| *** zackf has quit IRC | 00:06 | |
| *** shakamunyi has joined #openstack-swift | 00:11 | |
| *** pbojanic has quit IRC | 00:19 | |
| *** MooingLemur has joined #openstack-swift | 00:19 | |
| *** MooingLemur has quit IRC | 00:25 | |
| *** nosnos has joined #openstack-swift | 00:26 | |
| *** matsuhashi has joined #openstack-swift | 00:30 | |
| *** nosnos has quit IRC | 00:31 | |
| *** nosnos has joined #openstack-swift | 00:32 | |
| *** shri has quit IRC | 00:34 | |
| openstackgerrit | paul luse proposed a change to openstack/swift: Add Storage Policy Support to Recon Middleware https://review.openstack.org/87387 | 00:43 |
|---|---|---|
| *** bsdkurt has quit IRC | 00:44 | |
| *** bsdkurt has joined #openstack-swift | 00:45 | |
| *** csd has quit IRC | 00:54 | |
| *** shakamunyi has quit IRC | 00:54 | |
| *** lpabon has quit IRC | 00:55 | |
| *** NM1 has joined #openstack-swift | 00:57 | |
| *** zackf has joined #openstack-swift | 01:00 | |
| openstackgerrit | paul luse proposed a change to openstack/swift: Slight modification to the object updater sweep code https://review.openstack.org/87045 | 01:01 |
| *** fifieldt has joined #openstack-swift | 01:10 | |
| *** ekarlso has quit IRC | 01:11 | |
| openstackgerrit | paul luse proposed a change to openstack/swift: Super small code cleanup in Updater https://review.openstack.org/87424 | 01:12 |
| *** ekarlso has joined #openstack-swift | 01:13 | |
| *** zackf has quit IRC | 01:13 | |
| *** Mikalv has quit IRC | 01:13 | |
| *** ekarlso has quit IRC | 01:13 | |
| *** Mikalv has joined #openstack-swift | 01:13 | |
| *** zackf has joined #openstack-swift | 01:14 | |
| *** ekarlso has joined #openstack-swift | 01:18 | |
| *** ryao has quit IRC | 01:18 | |
| *** ryao has joined #openstack-swift | 01:19 | |
| *** ryao has quit IRC | 01:19 | |
| *** ryao has joined #openstack-swift | 01:24 | |
| *** ryao has quit IRC | 01:25 | |
| *** ryao has joined #openstack-swift | 01:29 | |
| *** ryao has quit IRC | 01:30 | |
| *** ryao has joined #openstack-swift | 01:34 | |
| *** gyee has quit IRC | 01:35 | |
| h6w | notmyname: Thanks. That makes some sense. :-) | 01:38 |
| *** mwstorer has quit IRC | 01:40 | |
| *** saschpe has quit IRC | 01:41 | |
| *** saschpe has joined #openstack-swift | 01:43 | |
| *** foexle has quit IRC | 01:44 | |
| *** shakamunyi has joined #openstack-swift | 01:51 | |
| *** MooingLemur has joined #openstack-swift | 01:52 | |
| *** shakamunyi has quit IRC | 01:55 | |
| *** lpabon has joined #openstack-swift | 01:59 | |
| *** lpabon has quit IRC | 02:02 | |
| *** haomaiw__ has quit IRC | 02:17 | |
| *** haomaiwang has joined #openstack-swift | 02:17 | |
| *** haomaiwang has quit IRC | 02:24 | |
| *** mlipchuk has joined #openstack-swift | 02:31 | |
| *** haomaiwang has joined #openstack-swift | 02:31 | |
| *** mlipchuk has quit IRC | 02:32 | |
| *** praveenkumar has joined #openstack-swift | 02:53 | |
| hugokuo | Good morning | 02:55 |
| *** haomaiwang has quit IRC | 02:55 | |
| *** gadb has joined #openstack-swift | 02:56 | |
| *** haomaiwa_ has joined #openstack-swift | 03:00 | |
| *** matsuhashi has quit IRC | 03:27 | |
| *** zhiyan_ is now known as zhiyan | 03:30 | |
| *** zhiyan is now known as zhiyan_ | 03:31 | |
| *** nosnos has quit IRC | 03:39 | |
| *** zackf has quit IRC | 03:41 | |
| *** zackf has joined #openstack-swift | 03:41 | |
| *** zhiyan_ is now known as zhiyan | 03:47 | |
| *** Anju has joined #openstack-swift | 03:53 | |
| *** nester has quit IRC | 04:01 | |
| briancline | has anyone using sublime found a good plugin arrangement to make it cooperate with flake8+hacking? | 04:01 |
| *** nester has joined #openstack-swift | 04:01 | |
| *** chandan_kumar has joined #openstack-swift | 04:07 | |
| *** Guest____ has joined #openstack-swift | 04:10 | |
| *** shakamunyi has joined #openstack-swift | 04:14 | |
| *** shakamunyi has quit IRC | 04:16 | |
| *** chandan_kumar has quit IRC | 04:17 | |
| *** ppai has joined #openstack-swift | 04:17 | |
| *** matsuhashi has joined #openstack-swift | 04:25 | |
| *** nosnos has joined #openstack-swift | 04:31 | |
| *** zackf has quit IRC | 04:32 | |
| *** zackf has joined #openstack-swift | 04:33 | |
| briancline | haven't found a "good" way yet without using build_systems settings in a project settings file | 04:43 |
| briancline | I'm lazy and want the hacking warnings to show up in the gutter just like normal pep8 violations :p | 04:44 |
| *** chandan_kumar has joined #openstack-swift | 04:50 | |
| *** matsuhashi has quit IRC | 05:03 | |
| *** matsuhashi has joined #openstack-swift | 05:04 | |
| *** tdasilva has quit IRC | 05:12 | |
| gadb | afternoon.. | 05:13 |
| gadb | my swift have 3 availablity zones consist in 9 object node | 05:14 |
| gadb | but, when i upload new object then data do not exist each unique zone. | 05:15 |
| gadb | of course. after as while swift try to handoff object. | 05:16 |
| gadb | but. i wonder why do not swift data placement "unique-as-possible" at one time. | 05:17 |
| gadb | i know. cause of handoff is networks or disk issue or something problum.. | 05:19 |
| gadb | but, my swift looks like all clear at me. | 05:20 |
| gadb | how can i trace reason of object handoff | 05:21 |
| *** haomaiwa_ has quit IRC | 05:25 | |
| *** haomaiwang has joined #openstack-swift | 05:25 | |
| *** pbojanic has joined #openstack-swift | 05:28 | |
| *** d89 has joined #openstack-swift | 05:28 | |
| *** NM1 has quit IRC | 05:39 | |
| *** NM1 has joined #openstack-swift | 05:39 | |
| *** NM1 has quit IRC | 05:44 | |
| *** nshaikh has joined #openstack-swift | 05:56 | |
| openstackgerrit | Michael Barton proposed a change to openstack/swift: Sendfile acceleration for GET requests https://review.openstack.org/82186 | 06:27 |
| *** Guest____ has quit IRC | 06:33 | |
| *** Guest____ has joined #openstack-swift | 06:34 | |
| *** portante has quit IRC | 06:35 | |
| *** nacim has joined #openstack-swift | 06:37 | |
| *** foexle has joined #openstack-swift | 06:38 | |
| *** portante has joined #openstack-swift | 06:39 | |
| *** ChanServ sets mode: +v portante | 06:39 | |
| *** pbojanic has quit IRC | 06:48 | |
| *** nacim has quit IRC | 06:50 | |
| *** portante has quit IRC | 06:57 | |
| *** matsuhashi has quit IRC | 07:01 | |
| *** matsuhashi has joined #openstack-swift | 07:01 | |
| *** portante has joined #openstack-swift | 07:02 | |
| *** ChanServ sets mode: +v portante | 07:02 | |
| *** matsuhashi has quit IRC | 07:02 | |
| *** psharma has joined #openstack-swift | 07:02 | |
| *** matsuhashi has joined #openstack-swift | 07:03 | |
| *** Guest____ has quit IRC | 07:04 | |
| *** cheri has joined #openstack-swift | 07:08 | |
| *** joeljwright has joined #openstack-swift | 07:26 | |
| *** cheri has quit IRC | 07:39 | |
| *** Anju has quit IRC | 07:40 | |
| *** cheri has joined #openstack-swift | 07:56 | |
| *** nacim has joined #openstack-swift | 08:05 | |
| *** tanee-away is now known as tanee | 08:05 | |
| *** tanee is now known as tanee-away | 08:05 | |
| *** tanee-away is now known as tanee | 08:06 | |
| *** tanee is now known as tanee-away | 08:06 | |
| *** tanee-away is now known as tanee | 08:06 | |
| *** zhiyan is now known as zhiyan_ | 08:10 | |
| *** zhiyan_ is now known as zhiyan | 08:10 | |
| *** matsuhashi has quit IRC | 08:13 | |
| *** matsuhashi has joined #openstack-swift | 08:14 | |
| *** zhiyan is now known as zhiyan_ | 08:14 | |
| openstackgerrit | Brian Cline proposed a change to openstack/swift: Uses None instead of mutables for function param defaults https://review.openstack.org/87490 | 08:14 |
| *** zhiyan_ is now known as zhiyan | 08:15 | |
| *** haomai___ has joined #openstack-swift | 08:17 | |
| *** fbo_away is now known as fbo | 08:18 | |
| *** matsuhashi has quit IRC | 08:18 | |
| *** haomaiwang has quit IRC | 08:20 | |
| openstackgerrit | Yuan Zhou proposed a change to openstack/swift: Fixes probe tests with non-default storage policy https://review.openstack.org/87492 | 08:23 |
| openstackgerrit | Yuan Zhou proposed a change to openstack/swift: Fixes probe tests with non-zero default storage policy https://review.openstack.org/87492 | 08:26 |
| *** matsuhashi has joined #openstack-swift | 08:28 | |
| *** Alex_Gaynor has quit IRC | 08:32 | |
| *** Alex_Gaynor has joined #openstack-swift | 08:38 | |
| *** gadb has quit IRC | 08:38 | |
| * yuan is away: I'm away | 08:44 | |
| *** yuan is now known as yuanzz | 08:45 | |
| *** JelleB is now known as a1|away | 09:08 | |
| *** haomai___ has quit IRC | 09:12 | |
| *** haomaiwa_ has joined #openstack-swift | 09:12 | |
| *** zhiyan is now known as zhiyan_ | 09:13 | |
| *** zhiyan_ is now known as zhiyan | 09:15 | |
| *** mkollaro has joined #openstack-swift | 09:17 | |
| *** haomaiw__ has joined #openstack-swift | 09:17 | |
| *** haomaiwa_ has quit IRC | 09:21 | |
| *** fifieldt has quit IRC | 09:33 | |
| *** mkollaro has quit IRC | 09:34 | |
| *** chandan_kumar has quit IRC | 09:38 | |
| *** chandan_kumar has joined #openstack-swift | 09:55 | |
| *** mkollaro has joined #openstack-swift | 10:02 | |
| *** mkollaro has quit IRC | 10:10 | |
| *** chandan_kumar has quit IRC | 10:23 | |
| *** nosnos has quit IRC | 10:27 | |
| *** chandan_kumar has joined #openstack-swift | 10:37 | |
| *** matsuhashi has quit IRC | 10:41 | |
| *** matsuhashi has joined #openstack-swift | 10:44 | |
| *** matsuhashi has quit IRC | 10:54 | |
| *** praveenkumar has quit IRC | 11:05 | |
| *** pbojanic has joined #openstack-swift | 11:16 | |
| *** mkollaro has joined #openstack-swift | 11:18 | |
| *** praveenkumar has joined #openstack-swift | 11:20 | |
| *** ppai has quit IRC | 11:23 | |
| *** ppai has joined #openstack-swift | 11:35 | |
| *** pbojanic has quit IRC | 11:38 | |
| *** chandan_kumar has quit IRC | 11:42 | |
| *** goneri has joined #openstack-swift | 11:45 | |
| *** chandan_kumar has joined #openstack-swift | 11:58 | |
| *** ppai has quit IRC | 12:01 | |
| *** mkollaro1 has joined #openstack-swift | 12:02 | |
| *** mkollaro has quit IRC | 12:02 | |
| *** mmcardle has quit IRC | 12:03 | |
| *** _bluev has joined #openstack-swift | 12:03 | |
| _bluev | Hello | 12:03 |
| _bluev | Has anyone seen container-server or object-server go to 100% CPU in a hard loop trying to send log message when the syslog daemon restarts ? | 12:09 |
| *** ppai has joined #openstack-swift | 12:14 | |
| openstackgerrit | Julien Danjou proposed a change to openstack/python-swiftclient: Allow to retrieve response from put_container https://review.openstack.org/87575 | 12:23 |
| _bluev | I think I've hit https://bugs.launchpad.net/swift/+bug/1156542 on about 8 backend servers ( ~20% ) | 12:25 |
| *** chandan_kumar has quit IRC | 12:26 | |
| *** ppai has quit IRC | 12:30 | |
| omame | _bluev: yeah, it's a known thing. just reload swift after syslog | 12:32 |
| openstackgerrit | paul luse proposed a change to openstack/swift: Bring Object Auditor Coverage To 99% https://review.openstack.org/87101 | 12:34 |
| *** chandan_kumar has joined #openstack-swift | 12:40 | |
| _bluev | omame: it looks like http://bugs.python.org/issue15179 *may* be the root of the issue | 12:42 |
| *** mmcardle has joined #openstack-swift | 12:45 | |
| *** cheri has quit IRC | 12:50 | |
| *** bada has joined #openstack-swift | 12:52 | |
| *** Alex_Gaynor has quit IRC | 12:53 | |
| *** mrsnivvel has joined #openstack-swift | 12:55 | |
| *** mrsnivvel has quit IRC | 12:57 | |
| *** mrsnivvel has joined #openstack-swift | 12:57 | |
| *** pbojanic has joined #openstack-swift | 12:59 | |
| *** tanee is now known as tanee-away | 13:03 | |
| *** mkollaro has joined #openstack-swift | 13:05 | |
| *** Alex_Gaynor has joined #openstack-swift | 13:06 | |
| *** nshaikh has quit IRC | 13:07 | |
| *** mkollaro1 has quit IRC | 13:09 | |
| *** pbojanic has quit IRC | 13:09 | |
| *** mrsnivvel has quit IRC | 13:23 | |
| *** j_king has joined #openstack-swift | 13:25 | |
| *** psharma has quit IRC | 13:35 | |
| swifterdarrell | _bluev: we independently traced it down to this: https://gist.github.com/temoto/8259860#file-eventlet-unix-socket-enotconn-v2-py | 13:57 |
| swifterdarrell | _bluev: one workaround is to use UDP for syslog instead of a unix domain socket | 13:57 |
| _bluev | swifterdarrell: do you think the fix put in place for http://bugs.python.org/issue15179 would resolve the issue ? | 13:59 |
| swifterdarrell | _bluev: unsure; can you tell what version of 2.7.x that fix has gone into? | 13:59 |
| swifterdarrell | _bluev: (IOW, the only way to be sure would be to test it) | 14:00 |
| _bluev | swifterdarrell: great question :-) . my google foo for changeset to python version mapping did not get me far | 14:01 |
| *** tanee-away is now known as tanee | 14:01 | |
| swifterdarrell | _bluev: my guess is you'd have to play Mercurial to get the answer | 14:01 |
| swifterdarrell | _bluev: 2.7.6 tag was Sun, 10 Nov 2013 | 14:02 |
| swifterdarrell | _bluev: 2.7.5 Sat, 11 May 2013 | 14:03 |
| swifterdarrell | _bluev: 2.7.4 Sat, 06 Apr 2013 | 14:03 |
| swifterdarrell | _bluev: 2.7.3 Mon, 09 Apr 2012 | 14:03 |
| swifterdarrell | _bluev: and the fix: Mon, 25 Jun 2012 | 14:04 |
| swifterdarrell | _bluev: So I'd expect that fix to be in Python 2.7.4 and later | 14:04 |
| *** piyush has joined #openstack-swift | 14:05 | |
| _bluev | swifterdarrell: thanks. Ubuntu 12.04 LTS has 2.7.3-0ubuntu3.4 presently and does not have the handler.py fix | 14:07 |
| swifterdarrell | _bluev: *nod* I'm really curious if that fixes the syslog issue; if you test, can you share your results here; not sure how long it'll be until I get a chance to test it | 14:08 |
| *** vr1 has joined #openstack-swift | 14:13 | |
| portante | vr1: hello | 14:13 |
| vr1 | hello | 14:13 |
| portante | so the way you use another backend is to change the config file | 14:13 |
| vr1 | ok | 14:13 |
| *** chandan_kumar has quit IRC | 14:13 | |
| portante | swift uses "paste" to deploy | 14:13 |
| portante | so you find a line in the object server config file like: "use = egg:swift#object" | 14:14 |
| portante | that is paste syntax for: "look up the python egg called swift, and reference the entry point named 'object'" | 14:15 |
| portante | essentially | 14:15 |
| * portante others more knowledgable about paste might add some clarification to that description | 14:15 | |
| portante | so the key is that #object | 14:16 |
| portante | take a look at your setup.cfg file in the root of the swift tree | 14:16 |
| portante | you see a section entitled, "[entry_points]" | 14:16 |
| portante | and in there you should then see under "paste.app_factory =", the 'object' and 'mem_object' definitions | 14:17 |
| portante | so then you know that you can change your object server config file to: "use = egg:swift#mem_object" | 14:18 |
| portante | and that will get you the in-memory object server backend | 14:18 |
| portante | vr1: does that make sense? | 14:18 |
| vr1 | yes very cool I test that | 14:20 |
| *** RockKuo has quit IRC | 14:23 | |
| *** fifieldt has joined #openstack-swift | 14:24 | |
| swifterdarrell | _bluev: I just tested a freshly-built python 2.7.6 on CentOS; I first confirmed the 100% CPU utilization in a proxy-server with normal syslog config when running "service rsyslog restart" | 14:38 |
| swifterdarrell | _bluev: then installed the 2.7.6 RPM I built and could no longer reproduce the problem | 14:38 |
| portante | swifterdarrell: so the CentOS version you are using starts out with what python? | 14:40 |
| portante | Is it 2.6? | 14:40 |
| portante | sorry for jumping in late if you have already explained this | 14:40 |
| swifterdarrell | portante: something like that; we've always built 2.7.3 for it | 14:40 |
| swifterdarrell | portante: so the "before" was 2.7.3 | 14:40 |
| swifterdarrell | portante: all the same build infrastructure built the 2.7.6 | 14:40 |
| swifterdarrell | portante: (it's a differently-named package which is "rooted" in a non-system location and it lives alongside the system python 2.6.x) | 14:41 |
| portante | ah, you are not replacing the base python image, but using a /opt installed one | 14:41 |
| swifterdarrell | portante: correct | 14:41 |
| portante | yes, great | 14:41 |
| portante | thanks | 14:41 |
| swifterdarrell | portante: np | 14:41 |
| *** d89 has quit IRC | 14:46 | |
| *** mwstorer has joined #openstack-swift | 14:50 | |
| vr1 | portante: it worked thanks alot | 14:51 |
| _bluev | swifterdarrell: good to know - thanks | 14:53 |
| swifterdarrell | _bluev: np, thanks for the heads-up! | 14:53 |
| _bluev | swifterdarrell: question is, should https://bugs.launchpad.net/swift/+bug/1156542 now be closed ? | 14:54 |
| portante | vr1: great! you are welcome | 14:55 |
| swifterdarrell | _bluev: dunno... I mean, swift + python 2.7.3 or lower has bad behavior when rsyslog is restarted with unix-domain-logging enabled | 14:56 |
| vr1 | actually I'm trying to make an ObjectController for a specific KVS | 14:56 |
| swifterdarrell | _bluev: it's not a "Swift bug" per se, but it's worth knowing it happens (maybe a launchpad bug isn't the best documentation for that...) | 14:56 |
| portante | vr1: let me know if I can help | 14:56 |
| swifterdarrell | _bluev: at a minimum, the bug should be updated w/the information that the proposed patch A) works; and B) is included in Python 2.7.4+ | 14:57 |
| _bluev | swifterdarrell: makes sense, I will do that later | 14:57 |
| *** tdasilva has joined #openstack-swift | 14:57 | |
| vr1 | portante: great thanks are you any of the maintainer of swift-obj ? | 14:57 |
| portante | you might want to take a look a the gluster-swift code, and the RADOS proposed backend at, https://review.openstack.org/#/c/60215/ | 14:57 |
| portante | vr1: I have worked on it from time to time | 14:58 |
| *** RockKuo has joined #openstack-swift | 14:58 | |
| portante | and I think swiftstack might have a backend somewhere for the seagate kinetics drive? | 14:58 |
| portante | not sure | 14:58 |
| openstackgerrit | paul luse proposed a change to openstack/swift: Add Storage Policy Support to Recon Middleware https://review.openstack.org/87387 | 14:58 |
| portante | notmyname, clayg: ^^^ | 14:58 |
| hugokuo | RockKuo: hi | 14:59 |
| vr1 | portante: great resource thanks | 14:59 |
| vr1 | for those object backends which already do support replication and manifests | 14:59 |
| vr1 | what are we supposed to do ? | 14:59 |
| portante | vr1: it depends on what your backend is offering, I believe | 15:00 |
| vr1 | because there is a kind of duplication with what swift initially offers | 15:00 |
| portante | vr1: you might want to read up on storage-policies before you continue | 15:00 |
| portante | peluse has some docs for it, I believe now | 15:01 |
| portante | vr1: what is the KVS file system? | 15:04 |
| vr1 | no it is a Key Value Store | 15:04 |
| vr1 | not a file system | 15:04 |
| *** RockKuo_iPad has joined #openstack-swift | 15:04 | |
| portante | what is the use case for a backend for that? Is it to get the swift API for it? | 15:05 |
| portante | Or in other words, how do you envision a swift client accessing your key/value store via the Swift API? | 15:05 |
| hugokuo | RockKuo_iPad: hi | 15:05 |
| vr1 | portante: accessing our kvs via swift | 15:07 |
| *** lpabon has joined #openstack-swift | 15:10 | |
| *** tanee is now known as tanee-away | 15:10 | |
| portante | so you are looking to add support for the swift API by placing swift configuration infront of your kvs | 15:10 |
| portante | vr1: are you sure that overhead is really what you want? | 15:11 |
| vr1 | what is the alternative ? | 15:11 |
| portante | it depends on the architecture of your kvs | 15:11 |
| portante | what is the current mode of access to your kvs? memcached type API? | 15:12 |
| vr1 | PUT/GET/DELETE | 15:12 |
| portante | how is your data stored? | 15:12 |
| vr1 | as objects | 15:12 |
| vr1 | not as files | 15:13 |
| notmyname | good morning world | 15:13 |
| portante | okay, but how are they represented on the storage as objects? | 15:13 |
| vr1 | they are kept as is | 15:13 |
| *** lpabon has quit IRC | 15:13 | |
| vr1 | unless you use EC in this case they are split | 15:13 |
| notmyname | creiht: redbo: dfg: anything come up in your testing so far since the py26 logging issue? ttx is about to ask me if rc2 is ok | 15:14 |
| vr1 | you mean the low level storage ? | 15:14 |
| notmyname | portante: swifterdarrell: anything you know of that has come up in the last couple of days? | 15:14 |
| portante | vr1: yes | 15:14 |
| portante | notmyname: nothing from my end | 15:14 |
| notmyname | portante: thanks | 15:14 |
| vr1 | they are written directly into a partition, not in a file | 15:15 |
| vr1 | the partition can be a file | 15:15 |
| vr1 | but not necessarily | 15:15 |
| vr1 | why are you asking ? | 15:15 |
| portante | vr1, just trying to see if how that partition method is done can become the backend to swift | 15:16 |
| vr1 | ah ok | 15:16 |
| portante | you have some on-disk format for your objects | 15:16 |
| vr1 | yes we do | 15:16 |
| portante | that might be a useful backend for swift | 15:16 |
| portante | but it seems like too much overhead to place a swift cluster in front of your existing KVS just to get the API, adding lots of latency | 15:17 |
| _bluev | does anyone here use swift-ring-master in production ? (besides RAX….) | 15:17 |
| *** fifieldt has quit IRC | 15:18 | |
| redbo | notmyname: nothing else has come up, but we haven't finished testing | 15:18 |
| vr1 | portante: if swift is fast it will be OK, do we have perf numbers for swift ? | 15:19 |
| notmyname | redbo: ok, thanks. when do you expect testing to be done? | 15:20 |
| portante | so a swift REST API request first goes to a proxy server process, which then forwards the request (with modifications) to the object server process, which would then forward the request to your KVS | 15:21 |
| *** vr2 has joined #openstack-swift | 15:21 | |
| redbo | yesterday. but our testing environment isn't being cooperative. let me go ask. | 15:21 |
| vr1 | is it recommended to run apache instead of the default one ? | 15:21 |
| vr1 | (http) | 15:21 |
| *** vr1 has quit IRC | 15:21 | |
| portante | huh? | 15:21 |
| portante | apache httpd vs. ... ? | 15:22 |
| *** foexle has quit IRC | 15:22 | |
| vr2 | there is a built-in wsgi server | 15:22 |
| vr2 | no ? | 15:22 |
| portante | yes, eventlet wsgi | 15:22 |
| portante | I see | 15:22 |
| portante | vr2, are you the one-in-the-same with vr1? | 15:23 |
| vr2 | I mean the swift frontend with apache has reasonable perf ? | 15:23 |
| notmyname | redbo: heh. thanks :-) | 15:23 |
| vr2 | portante: yes I am (wifi :) ) | 15:23 |
| portante | vr2, it does, depending on the load, small vs large objects | 15:24 |
| portante | lots of factors involved | 15:24 |
| notmyname | portante: do you run/test with httpd vs eventlet? | 15:25 |
| portante | we have only used eventlet here for our testing | 15:25 |
| vr2 | the same for me | 15:25 |
| vr2 | eventlet | 15:25 |
| vr2 | this is the default in devstack | 15:25 |
| notmyname | all the large swift clusters I know of (and the small ones too for that matter) all use eventlet in prod | 15:26 |
| *** zackf has quit IRC | 15:26 | |
| *** RockKuo_iPad has quit IRC | 15:26 | |
| portante | vr2: not matter how well swift performs, a swift cluster in front of your kvs will always add latency/overhead to a request | 15:26 |
| portante | can you share more about how your kvs is architected? is it open source? | 15:27 |
| vr2 | actually in the past we made the mistake to mimic S3 api, we dont want to do the same mistake with swift | 15:27 |
| vr2 | if you mimic an API, you are never totally compatible | 15:27 |
| portante | vr2, that is why gluster-swift was born | 15:28 |
| portante | ceph and now oracle offer their own swift implementation, saying it is compatible | 15:29 |
| vr2 | yes actually I belong to scality | 15:29 |
| portante | but they cannot leverage the growing ecosystem around object storage | 15:29 |
| *** _bluev has quit IRC | 15:29 | |
| *** tanee has joined #openstack-swift | 15:31 | |
| *** vr2 has quit IRC | 15:32 | |
| *** vr1 has joined #openstack-swift | 15:32 | |
| portante | is scality open source? | 15:32 |
| portante | vr2? | 15:32 |
| vr1 | portante: I'm back | 15:32 |
| vr1 | scality is partly close | 15:33 |
| portante | vr1: ah, the ol' switch-er-roo | 15:33 |
| portante | vr1: oh | 15:33 |
| vr1 | close source essentially | 15:33 |
| portante | okay | 15:34 |
| *** Trixboxer has joined #openstack-swift | 15:35 | |
| notmyname | ---->> last call for an RC 3 is tomorrow <<---- | 15:35 |
| *** tanee-away has quit IRC | 15:36 | |
| * portante heads to lunch | 15:36 | |
| redbo | We had one machine with a bunch of disks and threads_per_disk=4 that'd just spam "can't create thread" to the error log. I'm guessing it hit some ulimit on stack size or something. I'm going to try looking into it. | 15:36 |
| redbo | notmyname: I'm told all of our regression tests are passing | 15:37 |
| notmyname | redbo: how many workers? | 15:37 |
| notmyname | redbo: cool, thanks | 15:37 |
| redbo | 24 I think | 15:38 |
| openstackgerrit | Chmouel Boudjnah proposed a change to openstack/python-swiftclient: Fix test_multithreading on Python 3 https://review.openstack.org/84104 | 15:38 |
| vr1 | portante: I noticed gluster-swift is actually a fork, is it a good idea ? | 15:39 |
| vr1 | why not just have written a gluster_server.py ? | 15:39 |
| notmyname | redbo: I know torgomatic has seen something like that before. probably the ulimit from thread_per_disk | 15:39 |
| notmyname | vr1: it's not a fork. it's an artifact of the fact that we can't yet reference a DiskFile implementation in a config file (which really needs to happen!) | 15:41 |
| notmyname | portante: please confirm ^ :-) | 15:41 |
| openstackgerrit | Christian Schwede proposed a change to openstack/python-swiftclient: Add functional tests for python-swiftclient https://review.openstack.org/76355 | 15:42 |
| vr1 | notmyname: ceph had succesfully implemented a diskfile there: https://review.openstack.org/#/c/60215/ | 15:42 |
| vr1 | without forking | 15:42 |
| vr1 | it is why I was wondering | 15:43 |
| *** zackf has joined #openstack-swift | 15:44 | |
| *** _bluev has joined #openstack-swift | 15:44 | |
| *** _bluev has quit IRC | 15:47 | |
| *** _bluev has joined #openstack-swift | 15:48 | |
| *** _bluev has quit IRC | 15:48 | |
| notmyname | I've got to step out for a few hours | 15:51 |
| notmyname | vr1: I think portante or tdasilva will be able to say more about the gluster-swift stuff vs the eNovance ceph DiskFile | 15:52 |
| *** gyee has joined #openstack-swift | 15:52 | |
| notmyname | redbo: if something comes up that warrants an rc3, please send me a text 254-722-3321 | 15:52 |
| redbo | k | 15:54 |
| notmyname | thanks | 15:54 |
| *** Gu_______ has joined #openstack-swift | 16:04 | |
| chmouel | anyone have issues if i kill https://github.com/openstack/python-swiftclient/blob/master/swiftclient/client.py#L109 it's not being used by anyone since we moved to requests | 16:10 |
| *** zhiyan is now known as zhiyan_ | 16:12 | |
| *** nacim has quit IRC | 16:13 | |
| *** foexle has joined #openstack-swift | 16:14 | |
| *** _bluev has joined #openstack-swift | 16:15 | |
| *** Diddi has joined #openstack-swift | 16:23 | |
| *** foexle has quit IRC | 16:23 | |
| portante | vr1: how it is a fork of swift? | 16:32 |
| portante | it does not have the entire swift code base in it | 16:33 |
| portante | it just provides DiskFile and DiskDir, leveraging all the other code in swift | 16:33 |
| vr1 | portante: ok | 16:33 |
| vr1 | portante: thanks for everything, if you are at the next summit, maybe we'll meet | 16:34 |
| vr1 | in atlanta | 16:34 |
| portante | sure, I'm am planning on being there | 16:34 |
| vr1 | my name is rancurel | 16:34 |
| vr1 | see you there | 16:34 |
| vr1 | I need to go | 16:34 |
| creiht | notmyname, portante: sorry... recovering from watching the elcipse last night | 16:34 |
| portante | vr1, thanks, see you tehre | 16:34 |
| vr1 | ++ | 16:34 |
| *** vr1 has left #openstack-swift | 16:34 | |
| portante | creiht: how was it? | 16:35 |
| creiht | very cool | 16:35 |
| portante | did you snap any pictures? | 16:35 |
| creiht | of course :) | 16:35 |
| creiht | https://www.flickr.com/photos/thethiers/sets/72157644001323554/ | 16:35 |
| portante | that is my next adventure getting a telescope to which I can hook up a camera | 16:35 |
| creiht | hehe | 16:35 |
| creiht | I still need to do that :) | 16:35 |
| creiht | That's just from my camera and a good lens :) | 16:36 |
| portante | wow, that color is just fantastic | 16:36 |
| creiht | thanks | 16:36 |
| creiht | it was pretty cool to watch | 16:36 |
| portante | I bet | 16:36 |
| * portante is jealous | 16:36 | |
| creiht | portante: I have a busy day today, but I still have it on my list to check out the in process func tests | 16:37 |
| portante | great, thanks, creiht | 16:37 |
| *** csd has joined #openstack-swift | 16:44 | |
| *** judd7 has joined #openstack-swift | 16:45 | |
| portante | notmyname: the above vr* discussions are why I think a session on the DiskFile backends would be a good thing to have at summit (in some form) | 16:46 |
| tdasilva | portante: +1 | 16:48 |
| *** mmcardle has quit IRC | 16:49 | |
| *** tdasilva has left #openstack-swift | 16:59 | |
| *** EmilienM has quit IRC | 17:03 | |
| *** simpleAJ has joined #openstack-swift | 17:06 | |
| simpleAJ | does anybody know if I need to set http://docs.openstack.org/developer/swift/development_saio.html#optional-setting-up-rsyslog-for-individual-logging to get logs for my dev box | 17:07 |
| simpleAJ | i am using https://github.com/swiftstack/vagrant-swift-all-in-one | 17:07 |
| *** EmilienM has joined #openstack-swift | 17:07 | |
| *** shri has joined #openstack-swift | 17:08 | |
| *** lpabon has joined #openstack-swift | 17:08 | |
| redbo | simpleAJ: if you don't do that step, everything will go into /var/log/syslog | 17:10 |
| redbo | it can be helpful to have it broken out into individual log files | 17:11 |
| simpleAJ | agree. | 17:14 |
| simpleAJ | thanks | 17:14 |
| MooingLemur | can container sync be parallelized more than 1 process per container server? | 17:29 |
| *** csd has quit IRC | 17:34 | |
| *** rupsky has joined #openstack-swift | 17:46 | |
| *** piyush has quit IRC | 17:48 | |
| *** mmcardle has joined #openstack-swift | 17:50 | |
| *** piyush has joined #openstack-swift | 17:51 | |
| *** mlipchuk has joined #openstack-swift | 17:51 | |
| *** piyush1 has joined #openstack-swift | 17:53 | |
| *** mmcardle has quit IRC | 17:54 | |
| *** piyush has quit IRC | 17:56 | |
| *** Gu_______ has quit IRC | 18:01 | |
| *** mlipchuk has quit IRC | 18:03 | |
| *** lnxnut has joined #openstack-swift | 18:12 | |
| *** csd has joined #openstack-swift | 18:15 | |
| *** ozone227 has joined #openstack-swift | 18:16 | |
| *** simpleAJ has left #openstack-swift | 18:16 | |
| *** mlipchuk has joined #openstack-swift | 18:19 | |
| *** _80386_ has joined #openstack-swift | 18:21 | |
| _80386_ | I have a ring question for anyone interested. I have 4 machines, 2 in one room and 2 in another. My customer requires 4 replicas. Would it be better to setup a 2 zone (1 per room) or 4 zone (1 per machine) ring? | 18:23 |
| *** tburnes has joined #openstack-swift | 18:30 | |
| portante | _80386_: what do you consider the failure domains to be for your hardware? | 18:33 |
| *** tburnes has quit IRC | 18:34 | |
| portante | do have sufficient redundancy in your racks? | 18:34 |
| portante | would one whole room be taken out by a power failure? | 18:34 |
| _80386_ | portante: the issue under discussion is the loss of a computer room. the loss of the room could be due to power or network outage. | 18:34 |
| _80386_ | portante: the racks are dual powered as are the servers | 18:35 |
| _80386_ | I was thinking 4z 1 per machine could lead to a quorum issue if one of the rooms goes offline | 18:36 |
| portante | you have that with 2z 1 per room, too | 18:36 |
| _80386_ | so I made a ring with 2 machines per zone, 2 zones and looked at the handoffs with 4 replicas. if 1 zone goes down there are still 4 possible write devices | 18:37 |
| _80386_ | the ring was layed out 2 primary storage locations per zone and 2 handoffs per zone | 18:37 |
| portante | so if you loose the network to a room, the other room will still store the data to two different machines | 18:40 |
| portante | right? | 18:40 |
| portante | it might be 4 possible devices total, but it guarantees two different machines | 18:41 |
| _80386_ | yes | 18:41 |
| portante | is that sufficient to carry over until the other room would be restored? | 18:41 |
| _80386_ | i believe so | 18:41 |
| portante | what is your workload, write intensive or read intensive? | 18:41 |
| _80386_ | write intensive at the moment | 18:42 |
| _80386_ | which is why quorum came up in the discussion | 18:42 |
| portante | is there anything that prevents the two rooms thinking the other room is not up, but both rooms are still responding to requests? Split-brain basically | 18:43 |
| _80386_ | not that I am aware of | 18:44 |
| _80386_ | http://paste.openstack.org/show/75808/ | 18:44 |
| portante | so about their requirement for 4 way replication ... | 18:46 |
| portante | what is their expectation with 4 way? Is it that they'll have 7 9s instead of 5 9s of data availability? | 18:47 |
| _80386_ | in a DR event, if they physically lose one of the rooms they want 2 remaining copies | 18:47 |
| *** lnxnut has quit IRC | 18:47 | |
| portante | okay | 18:47 |
| portante | seems like that config provides that at leaste | 18:48 |
| portante | least | 18:48 |
| _80386_ | appreciate it portante | 18:49 |
| *** mmcardle has joined #openstack-swift | 18:50 | |
| *** mmcardle has quit IRC | 18:54 | |
| *** zigo has quit IRC | 19:04 | |
| *** ozone227 has quit IRC | 19:05 | |
| *** ozone227 has joined #openstack-swift | 19:06 | |
| *** zigo has joined #openstack-swift | 19:07 | |
| *** wasmum has joined #openstack-swift | 19:21 | |
| *** _bluev has quit IRC | 19:22 | |
| *** bmac423 has joined #openstack-swift | 19:24 | |
| bmac423 | Hi all, I was on last friday with some swift-informant issues. I'm getting this error... http://paste.openstack.org/show/75812/ | 19:26 |
| bmac423 | I'm running on centos with the latest swift-informant release, python 2.6 | 19:29 |
| *** judd7_ has joined #openstack-swift | 19:33 | |
| *** wasmum_ has joined #openstack-swift | 19:34 | |
| *** shakamunyi has joined #openstack-swift | 19:34 | |
| *** judd7 has quit IRC | 19:34 | |
| openstackgerrit | Chmouel Boudjnah proposed a change to openstack/python-swiftclient: Remove validate_headers functions https://review.openstack.org/87772 | 19:36 |
| *** Trixboxer has quit IRC | 19:37 | |
| *** shakamunyi has quit IRC | 19:38 | |
| *** chandan_kumar has joined #openstack-swift | 19:39 | |
| *** wasmum- has joined #openstack-swift | 19:42 | |
| *** mmcardle has joined #openstack-swift | 19:42 | |
| *** wasmum- is now known as fin1te | 19:43 | |
| _80386_ | was wondering which one of you guys was wasmum | 19:43 |
| *** gyee has quit IRC | 19:43 | |
| wasmum | im wasmum | 19:45 |
| wasmum_ | im wasmum | 19:45 |
| *** bach_ has joined #openstack-swift | 19:45 | |
| *** bmac423 has quit IRC | 19:46 | |
| *** wasmum- has joined #openstack-swift | 19:47 | |
| *** wasmum- is now known as dc588 | 19:48 | |
| *** dc588 has quit IRC | 19:48 | |
| *** mmcardle has quit IRC | 19:48 | |
| *** chandan_kumar has quit IRC | 19:48 | |
| openstackgerrit | Chmouel Boudjnah proposed a change to openstack/python-swiftclient: Remove validate_headers https://review.openstack.org/87772 | 19:49 |
| *** csd has quit IRC | 19:51 | |
| *** wasmum has left #openstack-swift | 19:51 | |
| *** wasmum_ has quit IRC | 19:52 | |
| *** chandan_kumar has joined #openstack-swift | 19:55 | |
| *** judd7 has joined #openstack-swift | 20:05 | |
| *** judd7_ has quit IRC | 20:06 | |
| peluse | clayg: you there? | 20:15 |
| clayg | peluse: sure! | 20:16 |
| *** chandan_kumar has quit IRC | 20:18 | |
| *** bach_ has quit IRC | 20:19 | |
| peluse | clayg: I'm thinking I'll go ahead and push a patch to udpate SAIO to have a 2nd pol so that the functinoal tests that yuan is doing get executed w/every run, what do you think? | 20:19 |
| clayg | peluse: still dunno, I think yuan's changes need to raise SkipTest if there's not more than one policy anyway... | 20:19 |
| clayg | peluse: I think that the gate functests won't get updated to have a second policy unless someone changes them in devstack | 20:20 |
| peluse | clayg: sorry, I meant when a dev runs them on their own which is better than nothing. his patch does indeed SkipTest if there's only one policy | 20:20 |
| clayg | peluse: I think that core developers will slowly add storage policies to their dev setups after they're merged and seeing the skips in functests will only highly they're missing something | 20:21 |
| peluse | clayg: so you think its a good idea then? | 20:21 |
| *** joeljwright has quit IRC | 20:22 | |
| clayg | peluse: also I think there's going to be desire for some sort of "how to convert a single policy saio to multi policy" doc for when folks are doing storage policy reviews, so if we HAVE to have that, it may make since to just have the saio doc link to that instead of having the instructions inlined everywhere? | 20:22 |
| clayg | peluse: I don't feel stronly either way - i guess I have a desire to see a "configure a running swift to use storage policies" doc - but that *can* be seperate from the saio changes if others think the storage policy setup makes sense in there... | 20:24 |
| peluse | claygy: OK, I have the directions wirtten up in the overview_policies_rst file. I can yank them into their own doc easily enough and link to them from the overview file as well as the SAIO file. I just thought it'd be good to change remkaerings and swift.conf in the default SAIO setup as well but I don't feel strongly either way as well | 20:27 |
| clayg | peluse: oh yeah I see your point, some of the saio stuff is scripted... yeah probably wouldn't hurt to have that stuff updated - good call! | 20:28 |
| peluse | clayg: rock n roll. will post it up there so its in sync with the doc patch (once I update it too). gracias | 20:28 |
| clayg | de nada | 20:29 |
| *** bach has joined #openstack-swift | 20:31 | |
| *** lnxnut has joined #openstack-swift | 20:34 | |
| *** lnxnut has quit IRC | 20:35 | |
| *** lnxnut has joined #openstack-swift | 20:35 | |
| openstackgerrit | Chmouel Boudjnah proposed a change to openstack/python-swiftclient: Fix test_raw_upload test https://review.openstack.org/87784 | 20:44 |
| *** mmcardle has joined #openstack-swift | 20:44 | |
| *** mmcardle has quit IRC | 20:49 | |
| *** csd has joined #openstack-swift | 20:55 | |
| *** csd_ has joined #openstack-swift | 20:56 | |
| *** gyee has joined #openstack-swift | 20:57 | |
| *** csd has quit IRC | 21:00 | |
| *** piyush1 has quit IRC | 21:04 | |
| *** zigo has quit IRC | 21:06 | |
| *** zigo has joined #openstack-swift | 21:08 | |
| *** bach_ has joined #openstack-swift | 21:10 | |
| *** bach has quit IRC | 21:11 | |
| *** lpabon has quit IRC | 21:16 | |
| chmouel | woo one last patch and swiftclient tests are passing under py34 | 21:23 |
| *** tdasilva has joined #openstack-swift | 21:28 | |
| *** haypo has left #openstack-swift | 21:28 | |
| *** shri has quit IRC | 21:34 | |
| *** tdasilva has quit IRC | 21:38 | |
| *** _80386_ has quit IRC | 21:42 | |
| *** mmcardle has joined #openstack-swift | 21:45 | |
| *** shri has joined #openstack-swift | 21:49 | |
| *** ozone227 has quit IRC | 21:49 | |
| *** mmcardle has quit IRC | 21:50 | |
| briancline | let me just blue-sky this real quick | 21:57 |
| briancline | how important is xattr metadata for tombstoned objects? | 21:58 |
| clayg | briancline: storage policies is gunna use it - why? | 21:58 |
| creiht | briancline: it can be helpful if you are troubleshooting | 21:58 |
| portante | briancline: I think it is used to prevent certain DELETE attacks as well | 22:00 |
| portante | briancline: what are you thinking of doing? | 22:00 |
| portante | chmouel: nice! | 22:00 |
| briancline | on one of my storage nodes during a rolling upgrade, all tombstones got quarantined since it was unable to do xattr reads on them | 22:01 |
| *** rupsky has quit IRC | 22:01 | |
| portante | huh? | 22:01 |
| portante | did you change the fs on which the tombstones were? | 22:02 |
| portante | can you say more? | 22:02 |
| portante | briancline: | 22:03 |
| briancline | I can, but in private if that's alright | 22:03 |
| portante | briancline: sure | 22:04 |
| *** shri1 has joined #openstack-swift | 22:05 | |
| *** shri has quit IRC | 22:06 | |
| *** saschpe has quit IRC | 22:07 | |
| *** kevinc_ has joined #openstack-swift | 22:09 | |
| *** nester has quit IRC | 22:16 | |
| *** saschpe has joined #openstack-swift | 22:16 | |
| *** csd_ has quit IRC | 22:21 | |
| openstackgerrit | Derek Higgins proposed a change to openstack/python-swiftclient: Add python-keystoneclient to requirements.txt https://review.openstack.org/79498 | 22:22 |
| *** mkollaro has quit IRC | 22:25 | |
| notmyname | good morning world | 22:39 |
| notmyname | I talked to a lot of people today at the red hat summit who are interested in working with or using swift | 22:42 |
| *** mmcardle has joined #openstack-swift | 22:47 | |
| notmyname | redbo: were the errors you saw earlier confirmed to be ulimit issues? | 22:50 |
| *** mmcardle has quit IRC | 22:51 | |
| *** tdasilva has joined #openstack-swift | 22:52 | |
| *** mmcardle has joined #openstack-swift | 23:06 | |
| *** praveenkumar has quit IRC | 23:08 | |
| *** mmcardle1 has joined #openstack-swift | 23:08 | |
| *** mmcardle has quit IRC | 23:08 | |
| *** mmcardle1 has quit IRC | 23:13 | |
| h6w | Morning all. | 23:29 |
| h6w | Great news notmyname! :-D | 23:30 |
| *** openstackstatus has quit IRC | 23:32 | |
| *** openstackstatus has joined #openstack-swift | 23:33 | |
| *** nosnos has joined #openstack-swift | 23:39 | |
| *** bach_ has quit IRC | 23:48 | |
| *** judd7 has quit IRC | 23:53 | |
| *** mwstorer has quit IRC | 23:56 | |
| *** mordred has quit IRC | 23:58 | |
| *** mordred has joined #openstack-swift | 23:58 | |
| *** ChanServ changes topic to "Restarting gerrit really quick to fix replication issue" | 23:59 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!