| *** zackf1 has quit IRC | 00:00 | |
| openstackgerrit | A change was merged to openstack/swift: Make swift-recon usable on hosts without IPv6 https://review.openstack.org/67776 | 00:07 |
|---|---|---|
| *** _bluev has quit IRC | 00:09 | |
| *** Midnightmyth has quit IRC | 00:17 | |
| *** matsuhashi has joined #openstack-swift | 00:24 | |
| openstackgerrit | paul luse proposed a change to openstack/swift: Replace Policy Index with Policy Name in Response Headers https://review.openstack.org/70265 | 00:25 |
| *** _bluev has joined #openstack-swift | 00:28 | |
| *** _bluev has quit IRC | 00:28 | |
| *** tnewsome has quit IRC | 00:40 | |
| *** pheadron1 has joined #openstack-swift | 00:51 | |
| *** pheadron has quit IRC | 00:54 | |
| *** shri has quit IRC | 01:21 | |
| *** shri has joined #openstack-swift | 01:22 | |
| *** shri has quit IRC | 01:22 | |
| *** haomaiwang has joined #openstack-swift | 01:30 | |
| *** haomaiwa_ has quit IRC | 01:34 | |
| *** nosnos has joined #openstack-swift | 01:41 | |
| *** pheadron1 has quit IRC | 01:56 | |
| openstackgerrit | A change was merged to openstack/swift: Add Storage Policy Support to the Updater https://review.openstack.org/64737 | 02:17 |
| *** matsuhashi has quit IRC | 02:34 | |
| *** matsuhas_ has joined #openstack-swift | 02:37 | |
| openstackgerrit | A change was merged to openstack/swift: Container Sync: Simple HTTP Proxy load balancing https://review.openstack.org/66989 | 02:38 |
| *** csd has quit IRC | 02:40 | |
| *** Diddi has quit IRC | 02:54 | |
| *** Diddi has joined #openstack-swift | 02:55 | |
| *** markd has quit IRC | 03:06 | |
| *** matsuhashi has joined #openstack-swift | 03:07 | |
| *** matsuhas_ has quit IRC | 03:08 | |
| *** matsuhashi has quit IRC | 03:08 | |
| *** matsuhashi has joined #openstack-swift | 03:20 | |
| *** shri has joined #openstack-swift | 03:21 | |
| *** byeager has joined #openstack-swift | 03:33 | |
| *** matsuhashi has quit IRC | 03:34 | |
| *** matsuhashi has joined #openstack-swift | 03:35 | |
| *** matsuhashi has quit IRC | 03:43 | |
| *** matsuhashi has joined #openstack-swift | 03:50 | |
| *** gyee has quit IRC | 04:04 | |
| openstackgerrit | A change was merged to openstack/swift: Add Storage Policy Support to the Auditor https://review.openstack.org/64461 | 04:07 |
| notmyname | portante: something going on with grenade that we need to look at? | 04:12 |
| *** gvernik has quit IRC | 04:16 | |
| *** Anju has quit IRC | 04:23 | |
| portante | notmyname: when a HUP signal is sent to the havana codebase all the WSGI server's get their shorts in a bunch | 04:24 |
| portante | I had looked into that in the past, and I think we solved it by happenstance on master, but have to refresh my memory (doing that now) | 04:25 |
| madhuri | portante: Hi, could you please tell me what does fast POST method means? | 04:26 |
| portante | madhuri: | 04:26 |
| portante | by default | 04:26 |
| portante | POST means COPY+new headers | 04:27 |
| madhuri | mast sunna diya mene chilke ko, bola k me ab presentation nahi dungi. kisi ko interst hi nahi h. already 6 7 bar cancel ho chuka h | 04:27 |
| portante | fast POST means just add new headers over old | 04:27 |
| portante | not abla | 04:27 |
| *** judd7 has quit IRC | 04:28 | |
| madhuri | portante: does it save metadata in .meta file | 04:28 |
| madhuri | portante: what is the need of .meta file? | 04:30 |
| notmyname | madhuri: fast POST uses a .meta file on disk | 04:31 |
| *** Anju has joined #openstack-swift | 04:32 | |
| madhuri | it means when we add any custom metadata, it stores the metadata in .meta file rather than setting xattr to .data file? | 04:35 |
| portante | madhuri: yes | 04:40 |
| portante | so to construct an object's meta data, the xattrs from the .data a read, then the xattrs from the .meta are read, and merge together where the user settable xattrs are taken from the .meta (content-length, etag, content-type, name are taken only from the .data, I believe) | 04:41 |
| *** tongli has quit IRC | 04:49 | |
| portante | notmyname: regarding that grenade thing | 04:54 |
| notmyname | yes? | 04:54 |
| portante | remember the infinite loop logging errors when the /dev/log file get yanked? | 04:55 |
| notmyname | yup | 04:55 |
| portante | the proxy server would spin in certain cases | 04:55 |
| portante | that is what this appears to be | 04:55 |
| notmyname | so an ordering issue then | 04:56 |
| portante | if you look at that stack trace, emit() is calling self.handleError(), which just calls into the same code path | 04:57 |
| portante | I don't think this is an ordering issue, though | 04:58 |
| portante | I don't think we construct our logging handler correctly | 04:58 |
| notmyname | portante: thanks for looking into that. FWIW (as I'm sure you confirmed too), this doesn't happen on my SAIO with `swift-init proxy restart --graceful` or `kill -HUP proxypid` | 04:58 |
| notmyname | what should we change? | 04:58 |
| portante | yes, because no error occurs writing to the socket the logger has | 04:59 |
| portante | I think swift/common/utils.py line 582, the write() method of that class | 05:00 |
| portante | utils installs a default logger,and the stdlib traceback.py ends up invoking it | 05:01 |
| notmyname | not on master | 05:01 |
| *** byeager has quit IRC | 05:01 | |
| notmyname | err...that line number isn't right for master | 05:01 |
| *** byeager has joined #openstack-swift | 05:02 | |
| portante | 732 in master | 05:03 |
| *** ppai has joined #openstack-swift | 05:05 | |
| madhuri | portante: why swift does so? why doesn't it go to set xattr to the .data file? | 05:06 |
| *** haomaiwang has quit IRC | 05:09 | |
| *** haomaiwang has joined #openstack-swift | 05:10 | |
| portante | madhuri: because it needs to read the original metadata and merge it with the new metadata and that operation is not atomic (at least, that is how I understand it) | 05:10 |
| portante | notmyname: the reason why this loop happens is because we install a utils.py LoggerFileObject instance in sys.stderr (see capture_stdio()) | 05:11 |
| notmyname | portante: madhuri: also because the actual name of the .data and .meta files matter for conflict resolution | 05:11 |
| madhuri | portante: but it can be set to .data file also | 05:12 |
| portante | madhuri: what do you mean? | 05:12 |
| portante | what is the "it" you refer to above? the original metadata of the file? | 05:13 |
| madhuri | why swift write custom metadata to .meta file? The custom metadata can be set to .data file also | 05:13 |
| madhuri | portante: yes | 05:13 |
| *** byeager has quit IRC | 05:14 | |
| portante | the metadata is stored as a python dictionary, pickled and then broken up into 254 byte chunks so that they will opportunistically fit into XFS's short-form xattrs within an inode | 05:15 |
| portante | at least, that was the originally intent as I understand it | 05:15 |
| notmyname | madhuri: you can't coordinate overwrites of the .data files (ie fsyncs) across the multiple replicas in the cluster. so you can't guarantee that the data is written the same way when there are multiple concurrent writers. so you have to write new metadata into a new file. slow POST (the default) does this via a copy to a new .data file. the fast POST option writes to a new .meta file | 05:16 |
| portante | however, today, we no longer recommend 1024 byte inodes, the default 256 byte inodes are now recommended | 05:16 |
| portante | notmyname: regarding the logging ... | 05:17 |
| notmyname | portante: ya, trying to trace through that code. I like it, but it's somewhat....convoluted | 05:17 |
| *** nshaikh has joined #openstack-swift | 05:17 | |
| notmyname | portante: do you have the stack trace link handy? | 05:19 |
| portante | because we replace sys.stderr and sys.stdout with a LoggerFileObject() that wraps the logger in use by wsgi, when the low-level python code (at /usr/lib/python2.7/logging/__init__.py:801) handles it | 05:19 |
| portante | an infinite loop is created | 05:19 |
| portante | posting the traceback ... | 05:19 |
| portante | https://gist.github.com/portante/8726954 | 05:20 |
| madhuri | notmyname: portante: could you please tell me is there any buffer management in swift? In case while writing data, the disk failure happens then how does the data is recovered? | 05:23 |
| notmyname | madhuri: I'd suggest reading through https://swiftstack.com/openstack-swift/architecture/ | 05:23 |
| notmyname | madhuri: also, https://swiftstack.com/blog/2012/09/13/how-openstack-swift-handles-hardware-failures/ is an overview of failure handling | 05:24 |
| notmyname | portante: why does this happen on SIGHUP vs any traceback? | 05:25 |
| portante | because I think the grenade jobs are nuking the /dev/log unix pipe by shutting down rsyslogd or something | 05:28 |
| portante | I think they are doing that before or in parallel with the shutdown of all the swift processes (and possibly other things) | 05:28 |
| portante | on errors sending to a unix socket, we appear to have an infinite loop | 05:29 |
| notmyname | I suppose "you're doing it wrong" isn't a helpful response to the bug? ;-) | 05:29 |
| portante | :) | 05:29 |
| portante | I think they are | 05:29 |
| notmyname | right, but that's kinda the whole point of swift: keep working when somebody or some thing does "it" wrong | 05:29 |
| portante | before restarting syslogd, they should everything else down, make sure they shutdown cleaning, then restart syslogd and bring up the new stuff | 05:30 |
| portante | ye | 05:30 |
| portante | s | 05:30 |
| portante | so I feel we should fix this loop | 05:30 |
| portante | don't know how just yet, but we should fix it | 05:30 |
| notmyname | ya, seems reasonable. here's hoping the fix is also reasonable ;-) | 05:32 |
| madhuri | notmyname: portante: thanks alot | 05:32 |
| portante | madhuri: you are welcome | 05:32 |
| notmyname | zaitcev: did that thing you just tweeted happen to you? that's hilarious | 05:33 |
| zaitcev | notmyname: Yes, it's some kind of spam. | 05:33 |
| zaitcev | notmyname: sent to zaitcev@redhat.com, no less. | 05:34 |
| notmyname | lol | 05:35 |
| zaitcev | notmyname: About that prohibition on copying from Oslo, how hard is it? I stole the part that invokes sys._current_frames() from there, which Solly in turn stole from Dan Berrange's prototype. | 05:40 |
| zaitcev | I mean I could've taken it from stackoverflow but this seems more proper somehow. | 05:41 |
| notmyname | zaitcev: well, the method is "we'll automatically copy/paste code into your repo and then you need to accept it after a bot proposes every new change that happens in the other repo" | 05:42 |
| notmyname | so it's more an issue with code in the repo that isn't actually managed by the maintainers of that repo. that's the bad part, and that's why I don't think we'll ever accept oslo-incubator code. make a library/module, and then the conversation can talk | 05:45 |
| *** byeager has joined #openstack-swift | 05:45 | |
| zaitcev | thanks... I think I'm safe, then | 05:46 |
| *** ChanServ sets mode: +v zaitcev | 05:57 | |
| *** byeager has quit IRC | 06:00 | |
| zaitcev | aww, I'm going to bed soon | 06:10 |
| zaitcev | Also, if I identify to services, I'll get the voice normally. I just avoid doing it, since my IRC is not SSL, so passwords flies like in telnet. | 06:11 |
| notmyname | ah ok | 06:11 |
| notmyname | screen + irssi? | 06:11 |
| notmyname | zaitcev: I figure a microsoft expert like yourself would have solved ssl in irc by now ;-) | 06:12 |
| *** odyssey4me has joined #openstack-swift | 06:32 | |
| *** zaitcev has quit IRC | 06:35 | |
| *** zaitcev has joined #openstack-swift | 06:35 | |
| *** ChanServ sets mode: +v zaitcev | 06:36 | |
| *** zaitcev has quit IRC | 06:42 | |
| *** haomai___ has joined #openstack-swift | 07:01 | |
| *** haomaiwang has quit IRC | 07:01 | |
| openstackgerrit | Kota Tsuyuzaki proposed a change to openstack/swift: Add missing sample config of object-replicator https://review.openstack.org/69301 | 07:04 |
| *** shri has quit IRC | 07:14 | |
| *** psharma has joined #openstack-swift | 07:26 | |
| *** shri has joined #openstack-swift | 07:52 | |
| *** shri has quit IRC | 07:53 | |
| *** matsuhashi has quit IRC | 08:37 | |
| *** sileht has quit IRC | 08:42 | |
| *** foexle has joined #openstack-swift | 08:43 | |
| *** sileht has joined #openstack-swift | 08:47 | |
| *** nacim has joined #openstack-swift | 08:47 | |
| *** haomai___ has quit IRC | 08:47 | |
| *** haomaiwang has joined #openstack-swift | 08:48 | |
| chmouel | cschwede: about https://review.openstack.org/#/c/62584/4/test/unit/bin/test_recon_client.py | 08:49 |
| chmouel | I commented on the review there | 08:49 |
| chmouel | but I think we should just go ahead and move the code to swift.cli directory | 08:49 |
| cschwede | chmouel: thanks for the review, looking | 08:49 |
| chmouel | pandemicsyn: ^^^ what do you think since you were the original author of swift recon | 08:50 |
| cschwede | chmouel: hmm, yes, that makes a lot of sense to me | 08:51 |
| cschwede | chmouel: doing this (only importing in bin/* from swift.cli) would make it a lot easier to test all that stuff | 08:51 |
| chmouel | cschwede: yeah right that would be the idea | 08:51 |
| cschwede | chmouel: without all the workarounds importing files with dashes in their name or missing ".py", suppressing creating of bytecode files etc | 08:52 |
| chmouel | let's see if notmyname or others have something to say about it and we can start doing the move for bin/* and get the coverage report to start show them (at 0 at first i guess) | 08:52 |
| chmouel | cschwede: yeah that looked scary :) | 08:52 |
| cschwede | chmouel: to make it easier for reviewers, should this be separated into two patches? one patch for moving code out of bin, follow up for test? | 08:53 |
| cschwede | chmouel: it is scary :) | 08:53 |
| cschwede | chmouel: and it was a pain to get this working... | 08:53 |
| chmouel | cschwede: I prefer that way but the others swift reviewers one large patch | 08:53 |
| chmouel | I would move the bin/swift to swift.cli add a minimal test that just pass so coverage can show | 08:54 |
| chmouel | and start from there | 08:54 |
| chmouel | (minimal test that import the module and just pass for the first test) | 08:54 |
| openstackgerrit | A change was merged to openstack/python-swiftclient: Add capabilities option https://review.openstack.org/66442 | 08:55 |
| *** fbo_away is now known as fbo | 08:57 | |
| cschwede | chmouel: ok, will start with moving code and adding some basic test | 08:58 |
| chmouel | you may want to ping others on US time just in case if other core have something to say about it | 08:59 |
| chmouel | cschwede: ^ | 08:59 |
| cschwede | chmouel: yepp, will do that | 08:59 |
| chmouel | cschwede: cool thanks | 08:59 |
| chmouel | and that's it we right on time for the scrum | 08:59 |
| chmouel | oops wrong window | 08:59 |
| openstackgerrit | A change was merged to openstack/python-swiftclient: Updated from global requirements https://review.openstack.org/63714 | 09:10 |
| openstackgerrit | A change was merged to openstack/python-swiftclient: assertEquals is deprecated, use assertEqual https://review.openstack.org/67419 | 09:13 |
| chmouel | creiht: add to the swift agenda for the next meeting the item to talk about py3 since you mentionned it https://wiki.openstack.org/wiki/Meetings/Swift | 09:27 |
| *** xga has joined #openstack-swift | 09:35 | |
| *** matsuhashi has joined #openstack-swift | 09:44 | |
| *** xga has quit IRC | 09:49 | |
| *** xga has joined #openstack-swift | 09:49 | |
| *** xga has quit IRC | 09:51 | |
| *** xga has joined #openstack-swift | 09:52 | |
| *** matsuhashi has quit IRC | 10:05 | |
| *** SnowDust has joined #openstack-swift | 10:12 | |
| *** matsuhashi has joined #openstack-swift | 10:16 | |
| *** gvernik_ has joined #openstack-swift | 10:16 | |
| *** ccorrigan has quit IRC | 10:26 | |
| *** Midnightmyth has joined #openstack-swift | 10:34 | |
| *** gvernik_ has quit IRC | 10:36 | |
| *** xga_ has joined #openstack-swift | 10:37 | |
| *** nacim_ has joined #openstack-swift | 10:38 | |
| *** haomaiwang has quit IRC | 10:39 | |
| *** nacim has quit IRC | 10:39 | |
| *** xga has quit IRC | 10:40 | |
| *** gvernik_ has joined #openstack-swift | 10:48 | |
| *** gvernik_ has quit IRC | 11:10 | |
| *** haomaiwa_ has joined #openstack-swift | 11:20 | |
| *** xga_ has quit IRC | 11:28 | |
| *** zanc has joined #openstack-swift | 11:52 | |
| *** zanc has joined #openstack-swift | 11:52 | |
| *** bada_ has joined #openstack-swift | 12:21 | |
| *** bada has quit IRC | 12:24 | |
| *** ppai has quit IRC | 12:33 | |
| *** xga has joined #openstack-swift | 12:40 | |
| *** kris_h has joined #openstack-swift | 12:59 | |
| *** kris_h has left #openstack-swift | 12:59 | |
| *** kris_h has joined #openstack-swift | 13:06 | |
| *** Trixboxer has joined #openstack-swift | 13:07 | |
| *** odyssey4me has quit IRC | 13:19 | |
| *** haomaiwa_ has quit IRC | 13:31 | |
| *** haomaiwang has joined #openstack-swift | 13:32 | |
| *** psharma has quit IRC | 13:36 | |
| *** matsuhashi has quit IRC | 13:41 | |
| *** gvernik_ has joined #openstack-swift | 13:43 | |
| *** kris_h has quit IRC | 13:44 | |
| *** matsuhashi has joined #openstack-swift | 13:50 | |
| *** matsuhashi has quit IRC | 13:57 | |
| *** matsuhashi has joined #openstack-swift | 14:04 | |
| *** haomai___ has joined #openstack-swift | 14:14 | |
| *** haomaiwang has quit IRC | 14:17 | |
| *** zanc has quit IRC | 14:23 | |
| openstackgerrit | Christian Schwede proposed a change to openstack/swift: Add some tests for bin/swift-recon https://review.openstack.org/70352 | 14:26 |
| openstackgerrit | Christian Schwede proposed a change to openstack/swift: Add some tests for bin/swift-recon https://review.openstack.org/62584 | 14:28 |
| *** matsuhashi has quit IRC | 14:28 | |
| *** mrsnivvel has quit IRC | 14:29 | |
| *** zanc has joined #openstack-swift | 14:34 | |
| *** SnowDust has quit IRC | 14:34 | |
| *** nosnos has quit IRC | 14:37 | |
| *** xga has quit IRC | 14:37 | |
| *** gvernik_ has quit IRC | 14:38 | |
| cschwede | chmouel: talking about https://review.openstack.org/#/c/62584/ : do you mean to put swift-recon and swift-recon-cron into swift.cli.recon? | 14:40 |
| chmouel | cschwede: oh i forgot about recon cron, can they share same code between the twos? | 14:41 |
| chmouel | or is it completely different | 14:41 |
| cschwede | chmouel: they have both a main() | 14:41 |
| cschwede | swift-recon-cron is only a few lines, but they don't share code | 14:42 |
| cschwede | chmouel: ah nevermind, i know what you mean. yes, that makes a lot of sense | 14:42 |
| * cschwede needs a coffee break | 14:43 | |
| chmouel | i think i may just do that as well :) | 14:43 |
| openstackgerrit | Christian Schwede proposed a change to openstack/swift: Add some tests for bin/swift-recon https://review.openstack.org/62584 | 14:45 |
| *** russellb is now known as rustlebee | 15:02 | |
| *** nacim_ has quit IRC | 15:03 | |
| *** nacim has joined #openstack-swift | 15:12 | |
| *** ccorrigan has joined #openstack-swift | 15:19 | |
| *** foexle has quit IRC | 15:21 | |
| *** foexle has joined #openstack-swift | 15:21 | |
| *** nacim has quit IRC | 15:23 | |
| *** gvernik_ has joined #openstack-swift | 15:34 | |
| *** xga has joined #openstack-swift | 15:34 | |
| *** nacim has joined #openstack-swift | 15:35 | |
| openstackgerrit | Sergio Cazzolato proposed a change to openstack/python-swiftclient: Bump hacking to 0.8 https://review.openstack.org/59790 | 15:35 |
| *** nshaikh has left #openstack-swift | 15:40 | |
| *** gvernik_ has quit IRC | 15:48 | |
| openstackgerrit | Christian Schwede proposed a change to openstack/swift: Add some tests for bin/swift-recon https://review.openstack.org/62584 | 15:51 |
| *** gvernik_ has joined #openstack-swift | 15:53 | |
| openstackgerrit | Christian Schwede proposed a change to openstack/swift: Add some tests for bin/swift-recon https://review.openstack.org/62584 | 15:53 |
| *** IRTermite has joined #openstack-swift | 16:08 | |
| *** leibinet has joined #openstack-swift | 16:11 | |
| *** csd has joined #openstack-swift | 16:17 | |
| creiht | chmouel: thx | 16:17 |
| *** rook][ has joined #openstack-swift | 16:21 | |
| *** zaitcev has joined #openstack-swift | 16:22 | |
| *** ChanServ sets mode: +v zaitcev | 16:22 | |
| *** hurricanerix has joined #openstack-swift | 16:24 | |
| briancline | in its current incarnation, there's no way of guaranteeing that a replica will exist in all regions of a global cluster, right? | 16:25 |
| briancline | assuming replica count <= region count | 16:26 |
| briancline | oops. i mean where replicas = regions | 16:26 |
| *** leibinet has left #openstack-swift | 16:32 | |
| *** byeager has joined #openstack-swift | 16:32 | |
| creiht | notmyname: do you know if they intend to flip the switch to be v3 auth only for the next release? | 16:33 |
| creiht | asking for a friend on another project | 16:33 |
| creiht | chmouel: -^ ? | 16:33 |
| chmouel | creiht: next release for juno it will just be deprecated not only | 16:34 |
| creiht | k | 16:34 |
| creiht | thx | 16:34 |
| *** gvernik_ has quit IRC | 16:44 | |
| dfg | chmouel: i tested that patch with safari and it seemed to work. i was using tempauth though | 16:51 |
| dfg | chmouel: i am going to make some changes to it though. i'm still not restricting cors if they have a specific origin(s) they want set. it never did that before but I want it to now atleast for x-web-true requests. | 16:52 |
| chmouel | dfg: so the x-web-true is only on staticweb requests right? | 16:52 |
| dfg | chmouel: on the other hand do we really want regualr (non-option) origin calls to just magically work for anything? i left that alone because I didn't want to break anybody but its kinda a bug | 16:53 |
| dfg | chmouel: ya, in swift. i use them for SOS too. but the x-web-true did change if you didn't have cors set up at all | 16:54 |
| chmouel | dfg: are you sure that works? maybe i missed something but it should not reply back | 16:54 |
| dfg | chmouel: so what happened? | 16:55 |
| dfg | chmouel: nm- let me just put up the new patch | 16:55 |
| chmouel | dfg: yep | 16:55 |
| notmyname | good morning everyone | 16:56 |
| chmouel | dfg: cause ia m trying to understand what you means :) and i think will be better with the patch | 16:56 |
| chmouel | notmyname: good morning | 16:56 |
| notmyname | briancline: if replicas == regions, then you'll have a replica in each region | 16:56 |
| dfg | chmouel: ok. | 16:56 |
| dfg | chmouel: are you expecting cors to work with no container metadata set? | 17:03 |
| chmouel | dfg: yes, people would set it up in proxy-server.conf | 17:05 |
| *** gyee has joined #openstack-swift | 17:05 | |
| dfg | chmouel: hm- well that should still work but I didn't test it with the proxy-server.conf stuff- so it probably doesn't :) | 17:10 |
| dfg | chmouel: ok- if they don't have container metadata cors and they don't have it set in the proxy-server.conf do you still want cors to work. because right now it does. not for OPTIONS calls but for simple cross origin requests | 17:12 |
| chmouel | ah well but you need the OPTIONS call to work for PUT right? | 17:13 |
| chmouel | dfg: | 17:13 |
| dfg | the cors_validation just adds in th origin that was passed to it all the time. now- can I just consider that a bug? | 17:13 |
| dfg | ya- for puts- but not for GETs | 17:13 |
| dfg | chmouel: ^^ | 17:13 |
| chmouel | hold on sorry on a video call at the same time :( | 17:14 |
| dfg | its alright- i'll go make coffee | 17:14 |
| *** xga has quit IRC | 17:29 | |
| *** gyee has quit IRC | 17:30 | |
| openstackgerrit | A change was merged to openstack/python-swiftclient: Install manpage in share/man/man1 instead of man/man1 https://review.openstack.org/67813 | 17:32 |
| *** byeager has quit IRC | 17:44 | |
| *** fbo is now known as fbo_away | 17:46 | |
| *** nacim has quit IRC | 17:49 | |
| notmyname | FYI, unittest failure in the gate https://jenkins03.openstack.org/job/gate-swift-python27/309/console | 17:55 |
| * creiht blames clayg :) | 17:59 | |
| creiht | I'll see if I can poke at it | 17:59 |
| creiht | notmyname: that test seems to fail intermittently | 18:03 |
| creiht | in jenkins | 18:03 |
| creiht | of course I can never get it to fail | 18:03 |
| *** IRTermite has left #openstack-swift | 18:03 | |
| *** IRTermite has joined #openstack-swift | 18:07 | |
| glange | you fixed it, good job | 18:10 |
| *** wer has quit IRC | 18:10 | |
| creiht | heh | 18:11 |
| *** wer has joined #openstack-swift | 18:11 | |
| notmyname | creiht: thanks for looking. here's some logstash results of other test runs too http://goo.gl/OpR38z | 18:23 |
| creiht | clayg: but if you have time to look, I would mind your eyes looking at it as well :) | 18:28 |
| clayg | creiht: sure no worries - but... what now? | 18:37 |
| creiht | hehe | 18:38 |
| creiht | clayg: isn't this the test that you added to my memcache path? | 18:38 |
| creiht | to test the timeouts | 18:38 |
| clayg | oh yeah sure, those all suck - i'll poke at it | 18:38 |
| creiht | haha | 18:38 |
| creiht | so the interesting thing that I have found | 18:39 |
| creiht | if I comment out the sleep before the assert that fails for jenkins, I get the same failure | 18:39 |
| creiht | it seems like for some reason things aren't yielding right | 18:39 |
| creiht | but I dunno... this test is kinda hard to follow ;P | 18:41 |
| *** shri has joined #openstack-swift | 18:41 | |
| creiht | clayg: actually, why would the pending go from 10 to 8? | 18:48 |
| creiht | should we maybe get rid of the sleep there, and change the assert to 10? | 18:48 |
| creiht | if I do that, the test passes here | 18:48 |
| creiht | not certain if that is really the desired behavior of the test though | 18:48 |
| *** markd has joined #openstack-swift | 18:50 | |
| clayg | creiht: sorry i'm in a meeting, portante ended up re-writing the test that eventually got merged I think | 18:52 |
| creiht | clayg: no worries | 18:53 |
| openstackgerrit | Chuck Thier proposed a change to openstack/swift: Attempt to fix periodic memcache timeout test https://review.openstack.org/70400 | 18:59 |
| creiht | notmyname: not really sure if that will fix the issue | 19:00 |
| creiht | but I *think* the test makes a little more sense now | 19:00 |
| creiht | clayg, portante, if you can get a min to look and let me know what you think -^ | 19:00 |
| *** byeager has joined #openstack-swift | 19:01 | |
| *** byeager has quit IRC | 19:02 | |
| *** byeager_ has joined #openstack-swift | 19:02 | |
| *** SnowDust has joined #openstack-swift | 19:05 | |
| *** gyee has joined #openstack-swift | 19:13 | |
| *** SnowDust has quit IRC | 19:15 | |
| *** IRTermite has quit IRC | 19:19 | |
| *** xga has joined #openstack-swift | 19:26 | |
| *** gvernik_ has joined #openstack-swift | 19:26 | |
| *** SnowDust has joined #openstack-swift | 19:28 | |
| *** xga has quit IRC | 19:30 | |
| *** IRTermite has joined #openstack-swift | 19:31 | |
| cschwede | swifterdarrell: ping | 19:41 |
| *** zackf has joined #openstack-swift | 19:44 | |
| *** byeager_ has quit IRC | 19:47 | |
| openstackgerrit | Chuck Thier proposed a change to openstack/swift: Attempt to fix periodic memcache timeout test https://review.openstack.org/70400 | 19:53 |
| creiht | gah | 19:53 |
| *** perfectsine has joined #openstack-swift | 19:53 | |
| openstackgerrit | Chuck Thier proposed a change to openstack/swift: Attempt to fix periodic memcache timeout test https://review.openstack.org/70400 | 19:54 |
| gholt | Perfect way to rejoin the team. | 19:54 |
| creiht | hah | 19:55 |
| * creiht wonders if anyone has created a openstack gerrit commit cheat sheet | 19:55 | |
| gholt | Hah, that change is pretty much like the one I did this morning: https://github.com/pandemicsyn/fgerrit/pull/19 | 19:55 |
| creiht | lol | 19:56 |
| *** byeager has joined #openstack-swift | 19:58 | |
| *** gvernik_ has quit IRC | 20:01 | |
| *** judd7 has joined #openstack-swift | 20:03 | |
| *** byeager has quit IRC | 20:05 | |
| *** bsdkurt has quit IRC | 20:08 | |
| notmyname | yay, creiht's back | 20:10 |
| rook][ | Hello all, has anyone setup a read-only role for swift in keystone? (not finding any examples via google) | 20:14 |
| torgomatic | rook][: you might have more luck asking in #openstack-dev; Swift has pluggable auth systems, so a lot of folks in here don't use Keystone at all | 20:17 |
| koolhead17 | rook][: https://github.com/openstack/swift/blob/master/swift/common/middleware/keystoneauth.py << Let me know if this might help :) | 20:17 |
| torgomatic | I mean, some do, but historically the amount of Keystone knowledge in here has been small | 20:17 |
| rook][ | thanks guys | 20:18 |
| koolhead17 | torgomatic: notmyname hellos | 20:18 |
| notmyname | koolhead17: hi | 20:18 |
| * koolhead17 wonders why Super Bowl is everywhere today!! | 20:18 | |
| *** gvernik_ has joined #openstack-swift | 20:22 | |
| creiht | heh | 20:24 |
| peluse | torgomatic: do you think we should require policy names? Now that we are doing /info and I just put a patch up to supply policy name instead of index on HEAD, blank policy names can be a little confusing | 20:30 |
| notmyname | peluse: yes, it should have a name :-) | 20:31 |
| torgomatic | peluse: yeah, if it's a client-facing thing, and clients use names, we should require names | 20:31 |
| peluse | Thanks guys. I'll submit a patch to make it required | 20:32 |
| briancline | generally speaking, despite the fact they serve mostly different purposes, is there any chance of container sync being deprecated / obliterated in the future, due to the option of using global clusters? | 20:32 |
| notmyname | peluse: but internal use of it should all be the index (as it is currently AFAIK). | 20:33 |
| notmyname | briancline: I doubt it | 20:33 |
| notmyname | briancline: reason being, they serve different purposes :-) | 20:33 |
| torgomatic | briancline: I can give you an upper bound of about 5 Gyears on the lifetime of container sync | 20:33 |
| notmyname | torgomatic: the 5th busy beaver number? | 20:34 |
| torgomatic | after that, the sun's expansion will destroy the earth, so no traces of container sync will remain :) | 20:34 |
| torgomatic | barring Swift-in-space, of course | 20:34 |
| torgomatic | multiglobal clusters? | 20:34 |
| notmyname | torgomatic: lol, interplanetary container sync | 20:34 |
| notmyname | torgomatic: actually, letterj was working on getting swift in space. literally | 20:34 |
| briancline | notmyname: torgomatic: good, I suspected as much, but wanted to confirm :) | 20:35 |
| torgomatic | notmyname: at that point, we'll have to support RFC 6921 | 20:35 |
| briancline | reduced redundancy storage on the moon would be nice | 20:35 |
| *** bsdkurt has joined #openstack-swift | 20:38 | |
| *** byeager has joined #openstack-swift | 20:40 | |
| *** byeager has quit IRC | 20:41 | |
| *** csd has quit IRC | 20:42 | |
| *** SnowDust has quit IRC | 20:45 | |
| briancline | is the process of syncing the actual objects in a container sync-enabled container a serial process? or does that depend on the number of container servers you're running | 20:59 |
| *** gvernik_ has quit IRC | 21:05 | |
| *** byeager has joined #openstack-swift | 21:05 | |
| *** csd has joined #openstack-swift | 21:12 | |
| openstackgerrit | Samuel Merritt proposed a change to openstack/swift: Merge remote-tracking branch 'upstream/master' into feature/ec https://review.openstack.org/70442 | 21:17 |
| *** openstackgerrit has quit IRC | 21:53 | |
| *** openstackgerrit has joined #openstack-swift | 21:53 | |
| *** byeager has quit IRC | 22:03 | |
| *** csd has quit IRC | 22:04 | |
| *** dencaval has quit IRC | 22:11 | |
| *** rook][ has quit IRC | 22:19 | |
| *** csd has joined #openstack-swift | 22:32 | |
| *** Trixboxer has quit IRC | 22:32 | |
| *** byeager has joined #openstack-swift | 22:47 | |
| openstackgerrit | A change was merged to openstack/swift-bench: Add MANIFEST.in https://review.openstack.org/66620 | 23:01 |
| notmyname | wow that was fast | 23:01 |
| *** byeager has quit IRC | 23:04 | |
| *** byeager has joined #openstack-swift | 23:04 | |
| markd | hello, new to openstack, I have been looking at the swift tests, and am wondering if there is any documentation like a test plan or pydoc of what the tests do/test. I have been searching all the docs, but so far have not found any. | 23:21 |
| notmyname | markd: welcome! | 23:21 |
| notmyname | markd: I'd say that mostly the devs who write swift also run it in prod and also do qa for their respective clusters. and as such there isn't a particular test spec. the test names are pretty descriptive | 23:22 |
| notmyname | markd: there are 3 kinds of tests in swift | 23:22 |
| notmyname | markd: unit tests, functional tests, and probe tests | 23:23 |
| notmyname | markd: unit and functional tests are pretty self-explanatory. and you can look at the API docs to see what needs to be tested in the functional tests | 23:23 |
| markd | notmyname, yep, have run those... OK, so I'll just study the code. Personally, I put a doc block at the start of my tests to indicate what it is testing and what if any parameters it needs.' | 23:23 |
| markd | thanks for the quick reply! | 23:24 |
| notmyname | markd: probe tests are "white box" testing, in that the tests start and kill various processes and ensure that things still work in the system | 23:24 |
| markd | ah, good to know. | 23:24 |
| notmyname | markd: also, there's been some work going on in the openstack tempest project to add swift functional testing there. it's fairly disconnected from the tests in the swift repo, but you will see more tests there as time goes on | 23:25 |
| notmyname | essentially, the tempest tests are functional tests | 23:26 |
| markd | notmyname, yep, just getting started with the tempest stuff. I'll look into that as well. Cool. Thanks. | 23:26 |
| notmyname | markd: cool. are you going to be writing new tests? either for tempest or int he swift codebase? | 23:26 |
| markd | notmyname, maybe... I work for Hitachi Data Systems and we are working on supporting the swift protocol (and other openstack protocols) for our product. Some of our changes may make it back to the core. I am a test guy so am looking at things from that perspective. | 23:28 |
| markd | So starting with the existing tests is a good first step. | 23:29 |
| notmyname | markd: any chance you know aaron martinez at HDS? | 23:30 |
| notmyname | markd: I sat next to him on a plane last week and we talked openstack and swift and HDS | 23:30 |
| markd | notmyname, hmmm, no I don't do you know what product he works on? I have only been with HDS for 4 months. | 23:30 |
| notmyname | ah, no worries. just curious if there was a connection :-) | 23:31 |
| markd | I'll make a note of it... and ask others.... I'm sure he is known. | 23:31 |
| markd | :-) | 23:31 |
| notmyname | markd: he's a sales engineer | 23:31 |
| notmyname | markd: so tell me more about HDS supporting swift | 23:32 |
| markd | ah, makes sense. | 23:32 |
| notmyname | markd: I have opinions on how this should be done ;-) | 23:32 |
| *** zackf has quit IRC | 23:32 | |
| markd | Well it appears that HDS is getting the openstack religion. I am new to the storage world, but HDS has a number of storage products. The one I am working on has not been released yet so I can't say much about it. We want our product to 'just work' with openstack. Our product would be a great fit for swift storage as we offer availability, partition tolerance AND consistency.... | 23:35 |
| markd | Right now just supporting NFS/Posix semantics, will be supporting CIFS. | 23:38 |
| notmyname | markd: for that kind of integration, HDS should be looking at how to use those storage volumes under Swift via the DiskFile abstraction | 23:41 |
| markd | OK, I will pass that on... just starting that effort so good to know. | 23:42 |
| notmyname | ya, unfortunately, friday afternoon at the end of the US business day isn't the best time to get a lot of active discussion :-) | 23:42 |
| markd | nope. In fact I gotta run. Thanks for the discussion and pointers! | 23:43 |
| *** byeager has quit IRC | 23:45 | |
| *** hurrican_ has joined #openstack-swift | 23:45 | |
| *** hurricanerix has quit IRC | 23:45 | |
| *** hurricanerix has joined #openstack-swift | 23:45 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!