| *** ho has joined #openstack-swift | 00:01 | |
| *** annegentle has joined #openstack-swift | 00:24 | |
| *** zhill has quit IRC | 00:35 | |
| *** evanjfraser has quit IRC | 00:36 | |
| *** evanjfraser has joined #openstack-swift | 00:36 | |
| *** annegentle has quit IRC | 00:42 | |
| openstackgerrit | Samuel Merritt proposed openstack/swift: EC: support multiple ranges for GET requests https://review.openstack.org/166576 | 00:44 |
|---|---|---|
| torgomatic | *phew* | 00:44 |
| torgomatic | mattoliverau: so there were a couple more problems in there, but I think I flattened them | 00:44 |
| *** ujjain has quit IRC | 00:48 | |
| *** vinsh has joined #openstack-swift | 00:48 | |
| openstackgerrit | Samuel Merritt proposed openstack/swift: EC: support multiple ranges for GET requests https://review.openstack.org/166576 | 00:49 |
| torgomatic | ...and I missed a test | 00:49 |
| *** kota_ has joined #openstack-swift | 00:50 | |
| *** panbalag has joined #openstack-swift | 00:51 | |
| ho | good morning guys! | 00:53 |
| *** vinsh has quit IRC | 00:55 | |
| *** anderstj has quit IRC | 00:59 | |
| *** alpha_ori has quit IRC | 01:00 | |
| *** mariusv has quit IRC | 01:00 | |
| *** amandap has quit IRC | 01:01 | |
| *** mordred has quit IRC | 01:01 | |
| *** tellesnobrega has quit IRC | 01:01 | |
| *** bus-104_ has joined #openstack-swift | 01:01 | |
| *** mariusv has joined #openstack-swift | 01:02 | |
| *** mariusv has quit IRC | 01:02 | |
| *** mariusv has joined #openstack-swift | 01:02 | |
| *** bus-104 has quit IRC | 01:05 | |
| *** mordred has joined #openstack-swift | 01:06 | |
| *** amandap has joined #openstack-swift | 01:07 | |
| *** anderstj has joined #openstack-swift | 01:07 | |
| *** alpha_ori has joined #openstack-swift | 01:08 | |
| clayg | torgomatic: nice! | 01:08 |
| clayg | ho: good morning | 01:08 |
| ho | clayg: hello | 01:09 |
| *** panbalag has quit IRC | 01:12 | |
| *** tellesnobrega has joined #openstack-swift | 01:15 | |
| *** haomaiwang has quit IRC | 01:41 | |
| *** annegentle has joined #openstack-swift | 01:43 | |
| *** annegentle has quit IRC | 01:48 | |
| *** annegentle has joined #openstack-swift | 01:50 | |
| *** haomaiwang has joined #openstack-swift | 01:55 | |
| *** annegentle has quit IRC | 02:00 | |
| *** vinsh has joined #openstack-swift | 02:06 | |
| *** panbalag has joined #openstack-swift | 02:07 | |
| *** panbalag has left #openstack-swift | 02:08 | |
| *** panbalag has joined #openstack-swift | 02:08 | |
| *** vinsh has quit IRC | 02:08 | |
| *** panbalag has quit IRC | 02:17 | |
| kota_ | clayg: Hi | 02:29 |
| kota_ | clayg: still here? | 02:29 |
| ho | torgomtic: hello, I have one question about the patch 166576. In the commit message there is a reference to RFC7232 (appendix a). That means I need to check RFC2616 with changes in RFC7232. right? | 02:33 |
| patchbot | ho: https://review.openstack.org/#/c/166576/ | 02:33 |
| kota_ | ho: bay-area seems closing :( | 02:36 |
| ho | kota_: yeah :-) around 18:30 there? | 02:38 |
| kota_ | ho: wait, opening my bay-area clock... | 02:38 |
| kota_ | ho: 19:30-sh because of daylight saving. | 02:39 |
| ho | kota_: i forgot it... | 02:40 |
| kota_ | ho: no worries, we don't have the rule in Japan :P | 02:41 |
| ho | kota_: haha. good point is we have more over-lap with eu guys :) | 02:43 |
| kota_ | ho: exactly :) | 02:43 |
| *** jrichli has joined #openstack-swift | 03:02 | |
| clayg | i'm back | 03:08 |
| ho | clayg: welcome back! but taking care of your health :) | 03:09 |
| clayg | my colleagues and friends around the world always seemed more worried about my odd hours that my office mates - if you hang out with me I think you just get to used to it ;) | 03:10 |
| clayg | I'm fine. Weekend is coming up. | 03:10 |
| clayg | kota_: i wanted to help you to validate the issue i'm seeing with jerasure and pyeclib < 1.0.7 - https://gist.github.com/clayg/959a5d77353a91a4b10a | 03:11 |
| clayg | kota_: it seemed like maybe it was thing - new pyeclib with fix is supposed to be released tomorrow - but too late to get into global requirements for kilo we think if the hack will hold us over into beta | 03:11 |
| ho | clayg: yeah. I can not wait weekend. | 03:14 |
| kota_ | clayg: I'm back too from lunch | 03:19 |
| kota_ | clayg: during scroll backing the log, maybe you know what I ask you and now I'm looking at the gist code... | 03:20 |
| kota_ | clayg: curious, exactly Kevin's commit fixes the issue so I confirmed the test_reconstructor passed with newest liberasurecode master and dirct using of reconstruct. | 03:33 |
| *** links has joined #openstack-swift | 03:45 | |
| kota_ | clayg: oh, I finally understood that bug happens only when the reconstruct target index is in parities. | 03:47 |
| kota_ | clayg: Once I wrote a test for non-ordered case here, https://bitbucket.org/bloodeagle40234/pyeclib/commits/dde4e182f702de9ce1a2ec454d15c42ab1cebf4c | 03:48 |
| kota_ | clayg: the bug never appeared because the test dropped a data fragment. | 03:48 |
| *** chlong has joined #openstack-swift | 03:48 | |
| *** chlong has quit IRC | 03:49 | |
| kota_ | clayg: ok, that's fine that the bug have been fixed on newest liberasurecode and I know the global requirement has already been freezed. | 03:49 |
| kota_ | clayg: Thanks for the information that is really helpful for me to understand everything! | 03:50 |
| kota_ | clayg: wow, found your comment, right now on the gerrit. much appreciated! | 03:56 |
| clayg | kota_: huh - that test in bitbucket looks reasonable to me? YOU wrote that!? | 03:57 |
| kota_ | clayg: yeah, that's my repository. | 03:58 |
| clayg | kota_: I don't get why that test didn't hit the case I saw - was it because your list had 13 of the 10+4 instead of only 10? | 03:58 |
| kota_ | clayg: In my understand, the bug hits when reconstructing the parity fragments, i.e. at the my case, index 0 must be in data fragments. | 04:00 |
| kota_ | clayg: so i still just am curious on your comment that you specified the reconstruction of index 2 (maybe in data fragments) also goes to fail. | 04:01 |
| kota_ | at the case 0, 1, 3, 4, 5, 6, [7] | 04:01 |
| clayg | holy crap patch 166576 and all that just worked on the replicated case already!? | 04:02 |
| patchbot | clayg: https://review.openstack.org/#/c/166576/ | 04:02 |
| clayg | kota_: can you help me fix the comment then? I think I acctually don't understand it. | 04:02 |
| clayg | when I wrote that gist I felt like I knew what I was talking about - now I don't think I know what I'm talking about - I think you know what you're talking about :) | 04:03 |
| kota_ | clayg: ok, I'll ack the comment on the gerrit :) | 04:04 |
| clayg | kota_: - totally works | 04:05 |
| clayg | it was the parity bit - since your test only hit 0 it worked (right?) | 04:05 |
| clayg | no wait... that test should have eventually has the shuffled frags[0] be a parity frag - wth? | 04:05 |
| *** jrichli has quit IRC | 04:07 | |
| kota_ | clayg: that test shuffled whole frags and use frags[1:], i.e. drop a fragment randomly eventhough the target might be in the available fragment. | 04:08 |
| kota_ | clayg: ya, it's worse for testing purpose...my bad. However available 1,2,3,4....13 and reconstruct 0 worked well. | 04:09 |
| clayg | wow so it's like if you want to rebuild a parity - but you don't have all the other data *and* parity (maybe) | 04:11 |
| kota_ | clayg: and then | 04:11 |
| kota_ | clayg: I pushed the failure case at https://bitbucket.org/bloodeagle40234/pyeclib/commits/26bc79ee1e1a0a1f21e789b9c1e3e9ce279fd889?at=master | 04:11 |
| clayg | what *really* - THAT fails and the other worked? | 04:12 |
| clayg | I thought for sure my test case was showing something to do with only haveing a subset of all the fragments | 04:12 |
| clayg | like I thought I verfied that if you have all segments except for the one you're trying to rebuild - it doesn't matter the order or if the missing fragment is data or parity - it works | 04:13 |
| kota_ | clayg: not yet tested whole thing. but to hit the issue, we should drop and try to reconstruct *parity* fragment. | 04:13 |
| kota_ | clayg: exactly, the order is not matter, I guess. | 04:14 |
| kota_ | clayg: no wait | 04:14 |
| clayg | oh i see - ok... maybe hold :P | 04:14 |
| kota_ | clayg: the oder is *one thing* to cause it. | 04:15 |
| clayg | in a 6+2 test 0, 1, 2, 3 works to rebuild 5 while 0, 1, 2, 4 does not | 04:16 |
| clayg | I think the failure mode (maybe there's others?) has specificially to do with rebuilding parity with only k fragments (and the k fragments aren't all the data fragments) | 04:17 |
| clayg | at least that's what I'm sticking with as being a thing that doesn't work - and that was why I saw :P | 04:17 |
| clayg | you should test if in a 10+4 0-8, 10, 11, 12 can rebuild 13 or not - but maybe more interesting if you can do it with just 0-8, 10 | 04:18 |
| clayg | ... or something | 04:19 |
| kota_ | clayg: ya, 10+4 0-8, 10, 11, 12 can rebuild 13 on the first sanity of my failure code, that was passed. | 04:20 |
| kota_ | clayg: seems a lack of data fragment trigger the failure... | 04:22 |
| kota_ | clayg: wait a while, i'll check it my assumption. | 04:23 |
| *** zaitcev has quit IRC | 04:23 | |
| kota_ | clayg: Add a test here https://bitbucket.org/bloodeagle40234/pyeclib/commits/2c028d16baf2ea4d4f475d41befb5e40908dfbf8?at=master | 04:34 |
| kota_ | clayg: according the test result, when the *data* fragment dropped *and* reconstruct to a *parity*, it goes to fail. | 04:36 |
| *** briancurtin has quit IRC | 04:46 | |
| *** fanyaohong has quit IRC | 04:46 | |
| *** goodes has quit IRC | 04:46 | |
| *** zhiyan has quit IRC | 04:46 | |
| *** jdaggett has quit IRC | 04:47 | |
| *** nottrobin has quit IRC | 04:47 | |
| *** briancurtin has joined #openstack-swift | 04:47 | |
| *** fanyaohong has joined #openstack-swift | 04:47 | |
| *** goodes has joined #openstack-swift | 04:47 | |
| *** cppforlife_ has quit IRC | 04:47 | |
| *** zhiyan has joined #openstack-swift | 04:47 | |
| *** nottrobin has joined #openstack-swift | 04:49 | |
| *** cppforlife_ has joined #openstack-swift | 04:49 | |
| *** jdaggett has joined #openstack-swift | 04:51 | |
| *** SkyRocknRoll has joined #openstack-swift | 05:13 | |
| *** thumpba has joined #openstack-swift | 05:14 | |
| kota_ | clayg: I pushed a current conclusion to the gerrit | 05:19 |
| kota_ | clayg: anways, that not so matter becase we can wait bump the global requirements after kilo-release :P | 05:22 |
| *** thumpba_ has joined #openstack-swift | 05:24 | |
| *** thumpba has quit IRC | 05:27 | |
| *** SkyRocknRoll has quit IRC | 05:27 | |
| *** dmorita has quit IRC | 05:43 | |
| *** kota_ has quit IRC | 06:02 | |
| *** kota_ has joined #openstack-swift | 06:02 | |
| *** krykowski has joined #openstack-swift | 06:04 | |
| *** ktsuyuzaki has joined #openstack-swift | 06:05 | |
| *** thumpba_ has quit IRC | 06:07 | |
| *** kota_ has quit IRC | 06:07 | |
| *** kota_ has joined #openstack-swift | 06:07 | |
| *** ktsuyuzaki has quit IRC | 06:09 | |
| charz | clayg: I think the problem is the py26 unittest sometime was stuck and can't finish until I chancel it. | 06:30 |
| clayg | charz: oh cool - can you use the scripts! | 06:33 |
| charz | clayg: I think I need to modify script on jenkins to avoid to send a empty report. | 06:35 |
| charz | clayg: yes, I'll use that. | 06:35 |
| clayg | heh, it's ok what ever works :P | 06:36 |
| clayg | kota_: you left!? | 06:37 |
| clayg | kota_: anyway - i think you've nailed it - thanks for pushing me! | 06:37 |
| *** kota_ has joined #openstack-swift | 06:44 | |
| charz | clayg: Oh, that's different problem, jenkins can't fetch the code from review.openstack.org at that moment. I add retry count on the task. | 06:47 |
| *** ktsuyuzaki has joined #openstack-swift | 06:47 | |
| *** kota_ has quit IRC | 06:48 | |
| * charz disable py26unit temporary | 07:04 | |
| clayg | take that py26! | 07:19 |
| *** jistr has joined #openstack-swift | 07:29 | |
| *** ktsuyuzaki has quit IRC | 08:07 | |
| *** Bsony has joined #openstack-swift | 08:12 | |
| *** jordanP has joined #openstack-swift | 08:20 | |
| *** acoles_away is now known as acoles | 08:27 | |
| acoles | morning | 08:29 |
| ho | acoles: morning! | 08:34 |
| acoles | clayg: the 'fragments' list is used by ECDiskFile purge() method | 08:39 |
| acoles | clayg: line 1898 https://review.openstack.org/#/c/170339/1/swift/obj/diskfile.py | 08:41 |
| acoles | ho: hello - have a good weekend, must be soon | 08:45 |
| ho | acoles: thanks! | 08:46 |
| acoles | huh, we only have 8 reviews in the ec-review chain, we had like 25 for policies | 08:48 |
| clayg | acoles: i found it - failing test acctually found it for me which is a good sign - thanks! | 08:54 |
| acoles | clayg: all of your comments on ssync_sender i had in my head but never made it to my keyboard in time;) | 09:00 |
| *** bkopilov has quit IRC | 09:01 | |
| acoles | as in coding it that way | 09:01 |
| acoles | now i have to learn a whole new set of review id's for git review -d ! | 09:02 |
| clayg | could be worse | 09:06 |
| *** krykowski has quit IRC | 09:25 | |
| *** haomaiwang has quit IRC | 10:43 | |
| *** nexusz99 has joined #openstack-swift | 10:51 | |
| ho | have nice weekend! | 10:59 |
| *** ho has quit IRC | 10:59 | |
| *** aix has joined #openstack-swift | 11:20 | |
| *** km has quit IRC | 11:32 | |
| *** silor has joined #openstack-swift | 11:47 | |
| *** haomaiwang has joined #openstack-swift | 11:59 | |
| *** jistr_ has joined #openstack-swift | 12:27 | |
| *** jistr has quit IRC | 12:30 | |
| *** jistr has joined #openstack-swift | 12:31 | |
| *** jistr_ has quit IRC | 12:32 | |
| *** silor has quit IRC | 12:33 | |
| *** kei_yama has quit IRC | 12:57 | |
| *** nexusz99 has quit IRC | 13:05 | |
| *** annegentle has joined #openstack-swift | 13:18 | |
| *** openstackgerrit has quit IRC | 13:36 | |
| *** openstackgerrit has joined #openstack-swift | 13:36 | |
| *** vinsh has joined #openstack-swift | 13:37 | |
| openstackgerrit | Merged openstack/swift-specs: Updates to the reconstructor section, quick scrub of everything https://review.openstack.org/142146 | 13:38 |
| *** vinsh has quit IRC | 13:43 | |
| *** jrichli has joined #openstack-swift | 14:02 | |
| *** zul has quit IRC | 14:13 | |
| *** zul has joined #openstack-swift | 14:14 | |
| *** annegentle has quit IRC | 14:25 | |
| *** jordanP has quit IRC | 14:26 | |
| *** vinsh has joined #openstack-swift | 14:33 | |
| *** links has quit IRC | 14:33 | |
| *** silor has joined #openstack-swift | 14:34 | |
| *** openstackgerrit has quit IRC | 14:39 | |
| *** openstackgerrit has joined #openstack-swift | 14:39 | |
| *** lpabon has joined #openstack-swift | 14:42 | |
| *** dmsimard has quit IRC | 14:45 | |
| *** dmsimard has joined #openstack-swift | 14:46 | |
| *** annegentle has joined #openstack-swift | 14:48 | |
| *** wer has joined #openstack-swift | 14:58 | |
| *** EmilienM is now known as EmilienM|afk | 15:07 | |
| *** annegentle has quit IRC | 15:17 | |
| *** annegentle has joined #openstack-swift | 15:17 | |
| *** links has joined #openstack-swift | 15:26 | |
| *** hugespoon has joined #openstack-swift | 15:29 | |
| *** EmilienM|afk is now known as EmilienM | 15:39 | |
| *** jistr has quit IRC | 15:53 | |
| *** jistr has joined #openstack-swift | 15:53 | |
| clayg | quiet in here | 15:54 |
| hugespoon | all, im running a SAIO setup in a container (xfs loopback FS mounted into the container) and everything seems to be working great, but when I run the functional tests, I get some failures on bad metadata testing… and thus far i have no idea why…this is 2.2.2 btw | 15:55 |
| clayg | acoles: I like that diff a lot - thanks all use that as a starting point for the day - you rock! | 15:58 |
| * peluse typing typing typing on doc | 15:59 | |
| acoles | clayg: hey, good morning. | 15:59 |
| acoles | peluse: you too ^^ | 15:59 |
| clayg | acoles: nice and review on the ec-sp change too - thanks! I should be busy ;) | 16:00 |
| acoles | clayg: totally agree about not duplicating the delete_list. So do you think its ok to change the return type of ssync_sender to include the timestamps | 16:00 |
| acoles | ? | 16:00 |
| acoles | like, we can assume no one 'out there' might be using that interface other than our replicator | 16:01 |
| acoles | i want the return to be a list of (hash, timestamp) tuples instead of just the hashes | 16:01 |
| *** kevinc has joined #openstack-swift | 16:03 | |
| *** annegentle has quit IRC | 16:07 | |
| *** jistr has quit IRC | 16:17 | |
| clayg | acoles: yeah that'd be fine - it would acctually be better for the replicator too - since it's using the list for cleanup same as the reconstructor - but instead of doing the right thing with a purge it's rmtree'in the hashdir - which is wrong! | 16:19 |
| clayg | so like everything is just getting better all of the time | 16:19 |
| clayg | acoles: at some point you're going to have to review the proxy stuff tho :P | 16:19 |
| acoles | clayg: righ on, the replicator can't be checking that no newer obj got between sync'ing and rmtree'ing :/ | 16:20 |
| clayg | afk, heading into the office + meetings | 16:20 |
| acoles | clayg: i'm on my up the chain man! | 16:20 |
| clayg | right on | 16:20 |
| acoles | like the infrastructor - someone who teaches infra? :D | 16:20 |
| hugespoon | acoles: or the builder of infra perhaps | 16:24 |
| acoles | heh | 16:27 |
| *** kevinc has quit IRC | 16:29 | |
| *** kevinc has joined #openstack-swift | 16:30 | |
| *** wer_ has joined #openstack-swift | 16:31 | |
| *** wer has quit IRC | 16:32 | |
| notmyname | hello, world | 16:36 |
| jrichli | hello! | 16:40 |
| jrichli | having a good Friday? | 16:41 |
| *** kevinc has quit IRC | 16:44 | |
| notmyname | a little bit of a slow start, but I picked up an awesome breakfast close to the office. so that's nice :-) | 16:50 |
| jrichli | sounds good. | 16:52 |
| *** silor has quit IRC | 16:55 | |
| *** annegentle has joined #openstack-swift | 16:58 | |
| openstackgerrit | Takashi Kajinami proposed openstack/swift: Add process name checking into swift-init https://review.openstack.org/116203 | 17:00 |
| *** annegentle has quit IRC | 17:04 | |
| *** annegentle has joined #openstack-swift | 17:04 | |
| *** Trixboxer has quit IRC | 17:10 | |
| *** lcurtis has joined #openstack-swift | 17:11 | |
| *** Guest8530 has quit IRC | 17:24 | |
| *** links has quit IRC | 17:27 | |
| *** david-lyle has quit IRC | 17:29 | |
| *** annegentle has quit IRC | 17:56 | |
| peluse | hey, it's Friday! weeeeeeee | 18:03 |
| clayg | i don't get it | 18:03 |
| *** rmcall has joined #openstack-swift | 18:08 | |
| peluse | special drink prices at 4:00 at most bars, what's not to get? | 18:10 |
| *** aix has quit IRC | 18:13 | |
| *** rmcall has quit IRC | 18:25 | |
| *** annegentle has joined #openstack-swift | 18:35 | |
| *** annegentle has quit IRC | 18:38 | |
| *** aix has joined #openstack-swift | 18:41 | |
| *** annegentle has joined #openstack-swift | 18:43 | |
| *** zhill has joined #openstack-swift | 18:46 | |
| clayg | ok this is going well - acoles fixups continue to be top notch - he has really been the champion of ec in these last couple of weeks - what a HERO man - could not have done it without you my friend! | 19:19 |
| peluse | I second that! went from "I think I can help out" to really being - well you said it! | 19:26 |
| *** mitz has quit IRC | 19:28 | |
| *** aerwin has joined #openstack-swift | 19:34 | |
| *** mitz has joined #openstack-swift | 19:37 | |
| *** annegentle has quit IRC | 19:49 | |
| *** annegentle has joined #openstack-swift | 20:07 | |
| *** lpabon has quit IRC | 20:07 | |
| *** annegentle has quit IRC | 20:13 | |
| acoles | peluse: clayg: cheers guys. its been a blast. | 20:18 |
| peluse | hoyy crap the sync_diskfile_builder callback is way cool! | 20:18 |
| *** annegentle has joined #openstack-swift | 20:19 | |
| acoles | peluse: there are only two things i like about python... lambdas, and that functions are first class objects, and there i get to do both in one line | 20:19 |
| peluse | yeah, neat stuff! | 20:21 |
| peluse | FYI completed docs are available with patch 169990 | 20:21 |
| patchbot | peluse: https://review.openstack.org/#/c/169990/ | 20:21 |
| peluse | (well, completed typing) | 20:21 |
| *** annegentle has quit IRC | 20:25 | |
| acoles | clayg i'll try to review at least patch 169986 tomorrow. ttfn | 20:38 |
| patchbot | acoles: https://review.openstack.org/#/c/169986/ | 20:38 |
| *** acoles is now known as acoles_away | 20:41 | |
| *** aix has quit IRC | 20:47 | |
| *** jrichli has quit IRC | 21:01 | |
| *** annegentle has joined #openstack-swift | 21:05 | |
| openstackgerrit | Samuel Merritt proposed openstack/swift: EC: support multiple ranges for GET requests https://review.openstack.org/166576 | 21:14 |
| *** lcurtis has quit IRC | 21:32 | |
| *** aerwin has quit IRC | 22:07 | |
| clayg | interesting, i just saw a probetest failure with the reaper on ec policy - and interestingly looking at the test and what I know of the object.replicas > container.replicas bug in the reaper - I don't see why I've not run into this failure before? | 22:10 |
| *** david-lyle has joined #openstack-swift | 22:12 | |
| *** Bsony has quit IRC | 22:13 | |
| clayg | trival fix - https://gist.github.com/clayg/ae6007b4a634a3872215 - not sure why the test wasn't complaining earlier... | 22:17 |
| *** annegentle has quit IRC | 22:28 | |
| *** annegentle has joined #openstack-swift | 22:28 | |
| *** annegentle has quit IRC | 22:43 | |
| notmyname | clayg: you going to include that in tonight's revision? | 23:05 |
| *** Viswanath has joined #openstack-swift | 23:31 | |
| *** Viswanath has quit IRC | 23:40 | |
| *** annegentle has joined #openstack-swift | 23:43 | |
| *** annegentle has quit IRC | 23:49 | |
| *** vinsh has quit IRC | 23:53 | |
| clayg | notmyname: yup - just finished the unitests | 23:58 |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!