| kota_ | if we have to call socket.close, we need to do that in MemcachePool.creat but I'm not sure what's happen if we got Timeout during socket.connect() | 00:01 |
|---|---|---|
| kota_ | do we need to call socket.close() to ensure closing? or just leave it (and expect to decrease reference counter and gc'ed ) as well as current swift? | 00:03 |
| notmyname | kota_: in my experience with getting swift running under pypy, it's almost always good to have swift explicitly close sockets and not rely on GC to take care of it | 00:04 |
| *** diogogmt has quit IRC | 00:04 | |
| kota_ | notmyname: good call | 00:04 |
| clayg | lazy garbage collectors | 00:05 |
| kota_ | notmyname, clayg: drafting as https://gist.github.com/bloodeagle40234/0fb2b0a65dba382f6ef7a642e41fe400 | 00:06 |
| notmyname | if you want some juicy tech details on the bug in eventlet related to closing sockets that is currently blocking swift-on-pypy, check otu https://github.com/eventlet/eventlet/issues/318#issuecomment-224756745 (resolved by https://github.com/eventlet/eventlet/pull/326) | 00:07 |
| notmyname | unfortunately, the eventlet maintainers are silent on the pull request, so the issue is still open | 00:07 |
| kota_ | it still need futher work though (remove comment out, and probably we can remove the meaningless sock arg from method call) | 00:07 |
| kota_ | notmyname: thanks! | 00:08 |
| kota_ | will read it. | 00:08 |
| clayg | kota_: seems reasonable | 00:09 |
| kota_ | clayg: thanks :-) will push a clean version soon (at least, in today in my time) | 00:10 |
| *** arch-nemesis has quit IRC | 00:11 | |
| openstackgerrit | Tim Burke proposed openstack/swift: Fix X-*-Container-Update-Override-* header/footer precedence https://review.openstack.org/338588 | 00:22 |
| openstackgerrit | Tim Burke proposed openstack/swift: Send correct SLO ETag for container updates https://review.openstack.org/337960 | 00:22 |
| clayg | notmyname: the whole thing is suspect, the call to dup in makefile makes no sense - the py3 version doesn't do - gevent doesn't do it - it's burried way back in the history of eventlet code when rdw was working on it - there was an awkward "refcount" implementation *in* eventlet that's all been washed over | 00:31 |
| *** klamath has quit IRC | 00:41 | |
| *** david-lyle has quit IRC | 00:42 | |
| *** m_kazuhiro has joined #openstack-swift | 00:52 | |
| *** Suyash has joined #openstack-swift | 01:04 | |
| *** tqtran has quit IRC | 01:07 | |
| *** klrmn has quit IRC | 01:12 | |
| *** _JZ_ has joined #openstack-swift | 01:13 | |
| openstackgerrit | yuyafei proposed openstack/swift: Add __ne__ built-in function https://review.openstack.org/337428 | 02:01 |
| *** diogogmt has joined #openstack-swift | 02:07 | |
| *** cdelatte has quit IRC | 02:14 | |
| *** klrmn has joined #openstack-swift | 02:18 | |
| *** chsc has joined #openstack-swift | 02:53 | |
| *** chsc has quit IRC | 02:53 | |
| *** chsc has joined #openstack-swift | 02:53 | |
| *** sanchitmalhotra has joined #openstack-swift | 02:54 | |
| *** chsc has quit IRC | 02:59 | |
| *** diazjf has joined #openstack-swift | 03:16 | |
| *** diazjf has quit IRC | 03:20 | |
| *** ManojK has joined #openstack-swift | 03:22 | |
| openstackgerrit | OpenStack Proposal Bot proposed openstack/swift: Updated from global requirements https://review.openstack.org/88736 | 03:31 |
| *** _JZ_ has quit IRC | 03:39 | |
| *** jamielennox is now known as jamielennox|away | 03:43 | |
| *** sgundur1 has left #openstack-swift | 03:49 | |
| *** ppai has joined #openstack-swift | 03:55 | |
| *** jamielennox|away is now known as jamielennox | 03:56 | |
| *** ManojK has quit IRC | 04:01 | |
| *** links has joined #openstack-swift | 04:03 | |
| *** tqtran has joined #openstack-swift | 04:06 | |
| *** sheel has joined #openstack-swift | 04:09 | |
| *** tqtran has quit IRC | 04:11 | |
| *** Suyash has quit IRC | 04:16 | |
| openstackgerrit | Mahati Chamarthy proposed openstack/swift: remove empty db hash and suffix directories https://review.openstack.org/331601 | 04:20 |
| *** nadeem has joined #openstack-swift | 04:21 | |
| *** jamielennox is now known as jamielennox|away | 04:35 | |
| *** jamielennox|away is now known as jamielennox | 04:42 | |
| *** vint_bra has joined #openstack-swift | 04:44 | |
| *** vint_bra has quit IRC | 04:45 | |
| *** vint_bra1 has joined #openstack-swift | 04:45 | |
| *** vint_bra has joined #openstack-swift | 04:45 | |
| *** vint_bra1 has quit IRC | 04:45 | |
| *** vint_bra1 has joined #openstack-swift | 04:47 | |
| *** vint_bra has quit IRC | 04:47 | |
| *** vint_bra1 has quit IRC | 04:51 | |
| *** Jeffrey4l_ has joined #openstack-swift | 04:51 | |
| *** psachin has joined #openstack-swift | 05:01 | |
| *** ChubYann has quit IRC | 05:37 | |
| *** siva_krish has joined #openstack-swift | 05:41 | |
| *** johanfi has quit IRC | 05:42 | |
| *** siva_krish has quit IRC | 05:44 | |
| *** acoles_ has quit IRC | 05:49 | |
| *** zaitcev has quit IRC | 05:50 | |
| *** acoles has joined #openstack-swift | 05:51 | |
| *** ChanServ sets mode: +v acoles | 05:51 | |
| openstackgerrit | Andreas Jaeger proposed openstack/swift: Add install-guide for swift https://review.openstack.org/330070 | 06:00 |
| *** siva_krish has joined #openstack-swift | 06:12 | |
| *** rcernin has joined #openstack-swift | 06:18 | |
| *** klrmn has quit IRC | 06:20 | |
| *** tqtran has joined #openstack-swift | 06:30 | |
| *** SkyRocknRoll has joined #openstack-swift | 06:34 | |
| *** tqtran has quit IRC | 06:34 | |
| *** hseipp has joined #openstack-swift | 06:52 | |
| *** kei_yama has quit IRC | 07:00 | |
| *** tesseract- has joined #openstack-swift | 07:03 | |
| *** kei_yama has joined #openstack-swift | 07:04 | |
| *** nadeem has quit IRC | 07:08 | |
| *** Lickitysplitted has quit IRC | 07:12 | |
| *** Lickitysplitted has joined #openstack-swift | 07:12 | |
| *** ppai has quit IRC | 07:32 | |
| *** SkyRocknRoll has quit IRC | 07:44 | |
| kota_ | Oh, what...??? I'm stll in a fog... | 07:49 |
| kota_ | clayge: Related to (but different from) the conversation at this morning, I'm now checking on sort of closing connection in proxy-server. | 07:50 |
| kota_ | ah, sorry, clayg | 07:50 |
| kota_ | clayg: and I'm wondering why https://github.com/openstack/swift/blob/master/swift/proxy/controllers/base.py#L1100-L1109 works correctly. | 07:51 |
| kota_ | clayg: On my thought, if the conn.getresponse() was timeout, conn reference will be lost and it will leak the socket inside. | 07:52 |
| *** ppai has joined #openstack-swift | 07:52 | |
| kota_ | clayg: but it doesn't actually, when I was trying to inject an error at the Line 1100, nothing leaked. | 07:53 |
| kota_ | clayg: and then, I noticed it is because we wrap the func with GreenPool (in fact, GreenAsyncPile though). | 07:54 |
| kota_ | clayg: https://gist.github.com/bloodeagle40234/e9a1f009336221c085d3d0086e621340 | 07:54 |
| *** oshritf has joined #openstack-swift | 07:54 | |
| *** oshritf_ has joined #openstack-swift | 07:54 | |
| kota_ | clayg: that is a snippet, I was trying to do. The code shows, if we loose the reference of connection, it causes socket leak but if we do so in GreenPile, the connection will be closed. | 07:56 |
| *** mvk has joined #openstack-swift | 07:57 | |
| kota_ | clayg: I thought GreenPool serves just a green thread for a hub of the networking tranpoline but... it semms like correcting the garbage? I'm not entirly sure and I didn't find how it was closed actually. | 07:59 |
| kota_ | ah, I wrote as to clayg, but if someone knows something on that, that's great to correct me anyway. | 08:00 |
| kota_ | what's happen in actually, and I'm not sure if we can trust the GreenPile to close the connection always. | 08:01 |
| *** cbartz has joined #openstack-swift | 08:04 | |
| *** rledisez has joined #openstack-swift | 08:06 | |
| *** ppai has quit IRC | 08:06 | |
| *** kei_yama_ has joined #openstack-swift | 08:08 | |
| *** kei_yama has quit IRC | 08:09 | |
| openstackgerrit | Davanum Srinivas (dims) proposed openstack/swift: [WIP] Testing latest u-c https://review.openstack.org/318441 | 08:10 |
| *** takashi has joined #openstack-swift | 08:11 | |
| mahatic_ | pdardeau: notmyname thanks for calling out on veg option, but I'm not going to make it to the mid-cycle this time (lot's of interesting topics on etherpad i'm going to miss out on :( ) | 08:15 |
| * mahatic_ got around the logs only now after a holiday at intel india | 08:16 | |
| *** openstackgerrit has quit IRC | 08:18 | |
| *** openstackgerrit has joined #openstack-swift | 08:18 | |
| *** pcaruana has joined #openstack-swift | 08:30 | |
| *** siva_krish has quit IRC | 08:34 | |
| *** siva_krish has joined #openstack-swift | 08:35 | |
| *** mmcardle has joined #openstack-swift | 08:44 | |
| *** thurloat has quit IRC | 08:48 | |
| *** thurloat has joined #openstack-swift | 08:48 | |
| *** asettle has joined #openstack-swift | 08:50 | |
| openstackgerrit | yuyafei proposed openstack/swift: Correct reraising of exception https://review.openstack.org/338774 | 09:00 |
| *** oshritf has quit IRC | 09:01 | |
| *** oshritf_ has quit IRC | 09:01 | |
| *** ouchkernel has quit IRC | 09:04 | |
| *** mmcardle has quit IRC | 09:09 | |
| *** ouchkernel has joined #openstack-swift | 09:10 | |
| *** takashi has quit IRC | 09:14 | |
| *** mmcardle has joined #openstack-swift | 09:15 | |
| *** mmcardle has quit IRC | 09:25 | |
| *** daemontool has joined #openstack-swift | 10:07 | |
| openstackgerrit | Kota Tsuyuzaki proposed openstack/swift: Ensure close socket for memcached if got timeout https://review.openstack.org/338819 | 10:08 |
| *** mmcardle has joined #openstack-swift | 10:08 | |
| *** oshritf_ has joined #openstack-swift | 10:14 | |
| *** oshritf has joined #openstack-swift | 10:14 | |
| *** thurloat has quit IRC | 10:25 | |
| *** thurloat has joined #openstack-swift | 10:26 | |
| psachin | ping clayg, https://review.openstack.org/#/c/334238/. Error and assert the message has been taken care off. https://github.com/psachin/swift/commit/da7ef05bb2aa0d6fc1a5a172b8d0af54aedccabb | 10:31 |
| patchbot | psachin: patch 334238 - swift - Fix swift-get-nodes arg parsing for missing ring | 10:31 |
| *** tqtran has joined #openstack-swift | 10:32 | |
| psachin | clayg, Not sure how to implement tests for invalid/defined policies. | 10:32 |
| *** tqtran has quit IRC | 10:37 | |
| *** siva_krish has quit IRC | 10:40 | |
| *** sanchitmalhotra has quit IRC | 10:42 | |
| *** mmcardle has quit IRC | 10:52 | |
| kota_ | wow....it's hard work to rebase off global ec staff to the current master. not a big from encryption but around chunk_hasher seems too affected... | 10:58 |
| kota_ | test seems to work well but not sure right now, I succeeded the rebasing... | 10:59 |
| kota_ | and probably it needs more clean up | 10:59 |
| openstackgerrit | Kota Tsuyuzaki proposed openstack/swift: PUT/GET path for global ec cluster https://review.openstack.org/219165 | 11:01 |
| openstackgerrit | Kota Tsuyuzaki proposed openstack/swift: Support EC duplication factor at reconstructor https://review.openstack.org/237867 | 11:06 |
| kota_ | probably it seems getting better to squash the later one into the previous one because about 100 lines changes in it. | 11:07 |
| *** mmcardle has joined #openstack-swift | 11:08 | |
| *** _JZ_ has joined #openstack-swift | 11:09 | |
| *** m3m0 has quit IRC | 11:15 | |
| openstackgerrit | Kota Tsuyuzaki proposed openstack/swift: Add Pros/Cons table for global clusetr consideration https://review.openstack.org/304382 | 11:21 |
| *** dfflanders has quit IRC | 11:23 | |
| *** mmcardle has quit IRC | 11:26 | |
| *** asettle has quit IRC | 11:26 | |
| openstackgerrit | Kota Tsuyuzaki proposed openstack/swift: CompositeRing Support https://review.openstack.org/271920 | 11:36 |
| kota_ | All of them are rebasing. | 11:36 |
| *** SkyRocknRoll has joined #openstack-swift | 11:38 | |
| *** cdelatte has joined #openstack-swift | 11:48 | |
| *** SkyRocknRoll has quit IRC | 12:03 | |
| *** arcimboldo has joined #openstack-swift | 12:08 | |
| *** NM has joined #openstack-swift | 12:18 | |
| openstackgerrit | Kota Tsuyuzaki proposed openstack/swift: PUT/GET path for global ec cluster https://review.openstack.org/219165 | 12:28 |
| openstackgerrit | Kota Tsuyuzaki proposed openstack/swift: Support EC duplication factor at reconstructor https://review.openstack.org/237867 | 12:29 |
| openstackgerrit | Kazuhiro MIYAHARA proposed openstack/swift: Remove an unused variable from tests.py https://review.openstack.org/338939 | 12:29 |
| *** tqtran has joined #openstack-swift | 12:33 | |
| *** tqtran has quit IRC | 12:38 | |
| *** m_kazuhiro has quit IRC | 12:42 | |
| *** m_kazuhiro has joined #openstack-swift | 12:46 | |
| *** asettle has joined #openstack-swift | 12:46 | |
| *** arcimboldo has quit IRC | 12:47 | |
| openstackgerrit | Ondřej Nový proposed openstack/swift: Add ringbuilder tests for --yes option https://review.openstack.org/326373 | 12:50 |
| openstackgerrit | Alistair Coles proposed openstack/swift: Fix unicode errors in object controller logging https://review.openstack.org/338950 | 12:52 |
| *** oshritf_ has quit IRC | 12:53 | |
| *** oshritf has quit IRC | 12:53 | |
| onovy | hi, https://review.openstack.org/#/c/326399/ https://review.openstack.org/#/c/253037/ this was waiting for +A after crypto freeze, which is over, right? | 12:53 |
| patchbot | onovy: patch 326399 - swift - Add simple multiple server type test | 12:53 |
| patchbot | onovy: patch 253037 - swift - Call swift-recon with more than one server type | 12:53 |
| acoles | cschwede: patch 338950 has the tests that I mentioned I was working on re. unicode errors. They don't yet test translations, but I think they could be extended to test unicode translations of the error messages. | 12:55 |
| patchbot | acoles: https://review.openstack.org/#/c/338950/ - swift - Fix unicode errors in object controller logging | 12:55 |
| cschwede | acoles: thx, will have a look! | 12:55 |
| *** m_kazuhiro has quit IRC | 12:56 | |
| onovy | acoles: thanks! | 12:56 |
| acoles | onovy: done | 12:56 |
| acoles | cschwede: I think some of those new tests will fail if the error message is translated to have non-ascii (without your fix for gettext another fix) | 12:57 |
| cschwede | acoles: sounds reasonable | 12:58 |
| cschwede | i thinkg clayg might be interested in patch 338950 as well | 12:58 |
| patchbot | cschwede: https://review.openstack.org/#/c/338950/ - swift - Fix unicode errors in object controller logging | 12:58 |
| *** arcimboldo has joined #openstack-swift | 12:59 | |
| *** klamath has joined #openstack-swift | 13:01 | |
| *** klamath has quit IRC | 13:01 | |
| *** links has quit IRC | 13:02 | |
| *** klamath has joined #openstack-swift | 13:02 | |
| jrichli | mahatic_: you will be missed at the hackathon! | 13:04 |
| *** ManojK has joined #openstack-swift | 13:06 | |
| *** diogogmt has quit IRC | 13:21 | |
| *** diogogmt has joined #openstack-swift | 13:26 | |
| *** oshritf_ has joined #openstack-swift | 13:27 | |
| *** oshritf has joined #openstack-swift | 13:27 | |
| *** diogogmt has quit IRC | 13:40 | |
| *** vinsh has joined #openstack-swift | 13:43 | |
| *** arcimboldo has quit IRC | 13:46 | |
| *** mmcardle has joined #openstack-swift | 13:51 | |
| *** mmcardle has quit IRC | 13:59 | |
| *** mmcardle has joined #openstack-swift | 14:02 | |
| *** arcimboldo has joined #openstack-swift | 14:02 | |
| *** mmcardle1 has joined #openstack-swift | 14:04 | |
| *** sheel has quit IRC | 14:05 | |
| *** mmcardle has quit IRC | 14:06 | |
| *** ManojK has quit IRC | 14:11 | |
| *** mmcardle has joined #openstack-swift | 14:12 | |
| *** mmcardle1 has quit IRC | 14:15 | |
| *** mmcardle1 has joined #openstack-swift | 14:24 | |
| *** mmcardle has quit IRC | 14:26 | |
| *** sgundur1 has joined #openstack-swift | 14:29 | |
| *** ManojK has joined #openstack-swift | 14:30 | |
| *** vinsh has quit IRC | 14:32 | |
| *** diogogmt has joined #openstack-swift | 14:32 | |
| *** Suyash has joined #openstack-swift | 14:33 | |
| openstackgerrit | Merged openstack/swift: Add simple multiple server type test https://review.openstack.org/326399 | 14:34 |
| openstackgerrit | Kota Tsuyuzaki proposed openstack/swift: CompositeRing Support https://review.openstack.org/271920 | 14:34 |
| acoles | jrichli: timburke I was about to abandon patch 327144 but thought some more and wonder if there is still some value in that modification - any thoughts? | 14:43 |
| patchbot | acoles: https://review.openstack.org/#/c/327144/ - swift (feature/crypto) - crypto - Don't encrypt update override etags for e... | 14:43 |
| timburke | acoles: yeah, now that we always use a random iv, i think there's no threat | 14:45 |
| acoles | timburke: ok, I'll rebase it | 14:49 |
| *** ametts has joined #openstack-swift | 14:51 | |
| jrichli | acoles: I'll take a look. can't right now, but will soon | 14:55 |
| acoles | jrichli: thanks. IIRC it was addressing an anomaly that you spotted. | 14:56 |
| acoles | we can discuss it more next week | 14:56 |
| *** siva_krish has joined #openstack-swift | 14:56 | |
| jrichli | acoles: ok, thanks. I took a quick look. i need to verify a couple things in our current impl now. | 14:57 |
| openstackgerrit | Mohit Motiani proposed openstack/swift: Fix typo in the account-server.conf manpage https://review.openstack.org/339045 | 14:57 |
| *** ManojK has quit IRC | 14:58 | |
| *** ManojK has joined #openstack-swift | 15:01 | |
| *** siva_krish has quit IRC | 15:03 | |
| *** diazjf has joined #openstack-swift | 15:03 | |
| *** diazjf1 has joined #openstack-swift | 15:05 | |
| *** diazjf has quit IRC | 15:08 | |
| *** vinsh has joined #openstack-swift | 15:09 | |
| *** mmcardle1 has quit IRC | 15:11 | |
| acoles | mmotiani: ^^ the same typo is in the container and object server conf manpages | 15:11 |
| mmotiani | acoles: Yeah, thanks. | 15:18 |
| mmotiani | Will fix that too | 15:18 |
| mmotiani | thanks :) | 15:18 |
| openstackgerrit | Mohit Motiani proposed openstack/swift: Fix typo in object-server.conf and container-server.conf manpage https://review.openstack.org/339063 | 15:23 |
| *** mmcardle has joined #openstack-swift | 15:23 | |
| *** nadeem has joined #openstack-swift | 15:24 | |
| *** arcimboldo has quit IRC | 15:24 | |
| *** nadeem has quit IRC | 15:25 | |
| *** nadeem has joined #openstack-swift | 15:25 | |
| *** ManojK has quit IRC | 15:26 | |
| *** chsc has joined #openstack-swift | 15:26 | |
| openstackgerrit | Alistair Coles proposed openstack/swift: Don't encrypt update override etags for empty object https://review.openstack.org/339071 | 15:28 |
| *** ManojK has joined #openstack-swift | 15:28 | |
| acoles | mmotiani: thanks | 15:29 |
| *** psachin has quit IRC | 15:37 | |
| *** oshritf has quit IRC | 15:40 | |
| *** oshritf_ has quit IRC | 15:40 | |
| *** pcaruana has quit IRC | 15:49 | |
| *** jmccarthy has quit IRC | 15:50 | |
| *** jmccarthy has joined #openstack-swift | 15:51 | |
| *** ManojK has quit IRC | 16:04 | |
| *** mmcardle has quit IRC | 16:09 | |
| *** rcernin has quit IRC | 16:10 | |
| notmyname | good morning | 16:12 |
| *** hseipp has quit IRC | 16:14 | |
| *** arch-nemesis has joined #openstack-swift | 16:14 | |
| *** siva_krish has joined #openstack-swift | 16:18 | |
| *** tesseract- has quit IRC | 16:23 | |
| *** chsc has quit IRC | 16:23 | |
| *** ManojK has joined #openstack-swift | 16:24 | |
| *** chsc has joined #openstack-swift | 16:32 | |
| *** chsc has joined #openstack-swift | 16:32 | |
| *** chsc has quit IRC | 16:33 | |
| *** chsc has joined #openstack-swift | 16:33 | |
| *** chsc has joined #openstack-swift | 16:33 | |
| *** dmorita has joined #openstack-swift | 16:39 | |
| *** asettle has quit IRC | 16:45 | |
| *** rledisez has quit IRC | 16:48 | |
| *** cdelatte has quit IRC | 16:58 | |
| *** cbartz has left #openstack-swift | 17:11 | |
| openstackgerrit | Merged openstack/python-swiftclient: Add python version constraint python>=2.7 https://review.openstack.org/334333 | 17:13 |
| openstackgerrit | Maria Malyarova proposed openstack/swift: Another amendment with missing parenthesis https://review.openstack.org/339145 | 17:15 |
| *** klrmn has joined #openstack-swift | 17:18 | |
| *** diazjf1 has quit IRC | 17:20 | |
| openstackgerrit | Merged openstack/swift: Fix intermittent bulk delete unit test failures https://review.openstack.org/335674 | 17:38 |
| openstackgerrit | Alistair Coles proposed openstack/swift: Add encryption package requirements to SAIO instructions https://review.openstack.org/339156 | 17:40 |
| *** acoles is now known as acoles_ | 17:41 | |
| *** daemontool has quit IRC | 17:52 | |
| *** catintheroof has joined #openstack-swift | 17:53 | |
| *** tqtran has joined #openstack-swift | 18:00 | |
| *** dmorita has quit IRC | 18:06 | |
| *** ManojK has quit IRC | 18:07 | |
| *** dmorita_ has joined #openstack-swift | 18:08 | |
| *** sheel has joined #openstack-swift | 18:12 | |
| *** gyee has joined #openstack-swift | 18:17 | |
| *** vinsh has quit IRC | 18:22 | |
| *** nadeem has quit IRC | 18:24 | |
| *** madorn has joined #openstack-swift | 18:25 | |
| madorn | is anyone aware of issues with cross-project ACL's when using keystoneauth? | 18:25 |
| *** bgmccollum has joined #openstack-swift | 18:25 | |
| *** nadeem has joined #openstack-swift | 18:25 | |
| *** nadeem has quit IRC | 18:26 | |
| madorn | it appears that in both mitaka and liberty openstack environments with swift, one cannot give a different project read/write access to a container | 18:26 |
| openstackgerrit | Merged openstack/swift: Fix typo in the account-server.conf manpage https://review.openstack.org/339045 | 18:28 |
| *** nadeem has joined #openstack-swift | 18:28 | |
| openstackgerrit | Merged openstack/swift: Fix typo in object-server.conf and container-server.conf manpage https://review.openstack.org/339063 | 18:30 |
| openstackgerrit | Merged openstack/swift: make print python3 compatible https://review.openstack.org/333801 | 18:30 |
| *** siva_krish has quit IRC | 18:32 | |
| madorn | madorn: figured it out... --os-storage-url | 18:35 |
| *** rcernin has joined #openstack-swift | 18:39 | |
| *** ChubYann has joined #openstack-swift | 18:41 | |
| *** siva_krish has joined #openstack-swift | 18:41 | |
| *** tqtran is now known as tqtran-afk | 18:46 | |
| *** diazjf has joined #openstack-swift | 18:50 | |
| *** ManojK has joined #openstack-swift | 18:50 | |
| *** daemontool has joined #openstack-swift | 18:54 | |
| clayg | is patch 339071 just belts and braces or some sort of rare failure in metadata handling? | 19:04 |
| patchbot | clayg: https://review.openstack.org/#/c/339071/ - swift - Don't encrypt update override etags for empty object | 19:04 |
| clayg | ... there's no associated bug - so i'm inclined to think it's just for "more better" | 19:04 |
| openstackgerrit | Tim Burke proposed openstack/swift: Run flake8 against scripts in bin https://review.openstack.org/339191 | 19:05 |
| openstackgerrit | Tim Burke proposed openstack/swift: Turn on H233 and start using print function https://review.openstack.org/339192 | 19:05 |
| clayg | timburke: cschwede: acoles_: nice working merging doc fixes! | 19:05 |
| openstackgerrit | Merged openstack/swift: Remove an unused variable from tests.py https://review.openstack.org/338939 | 19:06 |
| jrichli | clayg: I remember that logically there was a hole in one of the crypto patches awhile back. This plugged it. I will have to revisit the details and compare with what is on master now. | 19:06 |
| jrichli | just haven't been able to do that yet today :-( | 19:06 |
| timburke | there's a chance it had to do with leaking etags when multiple objects were PUT to the same path, but i need to think about whether it's still an issue now that we're always using random IVs | 19:08 |
| jrichli | timburke: yes, but technically, we do have the hmac where it will also result in the same value for the same path. but maybe that doesn't get involved here at all. again, i have to see details | 19:09 |
| *** diazjf has quit IRC | 19:09 | |
| jrichli | oh, right ... this is just for container listing stuff, right? | 19:10 |
| *** diazjf has joined #openstack-swift | 19:10 | |
| timburke | jrichli: yeah, the hmac seems unavoidable... and yeah, that patch just has to do with container listings | 19:10 |
| clayg | ugh, but I prefer print as a statement :'( | 19:10 |
| jrichli | but some of the tests added a case for other mw's adding footers. i wanted to be sure we are testing all that | 19:11 |
| timburke | clayg: we keep getting piecemeal patches to convert it, and if we're ever going to move to py3 it'll be necessary...may as well get used to it now :-/ | 19:11 |
| timburke | jrichli: speaking of other middlewares and container listing overrides, you might be interested in patch 337960 | 19:12 |
| patchbot | timburke: https://review.openstack.org/#/c/337960/ - swift - Send correct SLO ETag for container updates | 19:12 |
| jrichli | timburke: its already on my list! now i just need time to go through my list :-/ | 19:13 |
| timburke | thanks :-) don't forget its parent, patch 338588 | 19:13 |
| patchbot | timburke: https://review.openstack.org/#/c/338588/ - swift - Fix X-*-Container-Update-Override-* header/footer ... | 19:13 |
| jrichli | kk | 19:13 |
| openstackgerrit | Merged openstack/swift: Another amendment with missing parenthesis https://review.openstack.org/339145 | 19:21 |
| *** ametts has quit IRC | 19:23 | |
| *** cdelatte has joined #openstack-swift | 19:25 | |
| openstackgerrit | Merged openstack/swift: added note to testFileSizeLimit functional test https://review.openstack.org/331318 | 19:26 |
| openstackgerrit | Merged openstack/swift: Add encryption package requirements to SAIO instructions https://review.openstack.org/339156 | 19:27 |
| *** Jeffrey4l_ has quit IRC | 19:31 | |
| *** Jeffrey4l_ has joined #openstack-swift | 19:31 | |
| *** siva_krish has left #openstack-swift | 19:33 | |
| openstackgerrit | Thiago da Silva proposed openstack/swift: Tighten memcached tests https://review.openstack.org/333555 | 19:35 |
| openstackgerrit | Tim Burke proposed openstack/swift: Run flake8 against scripts in bin https://review.openstack.org/339191 | 19:38 |
| openstackgerrit | Tim Burke proposed openstack/swift: Turn on H233 and start using print function https://review.openstack.org/339192 | 19:38 |
| timburke | tdasilva: mind looking at the parent change, too? patch 333554 | 19:39 |
| patchbot | timburke: https://review.openstack.org/#/c/333554/ - swift - MockMemcached cleanup | 19:39 |
| tdasilva | timburke: approving that now | 19:39 |
| tdasilva | looks good | 19:39 |
| timburke | ah, perfect :-) thanks | 19:39 |
| tdasilva | timburke: great job btw | 19:40 |
| clayg | shouldn't the "run flake8 on bin scripts; and add more checks to flake8 and fixes in bin scripts" patches be squashed or reversed? | 19:43 |
| openstackgerrit | Merged openstack/swift: Add install-guide for swift https://review.openstack.org/330070 | 19:48 |
| *** asettle has joined #openstack-swift | 19:51 | |
| briancline | oh wow, patch 335373 got merged | 19:54 |
| patchbot | briancline: https://review.openstack.org/#/c/335373/ - swift - Prevent down nodes failing PUTs with non-ascii obj... (MERGED) | 19:54 |
| clayg | mahatic_: timburke fixed the tox/flake/pep8/merge issue in patch 339192 - but gerrit didn't remove your -1 because it thinks it was "just" a rebase | 19:55 |
| patchbot | clayg: https://review.openstack.org/#/c/339192/ - swift - Turn on H233 and start using print function | 19:55 |
| briancline | so was i just not moving quickly enough on that one? | 19:56 |
| *** asettle has quit IRC | 19:56 | |
| clayg | timburke: oh geez - I still don't know how to read new gerrits related changes list - the order you have the patches is correct obvs. - sorry | 19:56 |
| clayg | briancline: what's wrong with it merging? | 19:57 |
| *** asettle has joined #openstack-swift | 19:59 | |
| *** asettle has quit IRC | 19:59 | |
| briancline | clayg: nothing, i just thought there was more to be done. i guess i was confused whether i should continue on with it after your patchset :/ so i started working on some updates to its tests | 20:02 |
| clayg | briancline: oh sorry, yeah that bug was a hozer - and acoles_ picked up the ball anyway | 20:08 |
| clayg | thanks for getting everyone pointed in the right direction - if you ever have time to work on something that isn't a critical bug maybe you'll get to enjoy the month long review process! | 20:09 |
| timburke | mmotiani: i think clayg meant to tag you about patch 339192 above | 20:20 |
| patchbot | timburke: https://review.openstack.org/#/c/339192/ - swift - Turn on H233 and start using print function | 20:20 |
| mmotiani | timburke: got it, thanks! | 20:28 |
| mmotiani | thanks clayg | 20:28 |
| tdasilva | timburke: re patch 339192, why do some of the files not require "from __future__ import print_function" ? | 20:29 |
| patchbot | tdasilva: https://review.openstack.org/#/c/339192/ - swift - Turn on H233 and start using print function | 20:29 |
| *** castulo has joined #openstack-swift | 20:30 | |
| castulo | hi guys, question, I have seen it takes a while after having set account quota in swift to be enforced, more than 10 minutes in my devstack environment. Is there a way I can tell when the quota is finally being enforced? Or is there a way to force the account quota to be enforced sooner? | 20:33 |
| *** daemontool has quit IRC | 20:36 | |
| clayg | timburke: thanks! | 20:39 |
| clayg | tdasilva: which files don't require it!? | 20:39 |
| clayg | tdasilva: maybe ones that don't use print? | 20:39 |
| tdasilva | https://review.openstack.org/#/c/339192/2/bin/swift-drive-audit | 20:39 |
| patchbot | tdasilva: patch 339192 - swift - Turn on H233 and start using print function | 20:39 |
| tdasilva | clayg: maybe I'm confused, just don't see it there for example | 20:40 |
| clayg | castulo: it's up to the speed of the container updater cycle time - in devstack you can run like idk, just run one cycle in the foreground? `swift-init container-updater once -nv` | 20:40 |
| clayg | tdasilva: how interesting! | 20:42 |
| castulo | clayg: thanks, I'll try that | 20:43 |
| clayg | tdasilva: wow, I guess like the parans can just be enclosing? like print (1 == 2) is reasonable when print is a statement | 20:44 |
| clayg | tdasilva: but it should be fixed in the patch, annoying to lint for I guess - print('foo', x) printing a tuple because the future import is missing is going to be annoying | 20:45 |
| clayg | change is fun! | 20:45 |
| clayg | rewrite it in golang | 20:45 |
| *** sheel has quit IRC | 20:45 | |
| *** diazjf has quit IRC | 20:46 | |
| *** tqtran-afk is now known as tqtran | 20:52 | |
| *** diazjf has joined #openstack-swift | 20:55 | |
| *** klamath_ has joined #openstack-swift | 21:06 | |
| *** klamath has quit IRC | 21:08 | |
| timburke | tdasilva: it comes down to whether the print usage would actually need the function instead of the statement. print("foo") works fine on both py2 and py3, without needing the __future__ import. print("foo", end="") doesn't and raises a SyntaxError. print() doesn't and fails in a subtle way | 21:12 |
| openstackgerrit | Merged openstack/swift: Add __ne__ built-in function https://review.openstack.org/337428 | 21:13 |
| timburke | i could've just imported print_function everywhere, but i was lazy | 21:13 |
| openstackgerrit | Merged openstack/swift: MockMemcached cleanup https://review.openstack.org/333554 | 21:14 |
| *** rcernin has quit IRC | 21:17 | |
| notmyname | FYI https://review.openstack.org/#/c/339175/ | 21:26 |
| patchbot | notmyname: patch 339175 - governance - [WIP] Community Input for alternate Programming La... | 21:26 |
| openstackgerrit | Nandini Tata proposed openstack/swift: Fixed inconsistent naming conventions https://review.openstack.org/339242 | 21:26 |
| *** NM has quit IRC | 21:31 | |
| *** ozialien10 has joined #openstack-swift | 21:35 | |
| clayg | timburke: well if there's any bugs with newlines that's all the more reason to move that code out of /bin into tested methods in /cli ;) | 21:37 |
| castulo | clayg: I set a container quota, then I ran the `swift-init container-updater once -nv` as you suggested, then I tried to upload an object larger than the quota and it was still uploaded... | 21:38 |
| *** arcimboldo has joined #openstack-swift | 21:39 | |
| castulo | clayg: anything else I can do to make swift enforce the quota? I'm seeing if I can do something in Tempest regarding bug https://bugs.launchpad.net/tempest/+bug/1310597 | 21:39 |
| openstack | Launchpad bug 1310597 in tempest "race between quota setting and quota being enforced" [High,Confirmed] | 21:39 |
| *** diogogmt has quit IRC | 21:40 | |
| openstackgerrit | Nandini Tata proposed openstack/swift: Fixed inconsistent naming conventions https://review.openstack.org/339242 | 21:40 |
| clayg | castulo: container quotas don't need the container-updater to run like account quotas do | 21:41 |
| clayg | castulo: they are subject to the recheck_container_existence timeout form memcache I 'spose - uploading a new object doesn't nessecarily invalidate the container_info cache values in memcache | 21:42 |
| *** diogogmt has joined #openstack-swift | 21:47 | |
| notmyname | castulo: also remember that quotas in swift aren't hard. it's known an documented and expected that there will be a little "slop" in the values that are enforced | 21:48 |
| clayg | notmyname: but it's not like we do `if quota_over_limit() and random.random() > 0.5` it's only going to wiff some something hasn't updated yet and stuff updates out-of-band | 21:49 |
| notmyname | of course. enforced for the current view that the container/account has, and those don't get updated synchronously with every object put/delete in the cluster | 21:50 |
| castulo | clayg: my mistake, what I set was the account quota, not container quota | 21:50 |
| notmyname | but I didn't want castulo to be expecting something that doesnt' happen | 21:50 |
| torgomatic | that'd be neat, though... between 90% and 100% of quota, start randomly failing requests in proportion to fullness (so 90% = no failures, 95% = half failures, 100% = all failures) | 21:52 |
| torgomatic | it'd be sort of like trying to pack a suitcase really full | 21:52 |
| notmyname | or trying tog et your kid to clean up his room | 21:52 |
| castulo | notmyname: so what do you mean by a "little slop in the values that are enforce"? Sorry if my question is too basic, I'm new to swift | 21:53 |
| notmyname | castulo: no worries. let me find the link... | 21:53 |
| notmyname | castulo: http://docs.openstack.org/developer/swift/api/container_quotas.html | 21:53 |
| notmyname | castulo: specifically, the last 2 paragraphs there | 21:53 |
| castulo | notmyname: I see, and does that apply also to the x-account-meta-quota-bytes? My issue at the moment is with account quota, not container quota | 21:57 |
| clayg | notmyname: that page renders sorta shit - do we need to update some sphinx templates or something? | 21:57 |
| notmyname | clayg: probably oslosphinx. that's where the theme is. | 21:58 |
| clayg | notmyname: the navigation is sort of crap too - those are "the new api docs"? | 21:59 |
| notmyname | the versions there? | 21:59 |
| clayg | notmyname: that and also there's no like link to the TOC that I can find? | 21:59 |
| clayg | castulo: "The same limitations apply for the account quotas as for the container quotas" -> http://docs.openstack.org/developer/swift/middleware.html#module-swift.common.middleware.account_quotas | 22:00 |
| torgomatic | castulo: yeah, it also applies to the account... actually it's even worse; for containers, "might not be immediately updated" means "is usually immediately updated, but sometimes has a delay" | 22:00 |
| torgomatic | for accounts, it's more like "is basically never immediately updated unless you get really lucky in your timing" | 22:01 |
| openstackgerrit | Shashirekha Gundur proposed openstack/python-swiftclient: fixing nit picks in python-swiftclient doc https://review.openstack.org/339249 | 22:01 |
| *** diazjf has quit IRC | 22:01 | |
| castulo | I see, and there is no way to really know when that update has been done, only by trying I guess? | 22:02 |
| *** ManojK has quit IRC | 22:02 | |
| openstackgerrit | Shashirekha Gundur proposed openstack/python-swiftclient: fixing nit picks in python-swiftclient doc https://review.openstack.org/339249 | 22:03 |
| torgomatic | yep; all the container -> account updates are done by background daemons running in the cluster somewhere, not by anything in the HTTP-serving path | 22:03 |
| torgomatic | (otherwise you'd incur massive update pressure on the account DB when the object PUT rate was high) | 22:04 |
| notmyname | well, you look at logs and other metrics as part of the overall cluster monitoring. eg watch the updater cycle times | 22:04 |
| *** ManojK has joined #openstack-swift | 22:04 | |
| torgomatic | right, but as a consumer of the HTTP API, you have no idea when your object PUT will be reflected in account stats | 22:05 |
| torgomatic | and hence quotas | 22:05 |
| clayg | there's no good way in a test - someone said quotas were a bad idea one time for this reason - then we did it anyway | 22:05 |
| notmyname | rigth | 22:05 |
| castulo | torgomatic: I understand now, then I don't think there is really anything that can be done to fix bug https://bugs.launchpad.net/tempest/+bug/1310597 in Tempest, probably just completely remove the test | 22:05 |
| openstack | Launchpad bug 1310597 in tempest "race between quota setting and quota being enforced" [High,Confirmed] | 22:05 |
| notmyname | clayg: well not just for this reason. I still don't like quotas, but I got over that a while back ;-) | 22:05 |
| torgomatic | clayg: can we find that person and hire them? :p | 22:05 |
| notmyname | torgomatic: I think it was me | 22:06 |
| notmyname | torgomatic: so....yes ;-) | 22:06 |
| torgomatic | hehe | 22:06 |
| *** darrenc is now known as darrenc_afk | 22:06 | |
| clayg | castulo: probably not in your stock unittest-y framework - you need something like a burn in CI where you do some upload yesterday that pushes over quota - then today make sure you can't upload anymore and delete the account and create a new one for tomorrow | 22:06 |
| clayg | i'm guessin you don't have infrastructure around long enough for that kind of test tho | 22:07 |
| clayg | so... what-do-you-do | 22:07 |
| notmyname | clayg: that sounds really similar to what hurricanerix had built a while back for rax deployments | 22:07 |
| torgomatic | castulo: yeah, it's not really amenable to a test from the consumer standpoint... it's really more suitable as a Swift probetest where the test code actually controls the running of the various daemons, so it can force an update when it wants | 22:07 |
| clayg | notmyname: yeah rax guys are all about that shit - i'm not sure how much "support" for that kind of process is baked into cloudcafe | 22:08 |
| clayg | torgomatic: all new middleware that requires probetests would be a great deterrent - i guess we have container_sync | 22:09 |
| *** ManojK has quit IRC | 22:09 | |
| notmyname | clayg: FWIW https://bugs.launchpad.net/oslosphinx/+bug/1600029 | 22:09 |
| openstack | Launchpad bug 1600029 in oslosphinx "short pages are poorly rendered" [Undecided,New] | 22:09 |
| castulo | clayg: cool, I will update the bug then and will bring up the discussion with the QA team to see if the test should be removed from Tempest then. | 22:09 |
| castulo | thanks for your help Guys! :) | 22:09 |
| jrichli | clayg torgomatic notmyname : is there something that a swift functional test could do to make this quota update happen? patch 314099 is sort of stuck at this same question | 22:14 |
| patchbot | jrichli: https://review.openstack.org/#/c/314099/ - swift - Container Object Count Quota Functional Test | 22:14 |
| openstackgerrit | Shashirekha Gundur proposed openstack/python-swiftclient: fixing nit picks in python-swiftclient doc https://review.openstack.org/339249 | 22:16 |
| *** arcimboldo has quit IRC | 22:18 | |
| clayg | notmyname: nice work! | 22:18 |
| clayg | what if you just uploaded bytes into a container until a HEAD request seemed to reliably return some number of bytes - then set the quota to half of that (via POST which clears memcache) - i bet then you could mostly sort of expect/hope that you get an over quota limit (for containers) | 22:20 |
| clayg | i bet 80% of the time it would work 100% of the time | 22:21 |
| *** ozialien10 has quit IRC | 22:22 | |
| jrichli | clayg: worth a shot, thanks | 22:24 |
| *** darrenc_afk is now known as darrenc | 22:26 | |
| clayg | jrichli: sorry, i was joking - i don't think trying to test these things from a blackbox perspective is an exercise that will yeild value | 22:27 |
| clayg | jrichli: torgomatic is right - we could do it in a whitebox probetest were we take control of all the asyncrouns actions and all failures are controlled | 22:28 |
| openstackgerrit | Mohit Motiani proposed openstack/swift: Add region in swift-ring-builder add https://review.openstack.org/319342 | 22:30 |
| *** catintheroof has quit IRC | 22:40 | |
| kong | hi, guys, do you if there is an efficient way to get object upload/download bytes if users successfully do that? | 22:55 |
| kong | I am thinking about the traffic billing | 22:55 |
| kong | I found an in-progress spec related to that, but maybe not implemented. https://specs.openstack.org/openstack/swift-specs/specs/in_progress/notifications.html | 22:56 |
| openstackgerrit | Merged openstack/swift: Python 3: Fix basestring, long and StringIO https://review.openstack.org/333297 | 23:01 |
| openstackgerrit | Merged openstack/swift: Python 3: dict.iteritems() and dict.itervalues() https://review.openstack.org/333298 | 23:03 |
| clayg | kong: everyone does network bytes for traffic billing from proxy logs, but it's not baked in because everyone seems to roll it different | 23:03 |
| clayg | kong: slogging is open source, at some level everyone's code looks basically like this -> https://github.com/notmyname/slogging/blob/master/slogging/access_processor.py | 23:03 |
| *** manous has joined #openstack-swift | 23:04 | |
| kong | clayg: thanks for your reply first. So, swift don't support emit event yet, right? | 23:05 |
| *** _JZ_ has quit IRC | 23:05 | |
| *** garthb has joined #openstack-swift | 23:05 | |
| notmyname | kong: correct. swift does emit statsd messages with some aggregated transfer metrics, but it's not appropriate for billing since it isn't per user/account/tenant/whatever | 23:07 |
| openstackgerrit | Merged openstack/swift: Correct reraising of exception https://review.openstack.org/338774 | 23:07 |
| openstackgerrit | Merged openstack/swift: Tighten memcached tests https://review.openstack.org/333555 | 23:07 |
| notmyname | kong: the best way to get info for billing from a swift cluster is to process the logs | 23:07 |
| kota_ | good morning | 23:09 |
| kong | notmyname: ok, seems like the log analysis is the preferred way | 23:09 |
| kota_ | clayg: do you have a time to look at https://gist.github.com/bloodeagle40234/e9a1f009336221c085d3d0086e621340 a bit? | 23:09 |
| kong | clayg, notmyname, thanks you guys! | 23:09 |
| notmyname | kong: there's also some degree of ceilometer integration, I think, that you could use. it sets up some polling workers and message queues that you could use. it would probably be ok for small, low-traffic clusters, but I have serious concerns about using it at scale | 23:10 |
| kong | notmyname: is there any doc i can refer to? for ceilometer integration? | 23:11 |
| *** siva_krish has joined #openstack-swift | 23:12 | |
| *** siva_krish has quit IRC | 23:12 | |
| torgomatic | if you're thinking of using a message queue for computing someone's bill, just remember the two hard problems of distributed computing | 23:12 |
| notmyname | kong: unfortunately, I don't know. I mean, I'm sure there are somewhere. I'd start exactly where you would: a google search for "openstack ceilometer" | 23:12 |
| torgomatic | 2. Exactly-once delivery | 23:13 |
| torgomatic | 1. Guaranteed order of messages | 23:13 |
| torgomatic | 2. Exactly-once delivery | 23:13 |
| torgomatic | (credit to https://twitter.com/mathiasverraes/status/632260618599403520?lang=en) | 23:13 |
| notmyname | torgomatic: yeah, I'm not a fan of it at all ;-) | 23:13 |
| kong | notmyname: hah, good suggestion | 23:13 |
| notmyname | kong: that's one reason why I (strongly) think using logs for billing processing is much better | 23:14 |
| kong | notmyname: ok, got you, thanks! | 23:14 |
| notmyname | kong: the other reason is because you can reply logs to validate calculations. you can't reply the message queue | 23:14 |
| kong | notmyname: yes, good point. Actually, we already have a script to do that by reading bytes-in/bytes-out, but I just wonder if there is a better way to do | 23:15 |
| kong | and finally I got answer here | 23:16 |
| notmyname | :-) | 23:16 |
| kong | thanks again | 23:16 |
| clayg | kota_: still trying to grok | 23:16 |
| clayg | kota_: does my output looks like your output? | 23:17 |
| clayg | ... i added a comment in the gist | 23:17 |
| kota_ | clayg: my output is the first one shows the established connection still existing but the second one in green pool doesn't | 23:18 |
| kota_ | clayg: oh, yes, the gist you pasted is same with me | 23:19 |
| clayg | ok, so step #1 - there is a different and it's reproduicible - awesome | 23:19 |
| clayg | step #2 - why is it different, what does it mean, is it bad? | 23:19 |
| kota_ | so, the difference is the place running, the first one is in just python, the second one is in GreenPool. | 23:20 |
| kota_ | but I'm not sure GreePool can collect the leak socket (or not)? | 23:20 |
| kota_ | that's my question. | 23:21 |
| kota_ | is it expected or not? | 23:21 |
| kota_ | ah, ok. explaining step by step. | 23:22 |
| *** ManojK has joined #openstack-swift | 23:22 | |
| kota_ | so we have a known issue if a instance of BufferedConnection is missing the reference, it could cause socket leak for the connection, right? | 23:23 |
| kota_ | (if we don't call .close() before removeing reference) | 23:23 |
| kota_ | correctly speaking, it might be HTTPConnection issue though. | 23:25 |
| clayg | heh, we have a "SteamingPile" ... errr there might be an "r" in there somewhere | 23:25 |
| *** Suyash has quit IRC | 23:27 | |
| kota_ | and i suppose it could happen even if it's in GreenPool because GreenPool supports network I/O concurrency, switching and I don't expect it supports something like gc on networking socket. | 23:27 |
| kota_ | but... it can collect the garbage socket so I'm confused. | 23:27 |
| *** ManojK has quit IRC | 23:30 | |
| *** ManojK has joined #openstack-swift | 23:33 | |
| openstackgerrit | Merged openstack/swift: Fixed inconsistent naming conventions https://review.openstack.org/339242 | 23:34 |
| *** nadeem has quit IRC | 23:35 | |
| *** chsc has quit IRC | 23:36 | |
| clayg | kota_: maybe it's a green vs. real socket monkey-patch kinda thing? | 23:37 |
| clayg | https://gist.github.com/clayg/c0e04175d120ffb33e531d02b6656655 | 23:37 |
| clayg | kota_: ^ in this case both sock get closed... but that's dropping the reference to the socket directly instead of by dropping the connection - so maybe it's not apples to apples | 23:38 |
| clayg | well idk, if I replace the first test with a green_socket it shows the leak behavior like the orig | 23:39 |
| clayg | where the netstat showed the established connection outside of the function call? | 23:39 |
| kota_ | wait a bit, trying to the new one | 23:40 |
| clayg | i'm sure buffered_http uses green httplib explicitly - so maybe it's just a different in how python works with the non-blocking socket if you don't poll it somehow like the eventlet hub does? | 23:40 |
| clayg | kota_: I expect in the next version of the script the port param will make it all the way out to int(sys.argv[1]) - before long we'll be importing argparse :\ | 23:40 |
| kota_ | yeah, both sock get closed... | 23:42 |
| *** manous has quit IRC | 23:43 | |
| kota_ | if we set a green socket to the first one, it seems still leaked, i can see FIN_WAIT | 23:45 |
| *** siva_krish has joined #openstack-swift | 23:45 | |
| *** diogogmt has quit IRC | 23:46 | |
| kota_ | hmm.... interesting. | 23:48 |
| *** siva_krish has quit IRC | 23:50 | |
| notmyname | wow. lots of patches have landed today | 23:51 |
| kota_ | clayg: the eventlet hubs to read is around here, https://github.com/eventlet/eventlet/tree/master/eventlet/hubs, right? | 23:52 |
| timburke | notmyname: yup. crypto's done! what else can we do ahead of the hackathon? | 23:52 |
| notmyname | timburke: merge all the patches! | 23:52 |
| timburke | damn straight! | 23:52 |
| *** diogogmt has joined #openstack-swift | 23:54 | |
| kota_ | ah, looks like there is sort of FdListener there, so it could close if the thread ends, maybe??? | 23:55 |
| kota_ | look at more deelply... | 23:55 |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!