| peluse | oh, is that what that is for?? | 00:00 |
|---|---|---|
| torgomatic | yep | 00:00 |
| peluse | then its not duplicated code and both etag related code snippets are needed | 00:00 |
| torgomatic | otherwise you could get something that's either an FA from a different object, or in replicated-land, just a different object | 00:00 |
| torgomatic | yep, that's important | 00:00 |
| torgomatic | and confusing | 00:00 |
| peluse | yeah, that would suck a little | 00:00 |
| * torgomatic is going to go put a comment on it | 00:01 | |
| peluse | OK, so that's the etag_buckets thing in the _get_or_head_response() in my patch... | 00:01 |
| peluse | the rest of the patch is about having the storage node communicate back other FIs that are available besides the one it has and then having the GET path circle back and try again using those if it doesn't have enough (in one etag bucket of course) | 00:02 |
| peluse | appreciate if you guys can take a gander. the test mods to make it work had me hacking like a fool so it could be cleaned up a little for sure but right now it works w/the limited new test code that I added | 00:02 |
| peluse | also added one property in diskfile that I don't really like as it required a change to mem_diskfile but was the quickest way I could see to plumb the alternate info back up the stack | 00:03 |
| clayg | peluse: so how far does it go past primaries into handoffs before circling back to primaries it through out? | 00:03 |
| openstackgerrit | Samuel Merritt proposed openstack/swift: Add comment about ResumingGetter.used_source_etag https://review.openstack.org/203877 | 00:04 |
| clayg | peluse: what are the best new tests? | 00:04 |
| peluse | one sec... | 00:04 |
| clayg | torgomatic: so what if the resumer on EC ends up picking a node that has a duplicate fi (one that's already streaming from another node?) | 00:05 |
| torgomatic | clayg: everything fails horribly | 00:06 |
| torgomatic | I think you may have anticipated that answer ;) | 00:06 |
| peluse | clayg, wrt the how far question, the resumingGetter will go through the node_iter assuming its getting 404's until its exhausted. then _get_or_head_response() will say "hey, I don't have enough but there are some nodes I talked to that have other FI's, let me try those" and a new node_iter with just those is used for anoter round of GETs | 00:06 |
| clayg | torgomatic: I think we picked the wrong set of failures to punt on | 00:06 |
| torgomatic | clayg: not the wrong set, just a subset of what it needs to be | 00:06 |
| peluse | heh, and the wrong day to stop sniffing glue | 00:06 |
| *** nadeem has quit IRC | 00:07 | |
| torgomatic | for the specific case of resuming, we need to make sure it's the same etag *and* some other stuff | 00:07 |
| *** DericHorn-HP has joined #openstack-swift | 00:07 | |
| torgomatic | where "other stuff" is nothing for replication, but something for EC | 00:07 |
| peluse | the new probe test for the etag_buckets thing is either from you or bill huber, not sure. its TestECObjectHandoffOverwrite and is just basic right now but does exercise the etag_bucket idea well | 00:07 |
| torgomatic | if we get a different source etag, then it's no good, FIs, FAs, ECs, and PDQs notwithstanding | 00:07 |
| peluse | the basic test for the alternate index thing is called test_alt_fragments_ok() and just tests one scenario now | 00:08 |
| *** thurloat is now known as thurloat_isgone | 00:09 | |
| clayg | peluse: it's a new test? | 00:11 |
| clayg | in proxy.controllers.test_obj? | 00:11 |
| torgomatic | clayg: it's in the patch he referenced, yeah | 00:12 |
| peluse | test_server.py class TestECMismatchedFA() | 00:12 |
| clayg | peluse: k, got it | 00:14 |
| peluse | clayg, once you read that test... I was thinking about a func to genercize shuffling the FA's around and feeding it various scenarios.. with GETs in between | 00:14 |
| torgomatic | yeah, a test helper for that would probably be useful, because we're gonna have a bunch of these | 00:15 |
| peluse | yup | 00:15 |
| peluse | I was more concerned about getting just one to work to start with :) | 00:15 |
| clayg | yeah that'd be pretty great - although I think doing it in proxy test_server talking to real object-server's is a both a) heavy-handed b) harder to extend to different ec-schemes | 00:15 |
| torgomatic | same etag, new FI; same etag, repeated FI; different etag... | 00:15 |
| torgomatic | maybe it's not actually that many | 00:15 |
| clayg | it'd be easier to do the generalization in test.proxy.controller.test_obj | 00:15 |
| clayg | imho | 00:15 |
| * peluse likes heavy handed shit :) | 00:16 | |
| peluse | plus I'm not the test master that you are :):) | 00:16 |
| clayg | peluse: - yeah torgomatic too | 00:16 |
| * peluse can barely mock his way out of a paper bag | 00:16 | |
| clayg | heh obj4_ring_path - maybe it's *not* that hard to add new scenarios - it's like proxy.test_server is slow already or anything | 00:17 |
| clayg | s/it's like/it's not like/g | 00:17 |
| peluse | yeah, I needed more devices to setup a decent scenario and chaning the 3rd was going to involve a shitpile of other changes to unrelated tests | 00:18 |
| clayg | ... yup | 00:18 |
| peluse | and I tell ya, at first I didn't make unique dev IDs on the ring and oh boy was that fun to debug in new test code... | 00:20 |
| clayg | lol | 00:20 |
| peluse | I was like WTF? How do I have primary nodes that are also handoffs??? | 00:20 |
| *** gyee has quit IRC | 00:20 | |
| *** chlong has joined #openstack-swift | 00:21 | |
| peluse | hey I gotta run next door to the grocery store or my family wont have dinner tonight (I feed them every other night and tonight is bread *and* water) so I'll be back in 30-40 min.... | 00:22 |
| clayg | peluse: I'd really like to see the new code in proxy.controllers.obj broken out into a few methods | 00:23 |
| *** dmorita has joined #openstack-swift | 00:31 | |
| *** DericHorn-HP has quit IRC | 00:33 | |
| *** DericHorn-HP has joined #openstack-swift | 00:35 | |
| *** DericHorn-HP has quit IRC | 00:35 | |
| peluse | clayg, can do... did you have something specific in mind or just less hacky? | 00:53 |
| clayg | peluse: heh - anything to break it up really | 00:54 |
| peluse | no problema | 00:55 |
| peluse | clayg, your contribution sugestion in the comments would be HUMONGOUS! | 00:56 |
| clayg | peluse: yeah - i'll work on that | 00:58 |
| peluse | and thanks for the quick comments too... | 01:02 |
| clayg | peluse: might help the next guy navigate the change too | 01:03 |
| clayg | peluse: oh... on second though I shouldn't have scored it - now probably no one will look at it till you push a new change? | 01:03 |
| peluse | yeah so thanks for nuthin' :) | 01:04 |
| clayg | ikr | 01:04 |
| clayg | :'( | 01:05 |
| *** ccavanna has joined #openstack-swift | 01:14 | |
| *** haomaiwang has quit IRC | 01:24 | |
| *** bitblt has quit IRC | 01:29 | |
| *** kota_ has joined #openstack-swift | 01:40 | |
| *** ChanServ sets mode: +v kota_ | 01:40 | |
| *** jasondotstar has joined #openstack-swift | 01:45 | |
| openstackgerrit | Matthew Oliver proposed openstack/swift: Add container and account reverse listings https://review.openstack.org/120709 | 02:43 |
| *** jasondotstar has quit IRC | 02:47 | |
| *** barker has joined #openstack-swift | 03:05 | |
| *** mahatic has joined #openstack-swift | 03:05 | |
| *** sanchitmalhotra has joined #openstack-swift | 03:13 | |
| *** barker has quit IRC | 03:20 | |
| *** mahatic has quit IRC | 03:29 | |
| *** barker has joined #openstack-swift | 03:30 | |
| openstackgerrit | Carlos Cavanna proposed openstack/swift: Improving statistics sent to Graphite. https://review.openstack.org/202657 | 03:46 |
| *** barker has quit IRC | 03:57 | |
| *** pbourke has quit IRC | 03:57 | |
| *** pbourke has joined #openstack-swift | 03:57 | |
| *** sanchitmalhotra1 has joined #openstack-swift | 04:00 | |
| *** sanchitmalhotra has quit IRC | 04:03 | |
| *** sanchitmalhotra has joined #openstack-swift | 04:03 | |
| *** annegent_ has joined #openstack-swift | 04:04 | |
| *** sanchitmalhotra1 has quit IRC | 04:05 | |
| *** annegentl_ has quit IRC | 04:07 | |
| *** sanchitmalhotra1 has joined #openstack-swift | 04:10 | |
| *** sanchitmalhotra has quit IRC | 04:12 | |
| openstackgerrit | Matthew Oliver proposed openstack/swift: Add container and account reverse listings https://review.openstack.org/120709 | 04:39 |
| jith_ | hi all.. what is swift_store_key in configuring swift for glance backend... is it password for swift user?? | 04:49 |
| *** zaitcev has quit IRC | 05:07 | |
| *** hrou has quit IRC | 05:59 | |
| *** Kennan has quit IRC | 06:01 | |
| *** Kennan has joined #openstack-swift | 06:02 | |
| *** kota_ has quit IRC | 06:06 | |
| *** ho has joined #openstack-swift | 06:09 | |
| mattoliverau | jith_: sorry I'm not sure, I haven't set it up. I'll have to go read the glace documentation. | 06:17 |
| ho | acoles_away, donagh: around? | 06:34 |
| openstackgerrit | Merged openstack/swift: Add comment about ResumingGetter.used_source_etag https://review.openstack.org/203877 | 06:34 |
| *** hseipp has joined #openstack-swift | 07:05 | |
| jith_ | mattoliverau: its ok.. thanks | 07:08 |
| *** losingle has joined #openstack-swift | 07:30 | |
| mattoliverau | jith_: swift_store_key is the key (password) of the swift user glace is using to connect to swift. | 07:34 |
| mattoliverau | jith_: http://docs.openstack.org/developer/glance/configuring.html#configuring-the-swift-storage-backend | 07:34 |
| jith_ | mattoliverau: thanks a lot... Another doubt swift_store_auth_address is the the address of proxy layer which accepts connection right?? Inside proxy conf file i have mentioned the keystone authentication.... means the user of swift should get authenticated with keystone... so should i specify the direct keystone auth adddress or swift proxy address in swift_store_auth_address? sorry for the... | 07:38 |
| jith_ | ...trouble :) | 07:38 |
| mattoliverau | jith_: that's ok, I work with a glance core, but he is in a different timezone, so I can ask him or introduce you some other time. I haven't set it up before, but looks like you can set it to keystone if thats what your using for swift authentication. https://www.rdoproject.org/forum/discussion/252/use-swift-as-backend-for-glance/p1 | 07:45 |
| *** jordanP has joined #openstack-swift | 07:50 | |
| *** losingle has quit IRC | 07:52 | |
| *** jistr has joined #openstack-swift | 07:55 | |
| jith_ | mattoliverau: thanks a lot.. sure.. sorry for late reply | 08:01 |
| *** chlong has quit IRC | 08:04 | |
| *** wbhuber_ has quit IRC | 08:16 | |
| *** 20WABQBTM has joined #openstack-swift | 08:37 | |
| *** 20WABQBTM has quit IRC | 08:37 | |
| *** haomaiwang has joined #openstack-swift | 08:37 | |
| *** aix has joined #openstack-swift | 08:43 | |
| *** aix has quit IRC | 08:58 | |
| *** rvasilets__ has joined #openstack-swift | 09:09 | |
| *** aix has joined #openstack-swift | 09:28 | |
| *** Kennan has quit IRC | 10:02 | |
| *** Kennan has joined #openstack-swift | 10:03 | |
| *** aix has quit IRC | 10:04 | |
| *** aix has joined #openstack-swift | 10:15 | |
| *** chmouel_ is now known as chmouel | 10:25 | |
| ho | acoles_away, donagh: around? | 10:44 |
| *** silor has joined #openstack-swift | 10:50 | |
| *** dmorita has quit IRC | 10:59 | |
| *** silor has quit IRC | 11:07 | |
| *** rvasilets__ has quit IRC | 11:34 | |
| *** sanchitmalhotra1 has quit IRC | 11:35 | |
| openstackgerrit | Hisashi Osanai proposed openstack/swift: Remove _keystone_identity method https://review.openstack.org/204050 | 12:08 |
| *** silor has joined #openstack-swift | 12:10 | |
| *** robefran has quit IRC | 12:11 | |
| *** silor1 has joined #openstack-swift | 12:15 | |
| *** peterbusque has joined #openstack-swift | 12:15 | |
| *** chlong has joined #openstack-swift | 12:17 | |
| *** silor has quit IRC | 12:18 | |
| *** silor1 is now known as silor | 12:18 | |
| *** SkyRocknRoll has joined #openstack-swift | 12:25 | |
| *** barker has joined #openstack-swift | 12:25 | |
| *** barker has quit IRC | 12:27 | |
| *** km has quit IRC | 12:29 | |
| *** ho has quit IRC | 12:34 | |
| *** robefran has joined #openstack-swift | 12:43 | |
| *** kei_yama has quit IRC | 12:55 | |
| *** annegentl_ has joined #openstack-swift | 13:04 | |
| *** annegent_ has quit IRC | 13:06 | |
| *** tsg has joined #openstack-swift | 13:13 | |
| *** hrou has joined #openstack-swift | 13:18 | |
| *** SkyRocknRoll has quit IRC | 13:20 | |
| *** wbhuber has joined #openstack-swift | 13:22 | |
| *** janonymous has joined #openstack-swift | 13:22 | |
| *** ccavanna has quit IRC | 13:38 | |
| *** pbourke has quit IRC | 13:39 | |
| *** annegentl_ has quit IRC | 13:39 | |
| *** peterbusque has quit IRC | 13:50 | |
| *** rbrooker has joined #openstack-swift | 13:50 | |
| *** peterbusque has joined #openstack-swift | 13:50 | |
| openstackgerrit | janonymous proposed openstack/swift: unit tests: Replace "self.assert_" by "self.assertTrue" https://review.openstack.org/204092 | 13:56 |
| *** tongli has joined #openstack-swift | 13:56 | |
| openstackgerrit | janonymous proposed openstack/swift: test/(functional, probe): Replace "self.assert_" by "self.assertTrue" https://review.openstack.org/204093 | 13:58 |
| *** hrou has quit IRC | 14:00 | |
| *** breitz has quit IRC | 14:06 | |
| *** breitz has joined #openstack-swift | 14:06 | |
| *** amoturi has joined #openstack-swift | 14:12 | |
| *** jasondotstar has joined #openstack-swift | 14:13 | |
| *** hrou has joined #openstack-swift | 14:16 | |
| *** ccavanna has joined #openstack-swift | 14:16 | |
| *** jrichli has joined #openstack-swift | 14:22 | |
| *** jlhinson has joined #openstack-swift | 14:24 | |
| *** jkugel has joined #openstack-swift | 14:26 | |
| tdasilva | good morning | 14:29 |
| *** david-lyle has quit IRC | 14:30 | |
| janonymous | Hi Folks, I had a small query that discussed it on openstack-dev also, but through the response i am only able to understand a small part of it : | 14:33 |
| janonymous | My Question is simple the jenkins gate jobs produce a neat testr_results.html.gz of the failures through which it is very easy to track the problem like : http://logs.openstack.org/66/193866/10/check/gate-keystone-python27/bcfc010/ ; but in swift no such thing is present | 14:33 |
| janonymous | I think the answer lies in : http://git.openstack.org/cgit/openstack-infra/project-config/tree/jenkins/jobs/swift.yaml ; publishers: - devstack-logs - console-log | 14:34 |
| janonymous | which has an additional - test-results parameter in publisher : http://git.openstack.org/cgit/openstack-infra/project-config/tree/jenkins/jobs/keystone.yaml | 14:35 |
| openstackgerrit | Carlos Cavanna proposed openstack/swift: Improving statistics sent to Graphite. https://review.openstack.org/202657 | 14:35 |
| janonymous | Could someone help me out on this... | 14:35 |
| *** rbrooker has quit IRC | 14:37 | |
| *** annegentle has joined #openstack-swift | 14:37 | |
| openstackgerrit | paul luse proposed openstack/swift: Add ability for GET path to see/select alternate frag archs https://review.openstack.org/201283 | 14:41 |
| *** lpabon has joined #openstack-swift | 14:44 | |
| *** admin6 has joined #openstack-swift | 14:45 | |
| peluse | jamielennox, I can try... taking a look now | 14:45 |
| peluse | hmmm, why are you asking in the swift channel if I may ask? | 14:48 |
| peluse | janonymous, those last 2 comments were for you :) | 14:49 |
| janonymous | Ohh.. :) | 14:49 |
| janonymous | i asked it on dev ... and the that's the answer i got | 14:50 |
| janonymous | so thought someone could answer here... | 14:50 |
| *** wbhuber_ has joined #openstack-swift | 15:00 | |
| *** robefran has quit IRC | 15:04 | |
| *** hrou has quit IRC | 15:04 | |
| *** minwoob has joined #openstack-swift | 15:05 | |
| *** robefran has joined #openstack-swift | 15:06 | |
| *** wbhuber has quit IRC | 15:09 | |
| *** chlong has quit IRC | 15:09 | |
| *** wer has quit IRC | 15:09 | |
| *** jamielennox has quit IRC | 15:09 | |
| *** CrackerJackMack has quit IRC | 15:09 | |
| *** briancline has quit IRC | 15:09 | |
| *** sudorandom has quit IRC | 15:09 | |
| *** wbhuber_ is now known as wbhuber | 15:10 | |
| janonymous | peluse: Did you had a look ? | 15:10 |
| *** proteusguy has quit IRC | 15:11 | |
| *** chlong has joined #openstack-swift | 15:15 | |
| admin6 | Hi all. I’m planning a swift deployment and currently trying to design the way I’ll store my objects into Swift. Something not clear for me is how swift manage concurrent write access to a container under heavy load, between write lock into sqllite and replication between container nodes… Is there a documentation regarding this specific behavior? | 15:23 |
| *** proteusguy has joined #openstack-swift | 15:26 | |
| *** thebloggu has joined #openstack-swift | 15:30 | |
| thebloggu | has anyone here ever tried to run a swift aio in docker for tests? how can I use xfs inside the container (to support extended attributes)? | 15:32 |
| *** hrou has joined #openstack-swift | 15:36 | |
| *** david-lyle has joined #openstack-swift | 15:38 | |
| chmouel | thebloggu: i don't the answer but docker -v would just bind mount whatever filesystem you have | 15:40 |
| chmouel | unless you have your docker running on AUFS (default ubuntu i think) then you can't | 15:40 |
| *** joeljwright has joined #openstack-swift | 15:43 | |
| *** ChanServ sets mode: +v joeljwright | 15:43 | |
| thebloggu | chmouel, right, so I should just mount an xfs filesystem with -v? isn't it possible to have a non persistent option? | 15:44 |
| chmouel | thebloggu: I think as long your /var/lib/docker is using the dm-lvm docker's driver then it *should* work | 15:45 |
| thebloggu | chmouel, according to `docker info` it's using: `Storage Driver: devicemapper`. I'm asking this because I was using swift earlier and got a "No data available" when I tried to read an objects metadata (manually, in the swift node) and I thought the problem could be not having support for extended attributes in the filesystem (even though PUT/GET requests of objects to swift work without a problem, even with custom metadata) | 15:51 |
| *** mfalatic has joined #openstack-swift | 15:56 | |
| clayg | has anyone looked seriously at the per-device-rsync-modules? | 15:57 |
| clayg | is the repliation version of swifterdarrell's server-per-disk | 15:57 |
| clayg | I don't know where the patch is | 15:57 |
| clayg | peluse: k, i'm spinning up your ec GET fix now - gunna see what I can with a test | 15:59 |
| *** annegentle has quit IRC | 15:59 | |
| *** nadeem has joined #openstack-swift | 16:02 | |
| *** nadeem has quit IRC | 16:02 | |
| *** nadeem has joined #openstack-swift | 16:03 | |
| *** jistr has quit IRC | 16:04 | |
| admin6 | I’m trying to design my future swift cluster. I have between 300.000 and 700.000 new files written every day (without counting read access to the objects). Do you think it’s reasonable regarding to concurrent access to the container sqlite file to create one container per day to store all my file? | 16:08 |
| *** gyee has joined #openstack-swift | 16:08 | |
| clayg | admin6: yeah one container per day would be fine | 16:13 |
| *** peterbusque has quit IRC | 16:13 | |
| clayg | admin6: do you have that kind of temporal knowledge when accessing the object? | 16:13 |
| ccavanna | clayg: I'll propose the documentation change we talked about yesterday in openstack-infra. | 16:13 |
| *** peterbusque has joined #openstack-swift | 16:14 | |
| clayg | ccavanna: the developer docs are in the swift tree | 16:14 |
| clayg | ccavanna: the swift all in one docs are something we "maintain" | 16:14 |
| clayg | ccavanna: where "maintain" ~= "trick people new to the project into helping us with" | 16:14 |
| notmyname | good morning, world | 16:14 |
| admin6 | clayg: temporal knowledge is not very important for me. my main concern is about concurrent write access to the container sqlite database | 16:15 |
| ccavanna | clayg: I fall into that category, perfectly :) | 16:15 |
| ccavanna | I think the change belongs to this page: http://docs.openstack.org/infra/manual/developers.html | 16:15 |
| ccavanna | But I didn't see the doc for that under swift. | 16:15 |
| clayg | admin6: well why not do a hash ring? | 16:15 |
| *** peterbusque has quit IRC | 16:15 | |
| *** peterbusque has joined #openstack-swift | 16:16 | |
| clayg | admin6: concurrent write access to a single container is pretty solid to the tune of 100's per second | 16:17 |
| clayg | are these half-million objects coming into a single contaier *all at once* or do they trickle in over the day? | 16:17 |
| *** lpabon has quit IRC | 16:17 | |
| admin6 | clayg: its all over the day, with peak at some hours | 16:18 |
| clayg | the thing with multiple containers it that as they get.... idk ~4M replication can get pretty annoying | 16:18 |
| clayg | admin6: this helps speed out the occassional .pending eating slowdown during the object write path -> https://review.openstack.org/#/c/189080/ | 16:19 |
| admin6 | good to hear that I can tune up to 100 write per second on a single container | 16:20 |
| clayg | admin6: yeah the trickier part there is getting enough concurrency in the guys doing the uploads - and enough object disks to support all those iops - the containers may not be what limits you depending on the size of your cluster? | 16:21 |
| clayg | ccavanna: oh yeah - that page isn't us | 16:22 |
| admin6 | clayg: ok, my cluster will start with about 400TB of data ans 600 Millions of objects | 16:23 |
| clayg | admin6: that's a great cluster! congradulations! | 16:24 |
| admin6 | clayg: one thing which is still not clear to me is how sql write to the container file and swift recplicas between the continers nodes works together. | 16:25 |
| *** mgarza has joined #openstack-swift | 16:25 | |
| clayg | admin6: is that 4 nodes 24 disks or so? | 16:26 |
| clayg | admin6: you know... container replication is kinda tricky... it's acctually one of the smarter parts of swift IMHO | 16:26 |
| *** geaaru has joined #openstack-swift | 16:27 | |
| admin6 | clayg: I’m confident :-) | 16:27 |
| clayg | admin6: there's a bunch of tricks in there - but at the end of the day it goes row by row to make sure each replica for the container has the latest transaction for every object | 16:27 |
| admin6 | clayg:… ok, never read that anywhere | 16:29 |
| notmyname | admin6: clayg gives a good summary of http://docs.openstack.org/developer/swift/overview_replication.html#db-replication | 16:30 |
| ccavanna | clayg: unless you think it belongs somewhere else. But that page seemed appropriate because it documents the steps for checking in/out code, which led to my problem with the tools. | 16:30 |
| notmyname | ccavanna: that page you linked is a great place for stuff that generally applies to all openstack projects. for stuff that's swift-specific, it's good to keep that in the swift source tree | 16:31 |
| admin6 | All: thanks guys for your very expert answers (as always). I’ll dig deeper into that and will be back soon ;-) | 16:34 |
| *** david-lyle has quit IRC | 16:34 | |
| clayg | notmyname: oh that's a really great summary, i didn't know we had that! | 16:36 |
| ccavanna | notmyname: I didn't find any doc page in swift that deals with this. Maybe I can create a new section under: http://docs.openstack.org/developer/swift/development_guidelines.html | 16:36 |
| notmyname | ccavanna: can you remind me what "this" is? ;-) | 16:37 |
| hrou | ccavanna: I think the issues you ran into were not swift specific and could apply to other components ? So that's probably a good place; Actually I was setting up a new VM this morning, followed SAIO instructions verbatim, and all tests (i.e. unit/probe/func) were 100% clean after (recent Ubuntu release). | 16:37 |
| notmyname | dfg_: ^ ;-) | 16:37 |
| clayg | notmyname: in fairness the saio docs start with "get a trusty vm" | 16:37 |
| ccavanna | notmyname: oh, right :). the issue was that after I rebased my changes, the swift scripts (such as swift-init) were out of sync with my code. | 16:38 |
| notmyname | clayg: what if I haven't spent that much time with the VM and dont' trust it yet? | 16:38 |
| clayg | hrou: ok, now it again only this time with precise - and post the errors ;) | 16:38 |
| ccavanna | And I was not able to start swift because the version of swift required (swift==2.3.1.dev33) did not match my version of swift (swift==2.3.1.dev34) | 16:39 |
| notmyname | ccavanna: ah, yeah. that's something that's general to openstack (ie anything using pbp/pip/setuptools and distro packages). probably would be good to document generally as a hint on how to solve it | 16:39 |
| notmyname | ccavanna: IIRC | 16:39 |
| *** admin6 has quit IRC | 16:39 | |
| hrou | clayg: ; - ) I agree though when I've had issues before it generally comes down to which Ubuntu release (or release of OS ** blah **) you use. | 16:39 |
| clayg | hrou: it's hard to keep a living document like that up-to-date for multiple distros :\ | 16:40 |
| hrou | clayg: Yea I can definitely see why that's a big pain! | 16:40 |
| ccavanna | notmyname: So I was thinking http://docs.openstack.org/infra/manual/developers.html | 16:40 |
| ccavanna | That's where more or less this belongs. I can mention my case with Swift as an example. | 16:41 |
| clayg | ccavanna: worth a shot! | 16:41 |
| notmyname | ccavanna: yeah, or maybe on http://docs.openstack.org/infra/manual/python.html#python-unit-tests (which I found linked on the "run unit tests" section of the doc you linked | 16:43 |
| clayg | ccavanna: if that doesn't work out for you (seriously good luck) - maybe a section that duplicates some of the "making your first change" part of that document in the swift developer docs - maybe the development guidelines should just a be a "developer guide" or something? | 16:43 |
| *** vinsh has quit IRC | 16:46 | |
| *** vinsh has joined #openstack-swift | 16:46 | |
| *** jlhinson_ has joined #openstack-swift | 16:46 | |
| *** jlhinson has quit IRC | 16:46 | |
| ccavanna | clayg: actually, I like that more. I'm not a big fan of duplicating documentation, but the advantage for Swift would be that everything will be in one place. | 16:48 |
| *** david-lyle has joined #openstack-swift | 16:48 | |
| ccavanna | If you think that's more appropriate, I can easily make the changes to the Swift doc and propose it. | 16:49 |
| notmyname | ccavanna: honestly, I don't think it matters where it is. what matters more is that it's available for new people coming in who might be hit with the problem. | 16:50 |
| notmyname | you've got great experience there since you're relatively new to swift | 16:50 |
| notmyname | if you'd be more likely to look in the infra manuals (because of existing links/docs), then let's put it there. if you look to the in-tree swift docs first, let's put it there | 16:51 |
| notmyname | ccavanna: so i think you've probably got a better sense of where to put it for new contributors than clayg or me (since we've been breathing swift for nearly 6 years) ;-) | 16:52 |
| klrmn | ccavanna: both? =) | 16:52 |
| notmyname | klrmn: I'm ok with that too :-) | 16:52 |
| *** vinsh has quit IRC | 16:52 | |
| ccavanna | Personally, I would have looked in Swift first, under Getting Started, in the Development section. | 16:55 |
| ccavanna | I can create a new topic, "Making your first change" or "Making your first contribution". | 16:55 |
| *** jasondotstar has quit IRC | 16:56 | |
| openstackgerrit | Clay Gerrard proposed openstack/swift: Fix Associated Projects link on Getting Started https://review.openstack.org/204167 | 16:58 |
| notmyname | clayg: what version of sphinx do you have installed? | 17:01 |
| notmyname | ugh. distro packages vs pip strikes again? I had python-sphinx installed. got an error building docs. removed it (with apt-get), now it works. but I don't see sphinx in dpkg or pip freeze! | 17:06 |
| notmyname | *magic* | 17:06 |
| clayg | notmyname: vagrant-swift-all-in-one has an autodoc command - it's *magic* | 17:07 |
| notmyname | yeah, it's pretty sweet | 17:07 |
| notmyname | that alone might make me start using vSAIO | 17:07 |
| clayg | oslosphinx==2.5.0 | 17:07 |
| clayg | Sphinx==1.1.3 | 17:07 |
| notmyname | yeah, i've got oslosphinx, just not plain sphinx | 17:08 |
| clayg | notmyname: better install it! | 17:08 |
| notmyname | why? it works without it | 17:08 |
| notmyname | ;-) | 17:08 |
| clayg | lol! | 17:08 |
| *** vinsh has joined #openstack-swift | 17:08 | |
| *** jordanP has quit IRC | 17:09 | |
| hrou | Funny enough all the issues I've had also always come down to distro packages vs pip (e.g. when I was installing an older version of swift on a recent release), fixing said issues is an art form I find, it can be mediating ; ) | 17:11 |
| vinsh | Hello swift. Anyone have hints for me on the following? when I try to list a specific container... from many different clients.. I always get the following failure "simplejson.scanner.JSONDecodeError: Expecting value: line 1 column 1 (char 0)" | 17:11 |
| vinsh | Wondered if that was something obvious you could point me at to look into. | 17:11 |
| notmyname | vinsh: what's the request you are making? can you pastebin an example? | 17:12 |
| vinsh | Yes. One sec. | 17:12 |
| vinsh | http://paste.openstack.org/show/396356/ | 17:12 |
| vinsh | Other containers in that tenant are fine.. something happened to this container (don't know what). and now it bombs out. | 17:13 |
| notmyname | I wonder if there are some weird object names in that container | 17:13 |
| vinsh | In a proxy log you see a 200 from the request. | 17:13 |
| vinsh | This is a container the monasca guys are feeding stuff into.. so that's possible :) | 17:14 |
| vinsh | I'm just unsure of a way to recover it or even look further. | 17:14 |
| notmyname | vinsh: do `swift --debug list OpenStackMetrics` and pastebin that (scrub tokens, if necessary) | 17:15 |
| *** jasondotstar has joined #openstack-swift | 17:16 | |
| *** lpabon has joined #openstack-swift | 17:17 | |
| vinsh | good idea... http://paste.openstack.org/show/396366/ | 17:17 |
| ccavanna | notmyname: Can I ask you a question about some tests? | 17:18 |
| notmyname | vinsh: oh, interesting | 17:19 |
| clayg | is that.... html? | 17:19 |
| notmyname | vinsh: first, see line 19. that's the equivalent curl request | 17:19 |
| clayg | i have to admit - if *i* was a json parser I would also blow up on html | 17:19 |
| vinsh | hah! | 17:19 |
| vinsh | why is it sending me html today? | 17:20 |
| notmyname | vinsh: second (and to the problem), looks like it might be a contaienr with staticweb enabled? | 17:20 |
| vinsh | OH... | 17:20 |
| clayg | notmyname: in tree static web will respect the accept-encoding or json query params | 17:20 |
| notmyname | which, honestly sounds like a bug in our client if it can't list staticweb containers | 17:20 |
| notmyname | clayg: yeah...but the pastebin... | 17:20 |
| notmyname | ccavanna: of course | 17:20 |
| clayg | notmyname: it's a bug in the middleware if format=json returns html | 17:21 |
| notmyname | yeah, I thought staticweb ignored the staticweb part if a token is in the request (unless there is also x-web-mode: true) | 17:22 |
| vinsh | How can I check if static web is enabled on this container? | 17:22 |
| clayg | stat -v | 17:22 |
| vinsh | oh. -v. thanks | 17:22 |
| clayg | honestly i don't see it in the debug response headers | 17:22 |
| notmyname | clayg: doesn't staticweb filter that out for the auto-gen'd listings? | 17:23 |
| vinsh | Yeah, nothing listed | 17:23 |
| notmyname | oh ok | 17:23 |
| *** aix has quit IRC | 17:23 | |
| notmyname | wow. pip is up to version 7.1.0. I had 1.5.6 installed | 17:25 |
| clayg | notmyname: version *7* - i just installed 6 - what are these fools doing!? | 17:25 |
| clayg | vinsh: i'm not 100% that html is totally stock staticweb - can you double check your pipeline? | 17:26 |
| vinsh | what do you mean by that? The proxy node middleware pipeline? | 17:26 |
| vinsh | [filter:staticweb] | 17:27 |
| vinsh | use = egg:swift#staticweb | 17:27 |
| vinsh | pipeline = catch_errors healthcheck proxy-logging cache ratelimit formpost tempurl authtoken staticweb slo keystone proxy-server | 17:27 |
| clayg | vinsh: weird | 17:28 |
| notmyname | clayg: are you trying it locally yet? I'm in the middle of running some testins locally | 17:28 |
| notmyname | testins? what's that | 17:28 |
| clayg | notmyname: sigh - no :'( but ok | 17:28 |
| notmyname | I can, but it will be a few minutes | 17:28 |
| vinsh | clayg: The problem container shows Content-Type: text/html; charset=UTF-8 where as other containers show Content-Type: text/plain; charset=utf-8 | 17:32 |
| vinsh | Ive never messed with changing or setting content type... didn't know you could. | 17:32 |
| notmyname | you don't/can't (directly) on a contianer | 17:32 |
| *** hseipp has quit IRC | 17:33 | |
| vinsh | Can we tell the client/swift to force json ? | 17:35 |
| notmyname | that's what the ?format=json is supposed to do | 17:35 |
| vinsh | :-/ | 17:35 |
| *** annegentle has joined #openstack-swift | 17:36 | |
| clayg | so it doesn't respect ?format=json | 17:37 |
| clayg | but it does respect x-auth-token | 17:37 |
| clayg | er.. it does respect ?format=json with an auth token | 17:37 |
| *** annegent_ has joined #openstack-swift | 17:39 | |
| notmyname | that's what I'd expect | 17:40 |
| clayg | vinsh: idk, womm -> https://gist.github.com/clayg/b7dc5f06d03529a2532f | 17:40 |
| clayg | vinsh: notmyname: so it looks like the x-container-metadata headers aren't returned when the request is trasnlated by static web | 17:41 |
| clayg | so these options may be set but stripped out before we can see them | 17:41 |
| *** annegentle has quit IRC | 17:41 | |
| clayg | it may very well be a bug in staticweb (what version is deployed?) - but as a work around if you have some containers working you may just need to POST to the container to disable those options | 17:41 |
| clayg | you can verify if they're set with swift-container-info if you have access to the container nodes | 17:42 |
| vinsh | I do have access to the container nodes | 17:42 |
| vinsh | Let me see if I understand. You suggest we post to the container to turn static web off on it (it may be on but we are unsure). then test? | 17:43 |
| clayg | notmyname: any chance keystone auth does somehting like env.pop('x-auth-token') before staticweb sees the response? | 17:43 |
| openstackgerrit | Robert Francis proposed openstack/swift: Add functional test test_copy_to_same to test_object.py https://review.openstack.org/203820 | 17:43 |
| clayg | vinsh: sees like worth a shot | 17:43 |
| vinsh | awesome, thanks for taking time to look at this with. | 17:43 |
| clayg | does anyone have the devstack/keystone swift development environ up? | 17:43 |
| vinsh | *me | 17:44 |
| clayg | vinsh: yup :\ | 17:44 |
| hrou | clayg, I have it setup on SAIO and vagrant, if that helps ? | 17:44 |
| notmyname | clayg: thanks. my tests just finished running | 17:45 |
| openstackgerrit | John Dickinson proposed openstack/swift: Update test-requirements to match global requirements. https://review.openstack.org/204179 | 17:45 |
| *** bitblt has joined #openstack-swift | 17:48 | |
| clayg | hrou: keystone? on a vagrant-swift-all-in-one? | 17:49 |
| vinsh | clayg: this is swift 2.2.2-0 btw. unsure of static web version. | 17:49 |
| clayg | hrou: anyway - yeah we need someone to check staticweb listings with keystone | 17:49 |
| clayg | vinsh: seems new enough to me! | 17:49 |
| hrou | clayg: Yep on vagrant-swift-all-in-one (I installed keystone after the fact) and on a standalone SAIO (i.e. none vagrant). | 17:50 |
| *** peterbusque has left #openstack-swift | 17:50 | |
| *** bitblt has quit IRC | 17:51 | |
| *** bitblt has joined #openstack-swift | 17:51 | |
| *** annegent_ has quit IRC | 17:51 | |
| clayg | hrou: can you add a option to vagrant-swift-all-in-one to setup a keystone that will pass functests? | 17:51 |
| notmyname | nice | 17:52 |
| clayg | if you do that I'll let you off the hook for the staticweb thing | 17:52 |
| clayg | ;) | 17:52 |
| *** jasondotstar has quit IRC | 17:52 | |
| hrou | clayg: I actually haven't tried to run func tests on my vagrant-swift-all-in-one && keystone env but let me give that a go right now ! | 17:53 |
| clayg | good luck! | 17:53 |
| hrou | clayg: In terms of static web, why the vagrant dependency, could I not simply test that in my non-vagrant setup for you ? : ) | 17:54 |
| clayg | hrou: yeah any keystone + staticweb would work - i was just trying to get free keystone support in vagrant-swift-all-in-one | 17:54 |
| clayg | hrou: the problem is really with keystone that unless it's actively maintained it rots | 17:55 |
| clayg | hrou: that's why I tend to use my devstack vm for keystone stuff - just cause other people care to keep it working | 17:55 |
| clayg | i just really need to write this test for peluse :D | 17:56 |
| vinsh | Having a hard time finding some docs on what I post to disable staic web. | 17:56 |
| clayg | vinsh: swift post container -m x-web-listings: | 17:57 |
| clayg | nope | 17:57 |
| clayg | i was *close* | 17:57 |
| hrou | clayg: Yea setting up keystone initially was a bit of a pain actually, so I can see issues (particularly if you move up your swift lvl). I think getting it built into vagrant and working is a great idea so I'll look into that regardless. | 17:57 |
| *** bitblt has quit IRC | 17:58 | |
| clayg | swift post -v container -m web-listings: | 17:58 |
| clayg | hrou: sure np, good luck with that too, very helpful, thank you! | 17:58 |
| *** bitblt has joined #openstack-swift | 17:58 | |
| clayg | vinsh: sorry my first command was wrong :\ | 17:59 |
| vinsh | Container list now works. | 17:59 |
| vinsh | BINGO. | 17:59 |
| clayg | vinsh: bummer :\ | 18:00 |
| clayg | vinsh: i mean it's good for you i suppose :P | 18:00 |
| *** silor has quit IRC | 18:00 | |
| vinsh | :-o | 18:00 |
| hrou | clayg: Sure thing and thanks ! keystone is a requirement for various middleware we're using / developing and it'd help to have an easy to setup env (for folks starting the project). | 18:00 |
| vinsh | You really narrowed down on that quickly. Impressed :) | 18:00 |
| hrou | clayg: I'll give the keystone + static web a go as well, but assuming you're doing the same, that's a race you're going to win any day ; ) | 18:01 |
| vinsh | hrou: what are you using to provision within your vagrant vms? | 18:01 |
| vinsh | Wonder if I can offer any help on the keystone part. | 18:01 |
| clayg | hrou: i really do need to write these tests for paul - I'm acctually ignoring the fact that staticweb is probably broken with keystone | 18:02 |
| occupant | it's funny how the openstack people straight up tell you keystone is awful and you shouldn't use it | 18:02 |
| *** annegentle has joined #openstack-swift | 18:02 | |
| clayg | occupant: don't listen to swift people about the rest of openstack - we're not experts | 18:02 |
| occupant | err swiftstack I mean | 18:02 |
| vinsh | We moved over to using fernet tokens with keystone.. and it works very well. | 18:02 |
| notmyname | vinsh: yeah, I've been wanting to try out fernet tokens | 18:02 |
| vinsh | mfisch will hopefully do a talk on it in Tokyo. | 18:03 |
| vinsh | with some of the fernet guys. | 18:03 |
| *** zaitcev has joined #openstack-swift | 18:03 | |
| *** ChanServ sets mode: +v zaitcev | 18:03 | |
| vinsh | clayg: hrou: If you do open a bug on this statickweb issue. Please add me or send me the id. Id love to learn more and follow. | 18:04 |
| notmyname | vinsh: I was just working on that :-) | 18:04 |
| notmyname | vinsh: bug tracker is at https://bugs.launchpad.net/swift/+bugs | 18:04 |
| clayg | notmyname: it's going to be unconfirmed until hrou finishes his keystone testing | 18:04 |
| notmyname | vinsh: filing it there would be a big help | 18:04 |
| notmyname | clayg: yeah | 18:04 |
| vinsh | Ok, i'll file it now. | 18:05 |
| notmyname | thanks | 18:05 |
| hrou | vinsh: Was just going to do a simple test to validate clayg / notmyname theory in my simple env, but you've just about confirmed it by disabling it. | 18:06 |
| notmyname | hrou: it would be good to see someone else with keystone who sees the bug | 18:06 |
| hrou | notmyname: yep ! Agreed just to make sure there's nothing else at play. | 18:06 |
| hrou | vinsh: btw did you have 'delay_auth_decision = true' set ? (sorry if you included that already). | 18:08 |
| *** lpabon has quit IRC | 18:08 | |
| vinsh | set to "1" | 18:09 |
| *** CrackerJackMack has joined #openstack-swift | 18:13 | |
| *** sudorandom has joined #openstack-swift | 18:13 | |
| *** jamielennox has joined #openstack-swift | 18:15 | |
| *** thebloggu has quit IRC | 18:16 | |
| *** jamielennox is now known as jamielennox|away | 18:16 | |
| *** david-lyle has quit IRC | 18:21 | |
| *** jamielennox|away is now known as jamielennox | 18:23 | |
| openstackgerrit | Azhagu Selvan SP proposed openstack/swift: Respect 'Accept' header in error responses https://review.openstack.org/204196 | 18:24 |
| ccavanna | Question about a unit test. In particular: test_policy_index (test.unit.common.middleware.test_proxy_logging.TestProxyLogging) fails when I check in the code, but does not fail in my unit tests. | 18:26 |
| ccavanna | The failure is related to my code change. I compared my version of the test with the latest on the repository, but I see no differences. | 18:26 |
| ccavanna | (My code change, by the way, is the logging of graphite stats for the commands organized by policy) | 18:28 |
| ccavanna | The error I see is as follows: | 18:29 |
| ccavanna | This line of code: | 18:29 |
| ccavanna | policy_name = POLICIES.get_by_index(policy_index).name | 18:29 |
| ccavanna | fails with: | 18:29 |
| ccavanna | AttributeError: 'NoneType' object has no attribute 'name' | 18:29 |
| ccavanna | So clearly POLICIES.get_by_index(policy_index) is returning None. | 18:30 |
| *** annegentle has quit IRC | 18:30 | |
| * notmyname has a lunch meeting | 18:31 | |
| ccavanna | One way for get_by_index to return None is if the index I pass is not valid ('' and None return 0, or the values is converted to int.) | 18:33 |
| ccavanna | Regardless of the code changes, I wonder why the test fails when I submit, but not in my unit tests. | 18:33 |
| clayg | peluse: so the test rig i want for the ec get stuff is a little further off from what we already have in proxy.controller.test_obj | 18:35 |
| clayg | peluse: I'm going to get on the train and think about what i want to do here | 18:36 |
| clayg | ccavanna: you need to wrap your test with the patch_policies decorator | 18:37 |
| clayg | ccavanna: you have a policy defined on your machine that is not defined on some other machine | 18:37 |
| clayg | you can probably just wrap the whole class - lots of tests use it - in test.unit.__init__.py - patch_policies - will solve all | 18:38 |
| clayg | bbiab | 18:38 |
| vinsh | clayg: notmyname: hrou: Static web related bug now filed: https://bugs.launchpad.net/swift/+bug/1476789 | 18:41 |
| openstack | Launchpad bug 1476789 in OpenStack Object Storage (swift) "staticweb enabled container will not list json" [Undecided,New] | 18:41 |
| openstackgerrit | Greg Lange proposed openstack/swift: go: Adds memcache ring https://review.openstack.org/203758 | 18:48 |
| *** jlhinson_ has quit IRC | 18:52 | |
| *** jasondotstar has joined #openstack-swift | 18:52 | |
| *** janonymous has quit IRC | 18:53 | |
| *** jlhinson has joined #openstack-swift | 18:55 | |
| openstackgerrit | Azhagu Selvan SP proposed openstack/swift: Respect 'Accept' header in error responses https://review.openstack.org/204196 | 18:55 |
| *** breitz has quit IRC | 19:03 | |
| ccavanna | clayg: Thanks for the tip. I patched the policies and now I can recreate the error. | 19:04 |
| ccavanna | clayg: is it worth keeping the patch (to keep one policy only) in the unit tests? That way there's a consistent policy setting for testing, like other tests do. | 19:05 |
| *** breitz has joined #openstack-swift | 19:08 | |
| *** gyee has quit IRC | 19:11 | |
| *** amit213 has quit IRC | 19:14 | |
| *** amit213 has joined #openstack-swift | 19:14 | |
| *** breitz has quit IRC | 19:17 | |
| *** jasondotstar has quit IRC | 19:18 | |
| *** breitz has joined #openstack-swift | 19:18 | |
| *** annegentle has joined #openstack-swift | 19:25 | |
| clayg | ccavanna: yeah always better to patch policies! | 19:30 |
| openstackgerrit | Carlos Cavanna proposed openstack/swift: Improving statistics sent to Graphite. https://review.openstack.org/202657 | 19:33 |
| *** david-lyle has joined #openstack-swift | 19:34 | |
| *** annegentle has quit IRC | 19:36 | |
| *** annegent_ has joined #openstack-swift | 19:37 | |
| clayg | where did I add Stub Reponses? | 19:54 |
| clayg | was that swiftclient? | 19:54 |
| clayg | oh yeah, that was swiftclient - nm | 19:55 |
| notmyname | bah! python's getting in my way | 19:57 |
| notmyname | foo = lambda: raise Exception() # syntax error | 19:58 |
| *** bitblt has quit IRC | 19:58 | |
| clayg | notmyname: that's bs - bet it works in python3 | 19:58 |
| redbo | yeah, I bet raise is a function | 19:58 |
| notmyname | probably :-/ | 19:58 |
| redbo | So you can call it from a statement and not just an expression. Or whatever silly python sounding thing. | 19:59 |
| clayg | redbo: see what problems you run into if you let your language have exceptions? better to just use return codes and be done with it. | 20:00 |
| clayg | lol @ statckovermind -> y = lambda: (_ for _ in ()).throw(Exception('foobar')) | 20:01 |
| notmyname | fails in py33 too | 20:01 |
| notmyname | clayg: if I submit code that does that, please -2 it ;-) | 20:02 |
| zackmdavis | `raise` is not a function | 20:03 |
| redbo | you should make a raise function and just have it raise its argument | 20:03 |
| redbo | then you could use it for lambdas everywhere | 20:03 |
| clayg | redbo: no no no *print* is a function - obviously *raise* is a statement - don't be a plebeian | 20:04 |
| clayg | that's it we're making python4 | 20:06 |
| clayg | mulligan! | 20:06 |
| redbo | well first we need to go add "from __future__ import raise_function" to all of our files | 20:07 |
| clayg | redbo: oh you're right | 20:07 |
| clayg | from six import where_the_fuck_is_it_this_week | 20:08 |
| *** thebloggu has joined #openstack-swift | 20:09 | |
| vinsh | Is there some flag one could set in say object-server.conf to disable a specific service like maybe object-auditor? or just object-server? | 20:13 |
| clayg | you can't `f = lambda : yield 1` either - and oddly `def f(): <whatever you want>` if only it was pep8 :'( | 20:13 |
| clayg | vinsh: you just take out the config section - and it wont' start | 20:14 |
| vinsh | Yeah. +1 | 20:14 |
| clayg | or chmod -x the bin | 20:14 |
| *** amoturi has quit IRC | 20:15 | |
| *** tongli has quit IRC | 20:36 | |
| *** ccavanna has quit IRC | 20:37 | |
| hrou | Hey All quick FYI re: static web and keystone ... so at this point I have such that it is indeed going through keystone first (i.e. X-Web_mode: true). And in this setup it is working just fine | 20:45 |
| *** joeljwright has quit IRC | 20:45 | |
| vinsh | Was maybe a rogue container on our end? | 20:46 |
| hrou | http://paste.openstack.org/show/397006/ | 20:46 |
| hrou | So 'test me' is the content of index.html | 20:46 |
| hrou | vinsh: I think in your case you were deferring keystone though (i.e. not using 'X-Web-Mode: true') so I need to try that | 20:47 |
| hrou | A bit on and off but will try that next! | 20:47 |
| vinsh | :) | 20:47 |
| clayg | hrou: so vinsh specific setup was *just* `x-container-meta-web-listings: true` and the fail was html response even with x-auth-token | 20:48 |
| openstackgerrit | John Dickinson proposed openstack/swift: do container listing updates in another (green)thread https://review.openstack.org/189080 | 20:48 |
| notmyname | clayg: I think I took care of what we talked about yesterday ^ | 20:48 |
| vinsh | I don't know exactly what x-container was set to.. since we couldn't list or stat it. I didn't check on a container node to see what it had. | 20:49 |
| clayg | ok, i'll dust off my timeouter_server.py script and try again | 20:49 |
| hrou | clayg: Thanks let me try that I had - Meta Web-Listings: true | 20:49 |
| hrou | Meta Web-Error: error.html | 20:49 |
| hrou | Meta Web-Index: index.html | 20:49 |
| hrou | - all set. | 20:49 |
| clayg | vinsh: well could you try swift stat again now and see if anything else was set? you made it sound like you just turned off that one key and it worked | 20:50 |
| vinsh | yeah. that was all i turned off and it worked. | 20:50 |
| clayg | ... err `swift post -m web-listings:` and it started working | 20:50 |
| hrou | Yep he showed me the output, he had nothing static web set at all at this point; so you must have been right clayg i.e. just web-listing | 20:51 |
| vinsh | exactly | 20:51 |
| clayg | so that seems to be the relevant value - the x-web-mode headers, and the index.html with x-container-meta-index-blah-blah is not strictly the minimum test case for the issue you observed | 20:51 |
| clayg | hrou: ok - but from your keystone test that works? | 20:51 |
| clayg | hrou: ok - it'll have to stay unconfirmed for awhile :'( | 20:51 |
| clayg | unless someone else has any idea what to try? | 20:52 |
| *** robefran has quit IRC | 20:52 | |
| *** lyrrad has quit IRC | 20:52 | |
| hrou | Still trying ; ) As you said, I'm not repro'ing 100% yet (but yea in general it looks like works so its not a rampant issue) more soon. | 20:52 |
| *** lyrrad has joined #openstack-swift | 20:54 | |
| hrou | vinsh, you didn't set 'X-Web-Mode' did you ? Actually you were just using the swift client. | 21:02 |
| vinsh | Yeah, just the client | 21:03 |
| *** annegent_ has quit IRC | 21:12 | |
| *** jasondotstar has joined #openstack-swift | 21:22 | |
| clayg | unless someone else has any idea what to try? | 21:32 |
| *** jkugel has quit IRC | 21:33 | |
| *** annegentle has joined #openstack-swift | 21:34 | |
| hrou | I have an idea, still working through it, at this point I can only get it to work with X-Web-mode (which we know vinsh) is not using, this is because keystone auth is failing without this (which makes sense, the odd part being I have delay_auth_decision = True). | 21:35 |
| notmyname | hrou: that's the same that vinsh had | 21:35 |
| *** annegentle has quit IRC | 21:35 | |
| hrou | I didn't think vinsh had x-web-mode set ? | 21:35 |
| *** annegent_ has joined #openstack-swift | 21:35 | |
| hrou | ** sorry I didn't mean not working => the failure vinsh saw. | 21:36 |
| *** jasondot_ has joined #openstack-swift | 21:36 | |
| hrou | ** I meant I was simply getting 403 on the container listing. | 21:36 |
| *** jasondotstar has quit IRC | 21:39 | |
| *** kragniz has quit IRC | 22:01 | |
| *** kragniz has joined #openstack-swift | 22:02 | |
| *** jodah has quit IRC | 22:05 | |
| *** thebloggu has quit IRC | 22:06 | |
| *** minwoob has quit IRC | 22:08 | |
| *** jodah has joined #openstack-swift | 22:11 | |
| *** alpha_ori has quit IRC | 22:11 | |
| *** alpha_ori has joined #openstack-swift | 22:14 | |
| mattoliverau | morning | 22:21 |
| *** jrichli has quit IRC | 22:24 | |
| *** mgarza has quit IRC | 22:24 | |
| notmyname | yo | 22:25 |
| *** nadeem has quit IRC | 22:34 | |
| *** ujjain has quit IRC | 22:34 | |
| peluse | hola | 22:37 |
| *** ujjain has joined #openstack-swift | 22:37 | |
| *** ujjain has joined #openstack-swift | 22:37 | |
| openstackgerrit | Michael Barton proposed openstack/swift: go: wait for availability on REPLICATE https://review.openstack.org/204292 | 22:46 |
| *** haomaiwang has quit IRC | 22:49 | |
| *** lyrrad has quit IRC | 23:01 | |
| *** km has joined #openstack-swift | 23:01 | |
| *** lyrrad has joined #openstack-swift | 23:04 | |
| *** jasondot_ has quit IRC | 23:12 | |
| *** jasondotstar has joined #openstack-swift | 23:17 | |
| openstackgerrit | OpenStack Proposal Bot proposed openstack/swift: Updated from global requirements https://review.openstack.org/88736 | 23:17 |
| *** DericHorn-HP has joined #openstack-swift | 23:18 | |
| *** kei_yama has joined #openstack-swift | 23:20 | |
| clayg | hrou: oh right, to test the w/o x-auth-token case you need to also `swift post container -r .r:*,.rlistings` | 23:23 |
| clayg | notmyname: patch 204292 <- the go people get all the cool stuff! | 23:26 |
| patchbot | clayg: https://review.openstack.org/#/c/204292/ | 23:26 |
| clayg | and all we have is stupid lambda's that don't even support raise statements | 23:26 |
| *** zaitcev has quit IRC | 23:27 | |
| *** david-lyle has quit IRC | 23:28 | |
| *** lyrrad has quit IRC | 23:40 | |
| *** robefran has joined #openstack-swift | 23:44 | |
| *** ho has joined #openstack-swift | 23:45 | |
| *** jasondotstar has quit IRC | 23:47 | |
| hrou | clayg, thanks ! That could have been it, Will give that a go in a bit (once I have access to my env again ; ) | 23:49 |
| *** jasondotstar has joined #openstack-swift | 23:50 | |
| *** wbhuber has quit IRC | 23:51 | |
| ho | good morning! | 23:53 |
| *** gyee has joined #openstack-swift | 23:54 | |
| mattoliverau | ho: morning | 23:58 |
| *** jlhinson has quit IRC | 23:59 | |
| ho | mattoliverau: morning! | 23:59 |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!