 |
| timburke | but swift allows (nearly) any utf-8 as a container name, so it'll be tricky to hide | 01:14 |
| tdasilva | i mean, is it a better API than what we have? honest question | 01:14 |
| timburke | generally? they're just different. i really like the power and reusability of SLOs; S3 can't do that. i rather like how much less you have to think about as a client for S3, though | 01:16 |
| timburke | for this in particular, i could see some value in having some automated script that rotates out the archive container monthly or something. then you could change who has permission to what based on date range | 01:17 |
| clayg | timburke: if you know of a valid utf-8 container name that we can scope to a clints account - but you can't acctually create via the API - i want to know about it | 01:20 |
| timburke | "a valid ... container name ... you can't actually create via the API" seems like a contradiction | 01:21 |
| clayg | like maybe there's some magic way we could create a container with a "/" *in* the name and still have a world that makes sense? because god oh dear god how many times I've wished we could carve out "dot containers" like we did "dot accounts" | 01:22 |
| clayg | oh I guess you just %2F - stupid urlencoding thought of everything | 01:23 |
| timburke | yeah, i was thinking about that, too. /v1/a/c%2fstill-part-of-the-container/o would be nice, but we unquote() everything pretty early on as i recall | 01:23 |
| timburke | and we'd be back to wondering what clients we're gonna break | 01:24 |
| * kota_ is catching up the conversation for new versioned-write namespaces | 01:26 | |
| kota_ | tdasilva: AFAIK, swift3 is using "+" for the versions-location container because "+" is not allowed for the bucket name in actual S3 | 01:29 |
| timburke | yup | 01:29 |
| kota_ | tdasilva: something like "original_container+versioned" | 01:29 |
| kota_ | timburke: looking at the back log, we cannot set both x-history-location and x-stack-location at a container, right? | 01:30 |
| timburke | and there's a similar thing with = in header keys that i'm using to do https://review.openstack.org/#/c/345739/ | 01:30 |
| patchbot | patch 345739 - swift3 - Add support for more characters in header keys | 01:30 |
| kota_ | timburke: I'm wondering how we can resolve the not allowed header conbination in the split brain case eventually. | 01:30 |
| timburke | kota_: correct. it will 400 if you try to set both in one request; in separate requests, last write wins | 01:30 |
| kota_ | timburke: currently, we are using same header key (i.e. x-versions-location) so that it will be resolved one of them. | 01:31 |
| clayg | kota_: because they both resolve to x-versions-location in the data model the last-write-wins is enforced in the consistency engine with no extra code (nice work there timburke !) | 01:31 |
| kota_ | mode, too. | 01:31 |
| clayg | kota_: yup timburke is only changing the presentation layer - data model stays the same (it's a good data model!) | 01:32 |
| kota_ | clayg: gotcha!, nice work, I didn't see the patch yet, just wondering from conversation | 01:32 |
| kota_ | clayg: thanks | 01:32 |
| tdasilva | so is the mode header also hidden from the user? | 01:33 |
| tdasilva | i also have not yet looked at the patch | 01:33 |
| tdasilva | :P | 01:33 |
| kota_ | tdasilva: :P | 01:33 |
| clayg | tdasilva: afaik it is simply not translated from it's sysmeta location | 01:34 |
| clayg | tdasilva: I'm mostly leaning towards x-versions-location and x-history-location (I think?) | 01:34 |
| clayg | acctually i'd be fine with x-versions-location and x-history-container | 01:35 |
| tdasilva | clayg: yeah, i think either works for me | 01:37 |
| MooingLemur | before I open a bug on this, was there a reason DLO and SLO do not take Referer auth into account? | 01:39 |
| MooingLemur | SLO fails to do the make_subrequest to fetch the manifest because there's no auth token for the request, and DLO has the same issue but this time it's a container list. | 01:40 |
| MooingLemur | oh well, bug filed. :) | 01:45 |
| clayg | HTTP_REFERER is in the make_env list? | 01:47 |
| clayg | i thought we even had a functest for public DLO's? | 01:47 |
| clayg | well public *lo's rather | 01:47 |
| MooingLemur | how about nonpublic ones with a specific referer? | 01:48 |
| clayg | i'm sure i'll see the lp bug in my inbox in a minute - i'll read it over make sure i'm understanding what you're saying is not working as expected | 01:48 |
| clayg | yeah i'd guess it's less likely we have a specific test for that - so much more possible it's broken - but in some non-obvious way since acl rules would seem to work for public containers and the referer header should be getting copied across to the sub requests | 01:49 |
| clayg | should be easy to duplicate and add a failing functional test for if needed | 01:50 |
| *** StraubTW has joined #openstack-swift | 01:50 | |
| jrichli | clayg: timburke had recently noted an inconsistency in having a body with the 416 case for multiget slos patch 352657 | 01:55 |
| patchbot | https://review.openstack.org/#/c/352657/ - swift - Support multi-range GETs for static large objects. | 01:55 |
| openstackgerrit | Tim Burke proposed openstack/swift: Use separate headers for versioned_writes' stack and history modes https://review.openstack.org/373537 | 01:59 |
| MooingLemur | clayg: I'm embarrassed to say it may have been fixed between 2.5.0 and 2.7.0. It appears to behave correctly with 2.7.0. | 01:59 |
| MooingLemur | without a patch | 01:59 |
| clayg | MooingLemur: you were asking for some other rsync fix to back ported too yeah? | 02:00 |
| clayg | MooingLemur: does this mean you're gunna get on the close to master train with the rest of us!? | 02:01 |
| MooingLemur | :) | 02:01 |
| clayg | MooingLemur: or just that you're going to continue to advocate for the changes that we should be backporting because we're obviously clueless :P | 02:01 |
| MooingLemur | In 2.7.0 I have the problem of object-replicator hanging. I haven't opened anything on that one yet :) | 02:01 |
| MooingLemur | however, I do have an open bug that I'm certain still exists in 2.9.0 :) | 02:02 |
| clayg | you mean thats a *new* problem you *didn't* have 2.5.0 (!?) | 02:02 |
| MooingLemur | yes | 02:03 |
| timburke | MooingLemur: your slo/dlo issue seems to have been fixed in 2.6.0 -- https://github.com/openstack/swift/commit/87f7e90 | 02:04 |
| MooingLemur | timburke: thanks :) | 02:04 |
| MooingLemur | https://bugs.launchpad.net/swift/+bug/1454433 | 02:04 |
| openstack | Launchpad bug 1454433 in OpenStack Object Storage (swift) "swift-ring-builder does not seem to properly parse setting weights for multiple devices at one time" [Undecided,New] | 02:04 |
| MooingLemur | that one appears to have happened when swift-ring-builder was being refactored to allow the "new" syntax | 02:05 |
| clayg | mattoliverau: you're such a champ! | 02:06 |
| MooingLemur | swift-ring-builder ... set_weight <spec1> <weight> <spec2> <weight> ... # for some reason it parses out spec1 and weight (expanding spec1 into as many devices as it points to), passes them into _set_weight_values, then on the second round of the loop, appends the devices from spec2 into the same devs array without clearing them, and then calls _set_weight_values again, but now it has all of th | 02:08 |
| MooingLemur | e devices from spec1 as well as spec2, ... | 02:09 |
| MooingLemur | ... and it tries to apply the weight of spec2 to all of it | 02:09 |
| tdasilva | good night | 02:10 |
| tdasilva | ccccccdkkgthegjdtedcelidgugkjrcluedgjllbeeht | 02:10 |
| MooingLemur | so the devs array gets longer and longer :) Correct behavior seems to be to assign the value to devs rather than devs.extend | 02:10 |
| MooingLemur | Hi, kitty. | 02:11 |
| kota_ | tdasilva: good night | 02:11 |
| MooingLemur | actually looks too keyboard mash to be cat typing. Also it contains "clue" :P | 02:12 |
| clayg | MooingLemur: it's probably one of those rsa security key thingy's | 02:15 |
| MooingLemur | clayg: http://www.thinkgeek.com/product/ae83/ | 02:15 |
| clayg | MooingLemur: and do you *want* this syntax? Maybe rather than trying to fix that mess which no one is using (successfully) we could take the oppertunity to make the interface closer to something that would be useful? What's your use-case? | 02:16 |
| clayg | MooingLemur: that is *fabulous* | 02:17 |
| kota_ | clayg: btw, could you keep attention for my global ec thing? | 02:19 |
| MooingLemur | clayg: I'm claiming the documented syntax: swift-ring-builder <builder_file> set_weight <search-value> <weight> [<search-value> <weight] ... is broken. It only works if you specify one search value and one weight. If you specify more than one, it ends up applying the weight of the final search value to *all* of the matched search values including the ones earlier on the command line. | 02:19 |
| kota_ | clayg: patch 219165 is now on the master and got +1 from Jenkins, I addressed your items at patch 343032 | 02:20 |
| patchbot | https://review.openstack.org/#/c/219165/ - swift - EC Fragment Duplication - Foundational Global EC C... | 02:20 |
| patchbot | https://review.openstack.org/#/c/343032/ - swift - Items for EC Duplication to consider | 02:20 |
| clayg | kota_: full ack - i'll loop back around to that | 02:21 |
| kota_ | clayg: ah, the converstaion is still progress, not urgent right now. just notification. | 02:21 |
| kota_ | clayg: thanks | 02:21 |
| clayg | MooingLemur: i believe you it's busted - the code there is fully of crufty jank - i'm just thinking the documented (and broken) synax is sort of not that great/useful to begin with | 02:21 |
| clayg | i don't want to fix it just because the docs say it work that way - esp if it's broken because it's so terrible to implement and also - not useful | 02:22 |
| MooingLemur | clayg: I can specify one search value and one weight per invocation of swift-ring-builder, however if I'm doing many (hundreds) of weight adjustments (while bringing on new storage), the amount of time to do that process is severely amplified. | 02:22 |
| clayg | right obviously - do you need explicit (device, weight) pairs or would [dev, dev, dev, dev] [weight] work in your use-case | 02:23 |
| MooingLemur | it's more useful to do the former, since the script that manages the slow reweighting already has knowledge of each device, its size, and current weight, but if I need to split out the invocations of swift-ring-builder so that I group them by weight assignment, I can do that | 02:25 |
| MooingLemur | but it seems like a silly limitation | 02:25 |
| *** _JZ_ has quit IRC | 02:25 | |
| clayg | is the script that manages the slow reweighting happen to be written in python by any chance? | 02:26 |
| MooingLemur | nope :) | 02:26 |
| clayg | one can dream | 02:26 |
| MooingLemur | perl | 02:26 |
| *** ChubYann has quit IRC | 02:30 | |
| *** dmorita has quit IRC | 02:40 | |
| *** ukaynar has joined #openstack-swift | 02:42 | |
| *** StraubTW has quit IRC | 02:53 | |
| mattoliverau | . | 02:54 |
| mattoliverau | clayg: I yeah, I remember that patch, yeah we fixed that referrer thing :) | 02:54 |
| *** takashi has joined #openstack-swift | 02:54 | |
| mattoliverau | also, for those of you who like me have to participate in many a video conference: https://www.youtube.com/watch?v=DYu_bGbZiiQ | 02:54 |
| *** dmorita has joined #openstack-swift | 02:56 | |
| *** dmorita has quit IRC | 03:01 | |
| *** winggundamth_ has quit IRC | 03:03 | |
| *** david-lyle has quit IRC | 03:04 | |
| clayg | mattoliverau: that's pretty good - tyler is my hero - esp when he's in the hallway talk to the wall | 03:04 |
| mattoliverau | lol yeah :) | 03:06 |
| jrichli | mattoliverau: love it! | 03:15 |
| *** vinsh has joined #openstack-swift | 03:32 | |
| *** chsc has joined #openstack-swift | 03:37 | |
| *** chsc has joined #openstack-swift | 03:37 | |
| *** klrmn has quit IRC | 03:47 | |
| *** chsc has quit IRC | 03:50 | |
| *** ukaynar has quit IRC | 03:55 | |
| *** ukaynar has joined #openstack-swift | 04:02 | |
| MooingLemur | clayg: says the developer to the end user "You're using that for WHAT?!" :) | 04:10 |
| MooingLemur | We encounter a little bit of that internally in our company from time to time. :) | 04:12 |
| *** ppai has joined #openstack-swift | 04:33 | |
| *** psachin has joined #openstack-swift | 04:38 | |
| *** vinsh has quit IRC | 04:39 | |
| *** psachin has quit IRC | 04:40 | |
| *** psachin has joined #openstack-swift | 04:41 | |
| zaitcev | well you saw that guy yesterday with Swift on top of ZFS | 04:44 |
| *** psachin has quit IRC | 04:45 | |
| zaitcev | Seems like one of the dumbest ideas ever on the surface of it, since you can't access on ZFS side anyway, without swift-on-file anyway | 04:45 |
| *** psachin has joined #openstack-swift | 04:47 | |
| *** ppai has quit IRC | 04:59 | |
| *** McMurlock1 has joined #openstack-swift | 05:00 | |
| *** McMurlock1 has quit IRC | 05:09 | |
| *** ppai has joined #openstack-swift | 05:12 | |
| *** asettle has joined #openstack-swift | 05:20 | |
| *** wuhg has joined #openstack-swift | 05:22 | |
| openstackgerrit | Matthew Oliver proposed openstack/swift: Use separate headers for versioned_writes' stack and history modes https://review.openstack.org/373537 | 05:23 |
| mattoliverau | timburke: ^^ Turns out the api-ref's parameters in parameters.yaml need to be in alphabetical order. I just fixed the ordering to Jenkins would pass. | 05:25 |
| *** tqtran has joined #openstack-swift | 05:25 | |
| *** asettle has quit IRC | 05:28 | |
| *** tqtran has quit IRC | 05:30 | |
| *** SkyRocknRoll has joined #openstack-swift | 05:38 | |
| clayg | timburke: lol - I just ran into this today -> https://bitbucket.org/ianb/pastedeploy/pull-requests/13/fix-string-formatting-args-when-raising/diff | 05:50 |
| *** Jeffrey4l has joined #openstack-swift | 05:51 | |
| clayg | ... the TypeError ... for which the fix is almost definately laying on my filesystem in saio - then later in an unrelated search I came across your patch! | 05:52 |
| *** tqtran has joined #openstack-swift | 05:53 | |
| openstackgerrit | OpenStack Proposal Bot proposed openstack/swift: Updated from global requirements https://review.openstack.org/88736 | 06:04 |
| *** rcernin has joined #openstack-swift | 06:07 | |
| *** ppai has quit IRC | 06:11 | |
| *** ukaynar has quit IRC | 06:33 | |
| openstackgerrit | Sharat Sharma proposed openstack/python-swiftclient: Replace MagicMock with Mock class https://review.openstack.org/373737 | 06:39 |
| *** ppai has joined #openstack-swift | 06:41 | |
| openstackgerrit | Merged openstack/swift: api-ref: fix some header definitions https://review.openstack.org/366232 | 06:44 |
| openstackgerrit | Clay Gerrard proposed openstack/swift: Make NullLogger more logger like https://review.openstack.org/373784 | 06:49 |
| *** kmARC has quit IRC | 06:49 | |
| openstackgerrit | OpenStack Proposal Bot proposed openstack/swift: Updated from global requirements https://review.openstack.org/88736 | 06:49 |
| *** hseipp has joined #openstack-swift | 06:55 | |
| *** kmARC has joined #openstack-swift | 07:00 | |
| clayg | do I tell them every with mock.patch() as returns a MagicMock? lp bug #1475722 | 07:05 |
| openstack | Launchpad bug 1475722 in python-neutronclient "Never use MagicMock" [Low,In progress] https://launchpad.net/bugs/1475722 - Assigned to lilintan (lilintan) | 07:05 |
| *** rledisez has joined #openstack-swift | 07:06 | |
| *** pcaruana has joined #openstack-swift | 07:13 | |
| *** tqtran has quit IRC | 07:19 | |
| *** sams-gleb has joined #openstack-swift | 07:23 | |
| *** silor has joined #openstack-swift | 07:27 | |
| openstackgerrit | Clay Gerrard proposed openstack/swift: Fix unstable json.dumps ordering in obj-server test https://review.openstack.org/373819 | 07:33 |
| *** dmorita has joined #openstack-swift | 07:41 | |
| *** geaaru has joined #openstack-swift | 07:43 | |
| *** dmorita has quit IRC | 07:46 | |
| openstackgerrit | zheng yin proposed openstack/python-swiftclient: remove swiftclient/openstack/common/* in the .coveragerc https://review.openstack.org/373846 | 07:53 |
| openstackgerrit | Kota Tsuyuzaki proposed openstack/swift: Support last modified on listing containers https://review.openstack.org/198634 | 07:56 |
| *** oshritf has joined #openstack-swift | 08:03 | |
| *** oshritf_ has joined #openstack-swift | 08:03 | |
| *** SkyRocknRoll has quit IRC | 08:10 | |
| openstackgerrit | Kota Tsuyuzaki proposed openstack/swift: Support last modified on listing containers https://review.openstack.org/198634 | 08:24 |
| kota_ | Updates my oldest. | 08:24 |
| *** SkyRocknRoll has joined #openstack-swift | 08:31 | |
| *** kei_yama has quit IRC | 08:43 | |
| *** asettle has joined #openstack-swift | 08:45 | |
| *** mvk has quit IRC | 09:00 | |
| *** acoles_ is now known as acoles | 09:02 | |
| psachin | Hello #openstack-swift, Deploying account and containers on separate servers for large infrastructure make sense? Is it possible? | 09:05 |
| *** mvk has joined #openstack-swift | 09:31 | |
| acoles | other cores: easy review and worth fixing before release - patch 373819 | 09:34 |
| patchbot | https://review.openstack.org/#/c/373819/ - swift - Fix unstable json.dumps ordering in obj-server test | 09:34 |
| ahale_ | psachin, its possible yes, and sometimes a good idea. we keep account and container together but apart from object | 09:41 |
| ahale_ | the acc/cont/obj rings are kind of independent , just as long as the ip's can all communicate | 09:44 |
| *** takashi has quit IRC | 09:50 | |
| *** SkyRocknRoll has quit IRC | 09:54 | |
| psachin | ahale_: what about the internal communication between account-container and object? Do you suggest dedicated network? | 09:55 |
| ahale_ | the proxy needs to be able to hit it all too so idk.. but a dedicated replication network can work. depends on your needs and scale i guess (i dont use dedicated network for different traffic but we do split replication and regular traffic to different listening processes on the same ip) | 09:58 |
| psachin | ahale_: Thanx for your suggestions | 10:06 |
| ahale_ | np, might get some other ideas later when all the US timezones wake up too :) | 10:06 |
| *** SkyRocknRoll has joined #openstack-swift | 10:10 | |
| *** tqtran has joined #openstack-swift | 10:17 | |
| *** dabukalam has quit IRC | 10:19 | |
| *** tqtran has quit IRC | 10:21 | |
| *** silor1 has joined #openstack-swift | 10:35 | |
| *** silor has quit IRC | 10:37 | |
| *** oshritf_ has left #openstack-swift | 10:39 | |
| *** silor has joined #openstack-swift | 10:40 | |
| *** silor1 has quit IRC | 10:40 | |
| openstackgerrit | Kota Tsuyuzaki proposed openstack/swift: Support last modified on listing containers https://review.openstack.org/198634 | 10:43 |
| *** dfflanders has quit IRC | 10:47 | |
| *** Jeffrey4l has quit IRC | 10:50 | |
| *** Jeffrey4l has joined #openstack-swift | 10:50 | |
| *** dabukalam has joined #openstack-swift | 11:37 | |
| *** daemontool has joined #openstack-swift | 11:39 | |
| *** dmorita has joined #openstack-swift | 11:43 | |
| *** dabukalam has quit IRC | 11:46 | |
| openstackgerrit | Alistair Coles proposed openstack/swift: Add more comment to authtoken sample options https://review.openstack.org/372573 | 11:46 |
| *** dmorita has quit IRC | 11:47 | |
| *** wuhg has quit IRC | 12:01 | |
| *** dabukalam_ has joined #openstack-swift | 12:07 | |
| *** dabukalam_ has quit IRC | 12:14 | |
| *** dabukalam_ has joined #openstack-swift | 12:18 | |
| openstackgerrit | Tuan Luong-Anh proposed openstack/swift: Fix using filter() to meet python2,3 https://review.openstack.org/368649 | 12:18 |
| *** dabukalam_ has quit IRC | 12:26 | |
| *** ppai has quit IRC | 12:41 | |
| *** david-lyle has joined #openstack-swift | 12:56 | |
| *** acoles is now known as acoles_ | 12:58 | |
| *** tgtanya has joined #openstack-swift | 12:59 | |
| *** SkyRocknRoll has quit IRC | 13:01 | |
| *** StraubTW has joined #openstack-swift | 13:02 | |
| *** SkyRocknRoll has joined #openstack-swift | 13:03 | |
| *** dabukalam has joined #openstack-swift | 13:06 | |
| *** dabukalam has quit IRC | 13:12 | |
| *** kei_yama has joined #openstack-swift | 13:14 | |
| *** klamath has joined #openstack-swift | 13:19 | |
| *** psachin has quit IRC | 13:21 | |
| *** dabukalam has joined #openstack-swift | 13:22 | |
| *** dabukalam has quit IRC | 13:23 | |
| *** psachin has joined #openstack-swift | 13:23 | |
| *** kei_yama has quit IRC | 13:25 | |
| *** vinsh has joined #openstack-swift | 13:37 | |
| jrichli | patch 357559 only needs one more +2 and is on our priority reviews list. 16 lines changed in 2 files. | 13:49 |
| patchbot | https://review.openstack.org/#/c/357559/ - swift - Include object sysmeta in POST responses | 13:49 |
| *** asettle has quit IRC | 13:54 | |
| *** tongli has joined #openstack-swift | 13:54 | |
| *** asettle has joined #openstack-swift | 13:55 | |
| *** vinsh has quit IRC | 13:56 | |
| *** lifeless has quit IRC | 13:57 | |
| *** lifeless has joined #openstack-swift | 13:59 | |
| *** acoles_ is now known as acoles | 14:03 | |
| *** chsc has joined #openstack-swift | 14:07 | |
| *** chsc has joined #openstack-swift | 14:07 | |
| *** mathiasb has quit IRC | 14:10 | |
| *** mathiasb has joined #openstack-swift | 14:11 | |
| *** tqtran has joined #openstack-swift | 14:19 | |
| *** tqtran has quit IRC | 14:23 | |
| *** vinsh has joined #openstack-swift | 14:24 | |
| *** psachin has quit IRC | 14:30 | |
| *** vinsh has quit IRC | 14:38 | |
| *** vinsh has joined #openstack-swift | 14:39 | |
| *** vinsh has quit IRC | 14:39 | |
| *** vinsh has joined #openstack-swift | 14:40 | |
| *** acoles is now known as acoles_ | 14:40 | |
| *** daemontool_ has joined #openstack-swift | 14:41 | |
| *** dmorita has joined #openstack-swift | 14:42 | |
| *** daemontool has quit IRC | 14:44 | |
| *** dmorita has quit IRC | 14:46 | |
| openstackgerrit | Donagh McCabe proposed openstack/swift: Document access control lists (ACLs) https://review.openstack.org/374215 | 14:47 |
| *** tgtanya_ has joined #openstack-swift | 14:49 | |
| *** acoles_ is now known as acoles | 14:49 | |
| openstackgerrit | Alistair Coles proposed openstack/swift: Add more comment to authtoken sample options https://review.openstack.org/372573 | 14:49 |
| *** daemontool_ has quit IRC | 14:58 | |
| *** arch-nemesis has joined #openstack-swift | 14:59 | |
| *** openstack has joined #openstack-swift | 15:01 | |
| *** sams-gleb has quit IRC | 15:04 | |
| *** _JZ_ has joined #openstack-swift | 15:08 | |
| *** oshritf has quit IRC | 15:11 | |
| *** klrmn has joined #openstack-swift | 15:20 | |
| *** sams-gleb has joined #openstack-swift | 15:23 | |
| *** sams-gleb has quit IRC | 15:24 | |
| *** mvk has quit IRC | 15:31 | |
| tdasilva | timburke, clayg, notmyname and other cores: are we trying to get patch 373537 done before release? can we add to Priority Reviews? | 15:32 |
| patchbot | https://review.openstack.org/#/c/373537/ - swift - Use separate headers for versioned_writes' stack a... | 15:32 |
| jrichli | acoles: re: container-sync : have there been any discussions around an approach to sync container metadata? | 15:32 |
| jrichli | tdasilva: FWIW I agree, it would be good to add to priority reviews. i will be taking a look today | 15:33 |
| *** klrmn has quit IRC | 15:33 | |
| openstackgerrit | Donagh McCabe proposed openstack/swift: Document access control lists (ACLs) https://review.openstack.org/374215 | 15:35 |
| *** kei_yama has joined #openstack-swift | 15:36 | |
| *** kei_yama has quit IRC | 15:41 | |
| acoles | jrichli: there is a bug report https://bugs.launchpad.net/swift/+bug/1464022 but looks like little activity for a while, I don't recall any discussion I had on that topic other than what is noted there | 15:42 |
| openstack | Launchpad bug 1464022 in OpenStack Object Storage (swift) "Container sync does not replicate container metadata" [Wishlist,Confirmed] - Assigned to Eran Rom (eranr) | 15:42 |
| tdasilva | jrichli: ack | 15:43 |
| jrichli | acoles: thanks | 15:43 |
| *** d0ugal has quit IRC | 15:48 | |
| *** d0ugal has joined #openstack-swift | 15:50 | |
| *** rcernin has quit IRC | 16:03 | |
| *** chianingwang has quit IRC | 16:04 | |
| *** chianingwang has joined #openstack-swift | 16:04 | |
| *** SkyRocknRoll has quit IRC | 16:11 | |
| *** pcaruana has quit IRC | 16:15 | |
| *** pcaruana has joined #openstack-swift | 16:17 | |
| *** arch-nemesis has quit IRC | 16:18 | |
| *** arch-nemesis has joined #openstack-swift | 16:18 | |
| openstackgerrit | Alistair Coles proposed openstack/swift: Check duplicate fragments do not cause 500 status https://review.openstack.org/374274 | 16:19 |
| *** arch-nemesis has joined #openstack-swift | 16:19 | |
| *** hseipp has quit IRC | 16:25 | |
| *** nadeem has joined #openstack-swift | 16:27 | |
| *** dmorita has joined #openstack-swift | 16:34 | |
| *** rledisez has quit IRC | 16:34 | |
| *** chsc has quit IRC | 16:54 | |
| *** ukaynar has joined #openstack-swift | 16:54 | |
| *** nikivi has joined #openstack-swift | 16:56 | |
| *** klrmn has joined #openstack-swift | 17:02 | |
| timburke | thanks jrichli and acoles! and mattoliverau, for fixing up my yaml! | 17:05 |
| acoles | timburke: what did I do ? | 17:06 |
| timburke | clayg: yeah, that was a weird one. figured i ought to submit a patch in case anyone else hits it | 17:06 |
| jrichli | my thoughts exactly :-) | 17:06 |
| acoles | yaml? | 17:06 |
| timburke | acoles: +A'ed the object sysmeta patch | 17:07 |
| timburke | and jrichli goaded you into it :-) | 17:07 |
| acoles | oic. and mattoliverau did something with yaml. got it | 17:07 |
| acoles | timburke: I previously goaded her into +A'ing one of clayg's. what goes round comes round :) | 17:08 |
| jrichli | lol | 17:08 |
| * jrichli is running to lunch | 17:08 | |
| *** dmsimard has joined #openstack-swift | 17:10 | |
| clayg | weee | 17:11 |
| tdasilva | acoles: re your comment about copying sysmeta to meta on POST requests, i think versioned writes would do that | 17:11 |
| tdasilva | acoles: https://github.com/openstack/swift/blob/master/swift/common/middleware/versioned_writes.py#L643 | 17:13 |
| clayg | psachin: I'm not aware of anyone that splits up accounts from conainers - they are so similar. | 17:13 |
| clayg | OTOH, I know lots of folks hat don't split a&c from o - and they all now realize it's terrible | 17:13 |
| timburke | tdasilva: likely, but for the moment we're only talking object sysmeta | 17:13 |
| clayg | notmyname says "always use SSD" - i'm mostly satisfied if you just use seperate disks | 17:15 |
| tdasilva | timburke: duh, true | 17:16 |
| tdasilva | nevermind | 17:16 |
| acoles | timburke: tdasilva yep that is a container request, but a good example of indiscriminate translation of resp headers | 17:16 |
| timburke | tdasilva: it'll certainly be something we want to think about when we're encrypting account/container meta, if we want to similarly expose encryption status | 17:17 |
| *** thebloggu has joined #openstack-swift | 17:22 | |
| *** nadeem has quit IRC | 17:24 | |
| *** nadeem has joined #openstack-swift | 17:24 | |
| *** tgtanya has quit IRC | 17:31 | |
| *** cdelatte has joined #openstack-swift | 17:32 | |
| *** tgtanya_ has quit IRC | 17:33 | |
| *** Suyi_ has joined #openstack-swift | 17:34 | |
| *** chsc has joined #openstack-swift | 17:38 | |
| openstackgerrit | Merged openstack/swift: Fix unstable json.dumps ordering in obj-server test https://review.openstack.org/373819 | 17:40 |
| *** geaaru has quit IRC | 17:44 | |
| *** acoles is now known as acoles_ | 17:57 | |
| *** hseipp has joined #openstack-swift | 18:04 | |
| openstackgerrit | Tim Burke proposed openstack/swift: Use separate headers for versioned_writes' stack and history modes https://review.openstack.org/373537 | 18:11 |
| zaitcev | Guys, is there a dedicated authentication server that supports v2, but it's not Keystone? Like swauth. | 18:18 |
| timburke | zaitcev: i know that swiftstack's auth middleware handles something kinda-vaguely-almost like v2 (if you squint a little) | 18:21 |
| timburke | something similar could probably be done for tempauth | 18:21 |
| zaitcev | Well, it's because I don't want to stand up MySQL and Keystone threatens to remove support for SQLite back-end. | 18:22 |
| zaitcev | Also, the requirements for Keystone are long. I just installed it again and it's 41 packages. | 18:22 |
| timburke | and i think that rax has something that does v2 but does not seem to be keystone. i forget what funniness i saw that led me to that conclusion... | 18:23 |
| openstackgerrit | Merged openstack/swift: Include object sysmeta in POST responses https://review.openstack.org/357559 | 18:24 |
| timburke | zaitcev: how much of the api do you want/need? do you want to be able to add/remove users, projects, etc. or just get & verify tokens? | 18:24 |
| zaitcev | I suppose I need to configure accounts and users somehow. | 18:25 |
| timburke | would a flat file work? like tempauth but with enough v2 support to get & verify tokens? | 18:27 |
| timburke | 'cause we could just split off the :, call the first piece "project" and the rest "user". if all you need is password auth, that should work well enough | 18:28 |
| *** rcernin has joined #openstack-swift | 18:30 | |
| zaitcev | I'm just trying to reproduce a problem in python-swifclient that occurs with v2. | 18:33 |
| *** hseipp has quit IRC | 18:35 | |
| *** klamath has quit IRC | 18:47 | |
| *** tqtran has joined #openstack-swift | 18:47 | |
| *** ig0r_ has joined #openstack-swift | 18:51 | |
| *** asettle has quit IRC | 18:58 | |
| tdasilva | there's no community meeting today, correct? | 18:58 |
| zaitcev | I think notmyname is in NY or DC or something | 19:00 |
| *** thebloggu has quit IRC | 19:04 | |
| *** catintheroof has joined #openstack-swift | 19:12 | |
| *** ukaynar has quit IRC | 19:27 | |
| openstackgerrit | Merged openstack/python-swiftclient: remove swiftclient/openstack/common/* in the .coveragerc https://review.openstack.org/373846 | 19:41 |
| clayg | tdasilva: I think that's correct - notmyname said we should have one if someone will chair - and we all sulked | 19:46 |
| *** silor has quit IRC | 19:48 | |
| *** cdelatte has quit IRC | 19:54 | |
| *** kei_yama has joined #openstack-swift | 20:00 | |
| *** kei_yama has quit IRC | 20:05 | |
| *** panbalag has joined #openstack-swift | 20:07 | |
| *** catintheroof has quit IRC | 20:08 | |
| *** panbalag has quit IRC | 20:12 | |
| *** sinval has joined #openstack-swift | 20:34 | |
| *** Jeffrey4l has quit IRC | 20:35 | |
| *** Jeffrey4l has joined #openstack-swift | 20:37 | |
| *** kei_yama has joined #openstack-swift | 20:44 | |
| *** cutforth has joined #openstack-swift | 20:52 | |
| *** kei_yama has quit IRC | 20:52 | |
| *** nikivi has quit IRC | 20:53 | |
| *** nikivi has joined #openstack-swift | 20:56 | |
| *** cutforth has quit IRC | 20:59 | |
| *** sinval has left #openstack-swift | 21:00 | |
| clayg | nova has a new contributor liaison! https://wiki.openstack.org/wiki/Nova/Mentoring#How_to_Get_Involved | 21:00 |
| clayg | we should have notmyname do that job | 21:00 |
| *** tongli has quit IRC | 21:00 | |
| *** nikivi has quit IRC | 21:01 | |
| zaitcev | It was one case where capitalization was essential. | 21:02 |
| *** cdelatte has joined #openstack-swift | 21:08 | |
| openstackgerrit | Clay Gerrard proposed openstack/swift: Delete old tombstones https://review.openstack.org/346865 | 21:10 |
| openstackgerrit | OpenStack Proposal Bot proposed openstack/python-swiftclient: Updated from global requirements https://review.openstack.org/89250 | 21:10 |
| openstackgerrit | Clay Gerrard proposed openstack/swift: Update diskfile docstring for get_ondisk_files https://review.openstack.org/374399 | 21:17 |
| timburke | clayg: i thought notmyname already does all the jobs? we need to clone him | 21:19 |
| *** rcernin has quit IRC | 21:24 | |
| *** wasmum has joined #openstack-swift | 21:24 | |
| *** ujjain has quit IRC | 21:29 | |
| *** ujjain has joined #openstack-swift | 21:30 | |
| *** ujjain has joined #openstack-swift | 21:30 | |
| *** nadeem has quit IRC | 21:43 | |
| *** ig0r_ has quit IRC | 21:45 | |
| *** nadeem has joined #openstack-swift | 21:56 | |
| mattoliverau | Morning | 22:01 |
| *** klamath has joined #openstack-swift | 22:10 | |
| *** StraubTW has quit IRC | 22:12 | |
| *** timburke has quit IRC | 22:16 | |
| *** timburke has joined #openstack-swift | 22:16 | |
| *** ChanServ sets mode: +v timburke | 22:16 | |
| openstackgerrit | Clay Gerrard proposed openstack/swift: Move documented reclaim_age option to correct location https://review.openstack.org/374419 | 22:22 |
| timburke | clayg: in my head, i read your description from https://bugs.launchpad.net/bugs/1626290 like one would "there's nothing in the rulebook that says a dog *can't* play football!" | 22:23 |
| openstack | Launchpad bug 1626290 in OpenStack Object Storage (swift) "[object-reconstructor] section is not documented" [Wishlist,New] | 22:23 |
| *** StraubTW has joined #openstack-swift | 22:59 | |
| *** mvk has joined #openstack-swift | 23:04 | |
| *** jamielennox is now known as jamielennox|away | 23:04 | |
| *** asettle has joined #openstack-swift | 23:04 | |
| *** asettle has quit IRC | 23:09 | |
| *** hoonetorg has quit IRC | 23:10 | |
| *** nadeem has quit IRC | 23:16 | |
| *** chsc has quit IRC | 23:16 | |
| *** klamath has quit IRC | 23:17 | |
| openstackgerrit | Clay Gerrard proposed openstack/swift: Move documented reclaim_age option to correct location https://review.openstack.org/374419 | 23:25 |
| clayg | how many skips are folks averaging in unitttests these days? I'd generally prefer we not have to have skips in *unit* tests? | 23:27 |
| clayg | i haven't really dug into what they are? | 23:28 |
| clayg | SKIP=14? is it just me!? | 23:28 |
| klrmn | clayg: aren't most of the skips around which features are supported for which swift versions? | 23:29 |
| clayg | oh, it's the new linkat O_TMP stuff | 23:29 |
| clayg | klrmn: not in swift's own unittests - they just test whatever features are supported int he commit your testing - there's a few crazy tests for the databases that like had to capture the old db schema's and test migration upgrades - and sometimes a mock might do something like "old object servers used to respond like this bullshit here" - but generally there's no unitests "run against old swift versions" per say | 23:31 |
| clayg | anyway - I think i just need to get on some xenial stuff? or upgrade my kernal or something? | 23:31 |
| *** nikivi has joined #openstack-swift | 23:32 | |
| clayg | notmyname: mattoliverau: you guys approved patch 162243 - how do i get my business up where I can run all the tests for the new hawtness?! | 23:33 |
| patchbot | https://review.openstack.org/#/c/162243/ - swift - Make object creation more atomic in Linux (MERGED) | 23:33 |
| clayg | i'm also seeing the return of the ChunkWriteTimeout(60) in the stderr's again | 23:34 |
| clayg | timburke: do you remember the hacks we had to do to track it down last time? | 23:34 |
| timburke | clayg: ugh. i think i put it out of my mind. only thing i remember was that it was a PITA | 23:35 |
| openstackgerrit | Clay Gerrard proposed openstack/swift: Move documented reclaim_age option to correct location https://review.openstack.org/374419 | 23:37 |
| clayg | timburke: i remember some hack i did somewhere in... nose? that kept the hub around long enought so that it would definately fire if the test that triggered it ran | 23:38 |
| clayg | but I don't recall what crazy stack/patch/debug I had to do to *find* the timeout created with 60 seconds? | 23:38 |
| clayg | .... I might have ... if timeout == 60: asdf somewhere in eventlet (!? | 23:39 |
| *** Suyi_ has quit IRC | 23:40 | |
| timburke | i thought we maybe just found (one of?) the tests that caused it and stuck a sleep() at the end. that seems to be what you did in https://gist.github.com/clayg/64d659bcae9991e5686c | 23:40 |
| *** nikivi has quit IRC | 23:41 | |
| openstackgerrit | Tim Burke proposed openstack/swift: Use separate headers for versioned_writes' stack and history modes https://review.openstack.org/373537 | 23:42 |
| *** jamielennox|away is now known as jamielennox | 23:43 | |
| clayg | timburke: yes! but how did we find that test!? | 23:47 |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!