| *** Jeffrey4l has quit IRC | 00:03 | |
| clayg | timburke: can you say all of that on the review "for other reviewers here's what you can do ..." | 00:03 |
|---|---|---|
| *** Jeffrey4l has joined #openstack-swift | 00:03 | |
| *** vint_bra has joined #openstack-swift | 00:08 | |
| *** asettle has joined #openstack-swift | 00:12 | |
| *** Worry has joined #openstack-swift | 00:13 | |
| *** asettle has quit IRC | 00:17 | |
| *** Worry has quit IRC | 00:27 | |
| *** krypto has quit IRC | 00:30 | |
| *** chsc has quit IRC | 00:33 | |
| clayg | tdasilva: you have the 1.4.0 release stuff right there! | 00:43 |
| *** jamielennox is now known as jamielennox|away | 00:45 | |
| clayg | ok, let's see what I can do with patch 408257 and patch 408280 | 00:45 |
| patchbot | https://review.openstack.org/#/c/408257/ - liberasurecode - 1.4.0 release | 00:45 |
| patchbot | https://review.openstack.org/#/c/408280/ - pyeclib - release 1.4.0 | 00:45 |
| *** tqtran has quit IRC | 00:54 | |
| *** jistr|mtgs has quit IRC | 01:00 | |
| *** jistr has joined #openstack-swift | 01:00 | |
| *** tovin07 has quit IRC | 01:08 | |
| tdasilva | clayg: hi, sorry i'm back now | 01:10 |
| tdasilva | clayg: yeah we need to merge those two patches | 01:10 |
| *** asettle has joined #openstack-swift | 01:13 | |
| *** Jeffrey4l has quit IRC | 01:18 | |
| *** asettle has quit IRC | 01:18 | |
| *** Jeffrey4l has joined #openstack-swift | 01:19 | |
| *** _JZ_ has quit IRC | 01:30 | |
| *** mingyu has joined #openstack-swift | 01:38 | |
| *** diogogmt has quit IRC | 01:49 | |
| *** jamielennox|away is now known as jamielennox | 01:50 | |
| *** vint_bra has quit IRC | 02:05 | |
| *** klrmn has quit IRC | 02:12 | |
| *** asettle has joined #openstack-swift | 02:14 | |
| *** asettle has quit IRC | 02:19 | |
| *** Jeffrey4l has quit IRC | 02:26 | |
| *** Jeffrey4l has joined #openstack-swift | 02:26 | |
| *** Jeffrey4l has quit IRC | 02:31 | |
| *** Jeffrey4l has joined #openstack-swift | 02:31 | |
| kota_ | hello | 02:36 |
| kota_ | a bit late morning due to no meeting | 02:36 |
| kota_ | it looks like a lot of ec conversation occurred while i was asleep | 02:38 |
| kota_ | let's scroll back to check.... | 02:38 |
| *** bkopilov has quit IRC | 02:43 | |
| *** cdelatte has quit IRC | 02:43 | |
| *** Jeffrey4l has quit IRC | 02:46 | |
| *** Jeffrey4l has joined #openstack-swift | 02:47 | |
| *** clu_ has quit IRC | 02:47 | |
| *** diogogmt has joined #openstack-swift | 02:50 | |
| *** dmorita has quit IRC | 03:05 | |
| mattoliverau | kota_: o/ | 03:07 |
| kota_ | mattoliverau: o/ | 03:07 |
| *** siva_krish has joined #openstack-swift | 03:07 | |
| *** saltsa has quit IRC | 03:12 | |
| *** saltsa has joined #openstack-swift | 03:13 | |
| *** asettle has joined #openstack-swift | 03:15 | |
| *** asettle has quit IRC | 03:20 | |
| *** dmorita has joined #openstack-swift | 03:26 | |
| *** dmorita has quit IRC | 03:30 | |
| *** klrmn has joined #openstack-swift | 03:33 | |
| *** Jeffrey4l has quit IRC | 03:38 | |
| *** Jeffrey4l has joined #openstack-swift | 03:38 | |
| *** Jeffrey4l has quit IRC | 03:42 | |
| *** Jeffrey4l has joined #openstack-swift | 03:43 | |
| openstackgerrit | Matthew Oliver proposed openstack/swift: Move documented reclaim_age option to correct location https://review.openstack.org/374419 | 03:50 |
| mattoliverau | ^^ just a rebase | 03:51 |
| *** psachin has joined #openstack-swift | 03:54 | |
| *** Jeffrey4l has quit IRC | 03:56 | |
| *** Jeffrey4l has joined #openstack-swift | 03:56 | |
| *** klrmn has quit IRC | 03:59 | |
| *** mingyu has quit IRC | 03:59 | |
| *** Jeffrey4l has quit IRC | 04:00 | |
| *** Jeffrey4l has joined #openstack-swift | 04:01 | |
| *** manous has quit IRC | 04:03 | |
| *** bkopilov has joined #openstack-swift | 04:03 | |
| *** Jeffrey4l has quit IRC | 04:06 | |
| *** Jeffrey4l has joined #openstack-swift | 04:06 | |
| *** Jeffrey4l has quit IRC | 04:09 | |
| *** Jeffrey4l has joined #openstack-swift | 04:09 | |
| *** asettle has joined #openstack-swift | 04:16 | |
| *** asettle has quit IRC | 04:21 | |
| *** Jeffrey4l has quit IRC | 04:24 | |
| *** Jeffrey4l has joined #openstack-swift | 04:24 | |
| *** Jeffrey4l has quit IRC | 04:25 | |
| *** Jeffrey4l has joined #openstack-swift | 04:27 | |
| mahatic_ | mattoliverau: thanks | 04:36 |
| openstackgerrit | Merged openstack/pyeclib: Add tests for the availability of individual backends https://review.openstack.org/408261 | 04:40 |
| kota_ | tdasilva, clayg, timburke: as well timburke said, I'd love to https://review.openstack.org/#/c/405926/ to get in to the 1.4.0 liberasurecode release. | 04:52 |
| patchbot | patch 405926 - liberasurecode - Add get_version functionality to liberasurecode lib | 04:52 |
| kota_ | tdasilva: clayg, timburke: reading back the log, the idea pyeclib (and future go lang ec lib) will be into liberasurecode as bindings are great | 04:53 |
| kota_ | tdasilva, clayg, timburke: however, we still have to look up pyeclib a little more time until the work finished and the version reference could be problematic for anyone who dosn't know what's going on at pyeclib/liberasurecode | 04:54 |
| kota_ | tdasilva, clayg: timburke's example looks quite to make me sense to get sanity the difference w/ or w/o the change so that it will help you to review the patch. | 04:56 |
| *** ppai has joined #openstack-swift | 05:03 | |
| *** links has joined #openstack-swift | 05:05 | |
| *** arch-nemesis has joined #openstack-swift | 05:14 | |
| *** asettle has joined #openstack-swift | 05:17 | |
| *** arch-nemesis has quit IRC | 05:18 | |
| *** asettle has quit IRC | 05:22 | |
| *** hseipp has joined #openstack-swift | 05:29 | |
| *** arch-nemesis has joined #openstack-swift | 05:39 | |
| *** tovin07 has joined #openstack-swift | 05:59 | |
| *** chsc has joined #openstack-swift | 06:03 | |
| *** chsc has joined #openstack-swift | 06:03 | |
| *** chsc has quit IRC | 06:08 | |
| *** peter-wang has joined #openstack-swift | 06:09 | |
| peter-wang | hi This is peter | 06:10 |
| peter-wang | I met 503 Service Unavailable when upload new file to swift | 06:12 |
| peter-wang | I add new storage devices to swift cluster | 06:12 |
| peter-wang | but seems not succeeded, the cluster capacity did not increase. | 06:13 |
| peter-wang | can anyone help me out? | 06:13 |
| kota_ | peter-wang: I don't think 503 is related to capacity problem because if no enough space is there to be written, probably it will be 507 InsufficientStorage instead of 503 | 06:16 |
| kota_ | peter-wang: so that you may hit another problem. | 06:16 |
| kota_ | peter-wang: did your proxy or object server log say anything? | 06:17 |
| *** asettle has joined #openstack-swift | 06:18 | |
| *** dmorita has joined #openstack-swift | 06:19 | |
| *** ChubYann has quit IRC | 06:20 | |
| *** asettle has quit IRC | 06:23 | |
| *** dmorita has quit IRC | 06:23 | |
| *** mingyu has joined #openstack-swift | 06:28 | |
| *** _david_cole_soho has joined #openstack-swift | 06:37 | |
| *** sams-gleb has joined #openstack-swift | 06:40 | |
| *** sams-gleb has quit IRC | 06:40 | |
| *** sams-gleb has joined #openstack-swift | 06:40 | |
| _david_cole_soho | I have upgraded a medium size cluster to newton and I’m finding that after I upload an object, I may get one etag back and later I get a different etag back. | 06:45 |
| _david_cole_soho | I wonder if one of my backends is running old code, gonna check that first | 06:46 |
| zaitcev | which of those etags is correct though? You can verify by running md5sum | 07:10 |
| _david_cole_soho | When I saw it change, the most recent one was correct | 07:11 |
| *** hseipp has quit IRC | 07:11 | |
| _david_cole_soho | I have a bunch of account/container/object paths from an application that is failing at the moment. Maybe if I use swift-get-nodes on them I will find a correlation | 07:13 |
| zaitcev | So the etag returned by PUT is incorrect? | 07:14 |
| *** mingyu has quit IRC | 07:14 | |
| zaitcev | But the one in HEAD is correct? | 07:14 |
| _david_cole_soho | I am not sure if the PUT etag is correct or not. I was looking at an object with Cyberduck (which most likely just HEAD’s the object) and the etag changed, from one minute to the next | 07:17 |
| _david_cole_soho | That sounds like the object being overwritten I know | 07:17 |
| _david_cole_soho | But the data in question is written by an application that only writes once, using a UUID based object path | 07:18 |
| _david_cole_soho | And this has only started to occur since I upgraded to Netwton over the weekend | 07:18 |
| *** asettle has joined #openstack-swift | 07:19 | |
| *** mingyu has joined #openstack-swift | 07:19 | |
| *** asettle has quit IRC | 07:23 | |
| *** SkyRocknRoll has joined #openstack-swift | 07:25 | |
| *** SkyRocknRoll has quit IRC | 07:25 | |
| *** SkyRocknRoll has joined #openstack-swift | 07:25 | |
| *** rcernin has joined #openstack-swift | 07:26 | |
| *** McMurlock1 has joined #openstack-swift | 07:32 | |
| *** oshritf has joined #openstack-swift | 07:32 | |
| _david_cole_soho | I’ve checked my ACO servers and they all seem to be running the newton package and the processes were all restarted on the 5th, so I don’t think it’s a mismatch. I’m going to try and confirm if PUT is returning a different etag to a later HEAD | 07:37 |
| zaitcev | With Cyberduck, I think it's interesting to (a) get logs and filter proxy logs by your own client IP, the TxID (transaction), then collect transaction logs. This should point out if a particular node reports wrong Etag. | 07:44 |
| zaitcev | And (b) check if etag flips back and forth - although that does not tell you anything really | 07:45 |
| *** mingyu has quit IRC | 07:48 | |
| _david_cole_soho | Using curl I have been able to upload the same object 20 times, the PUT returns the same ( correct!) etag but a GET on the container shows me some of the objects I uploading have a different etag | 07:51 |
| zaitcev | yeah, your updater lags | 07:51 |
| zaitcev | probably you can find a megabyte-long pending/ directory somewhere | 07:52 |
| _david_cole_soho | Thank you for the help, I will look at the updater and see if there is a big pending directory | 07:53 |
| _david_cole_soho | We enabled encryption at rest when we upgraded to newton a few days ago, I hope that is not involved... | 07:53 |
| zaitcev | ugh | 07:54 |
| zaitcev | Well, that's for Janie or whoever knows how that works. I don't use it, so no idea. | 07:54 |
| _david_cole_soho | I may disable it if nothing else is wrong | 07:55 |
| zaitcev | Encryption creates a challenge for etags, because now the object's on-disk etag differs from what the client wants, which is plaintext etag. | 07:55 |
| zaitcev | I heard that some extraordinary tricks are used to shuffle etags, I wonder if it's a bug | 07:56 |
| zaitcev | Auditor can only verify the encrypted etag (obviously...), so that one is attached to object as the Etag parameter, and the data's etag is stuffed somewhere, like sysmeta... I dunno really, sorry. | 07:57 |
| *** hseipp has joined #openstack-swift | 08:01 | |
| _david_cole_soho | I’ve matched one of the incorrect etags in the container GET to a .data file on the object server, which of course which by encrypted. That’s concerning, the encrpted file etag is presented to the user, which I don’t think it ever should be | 08:09 |
| *** dmorita has joined #openstack-swift | 08:11 | |
| *** dmorita has quit IRC | 08:15 | |
| *** asettle has joined #openstack-swift | 08:20 | |
| *** mingyu has joined #openstack-swift | 08:20 | |
| *** asettle has quit IRC | 08:24 | |
| _david_cole_soho | A pastebin showing what I am seeing regarding encrypted object etags sporatciatlly appearing in container listings: http://paste.openstack.org/show/591755/ | 08:25 |
| *** jamielennox is now known as jamielennox|away | 08:33 | |
| *** rledisez has joined #openstack-swift | 08:34 | |
| *** mingyu has quit IRC | 08:40 | |
| *** mingyu has joined #openstack-swift | 08:43 | |
| *** jordanP has joined #openstack-swift | 08:43 | |
| openstackgerrit | Cao Xuan Hoang proposed openstack/swift: Use assertGreater(len(x), 0) instead of assertTrue(len(x) > 0) https://review.openstack.org/408488 | 08:47 |
| zaitcev | _david_cole_soho: file a bug in the launchpad, please. Sorry, this may not have an immediate resolution | 08:50 |
| _david_cole_soho | will do, thanks for your help | 08:51 |
| zaitcev | _david_cole_soho: if nobody picks in a day or two, try to poke acoles_ or jamielennox|away | 08:51 |
| zaitcev | probably what happens here, if the object server succeeds in update, it updates correctly, but if the update goes into pending, then the updater screws it up | 08:53 |
| *** cbartz has joined #openstack-swift | 08:53 | |
| *** cbartz has quit IRC | 08:53 | |
| _david_cole_soho | it’s interesting that over time, the first set of objects that have incorrect etags in the listing are corrected, but others that were fine go bad | 08:55 |
| zaitcev | althugh, I dunno... your pastebin says the md5 of the encrypted object is not 2ca803d66d53d85c3b7c6ded5cdcd2e8, but d2cffe8cadfd3e7f7cf6d998285da18f | 08:55 |
| zaitcev | anyway, I need to drop | 08:55 |
| zaitcev | Let Jamie & Co. to figure it out | 08:56 |
| *** pcaruana has joined #openstack-swift | 08:58 | |
| *** peter-wang has quit IRC | 09:15 | |
| *** peter-wang has joined #openstack-swift | 09:16 | |
| mahatic_ | _david_cole_soho: hello, did you happen to verify this part you've mentioned? "<_david_cole_soho> I wonder if one of my backends is running old code, gonna check that first" | 09:19 |
| mahatic_ | It could be the case that one of the backends is not upgraded and the data is not being decrypted - probably the reason for you to be seeing encrypted etag values | 09:20 |
| *** asettle has joined #openstack-swift | 09:24 | |
| _david_cole_soho | mahatic_: I confirmed that the Ubuntu packages were all correct on the back end - all running Newton. I believe all were restarted after the package install, but I suppose it might be possible that one was not. | 09:26 |
| openstackgerrit | Sachin Patil proposed openstack/swift: Fix swift-get-nodes arg parsing for missing ring https://review.openstack.org/406012 | 09:27 |
| *** mvk has quit IRC | 09:39 | |
| *** mingyu has quit IRC | 09:42 | |
| *** hseipp has quit IRC | 09:43 | |
| mahatic_ | _david_cole_soho: ok, could you file a bug after confirming that and with more details on your cluster config? | 09:44 |
| mahatic_ | I'm gonna try and reproduce it | 09:45 |
| _david_cole_soho | mahatic_: sure, will do | 09:45 |
| *** hogepodge has quit IRC | 09:45 | |
| *** hogepodge has joined #openstack-swift | 09:46 | |
| mahatic_ | also you could make sure that your conf adheres to this: http://docs.openstack.org/developer/swift/overview_encryption.html#deployment-and-operation | 09:49 |
| *** sams-gleb has quit IRC | 09:57 | |
| *** sams-gleb has joined #openstack-swift | 09:58 | |
| *** sams-gleb has quit IRC | 10:02 | |
| *** mvk has joined #openstack-swift | 10:11 | |
| mahatic_ | _david_cole_soho: FWIW, I tried to reproduce, my container listing shows the correct etags for all the objects (I tried 20 objects with similar data uploaded with diff names) | 10:16 |
| openstackgerrit | Christopher Bartz proposed openstack/swift: tempurls with a prefix-based scope https://review.openstack.org/274048 | 10:24 |
| *** peter-wang has quit IRC | 10:24 | |
| *** si1ver has quit IRC | 10:24 | |
| *** cbartz has joined #openstack-swift | 10:25 | |
| *** _david_cole_soho has quit IRC | 10:33 | |
| *** links has quit IRC | 11:00 | |
| *** hogepodge has quit IRC | 11:01 | |
| *** acoles_ is now known as acoles | 11:07 | |
| *** links has joined #openstack-swift | 11:16 | |
| *** Jeffrey4l has quit IRC | 11:20 | |
| *** hogepodge has joined #openstack-swift | 11:21 | |
| *** hogepodge has quit IRC | 11:26 | |
| acoles | is david cole still here? | 11:29 |
| acoles | joel? | 11:30 |
| *** jarbod has joined #openstack-swift | 11:32 | |
| acoles | fwiw the mixed etag issue described could be due to some object servers not being upgraded. Obviously I can't say for sure but it would explain the symptoms, | 11:35 |
| acoles | mahatic_: zaitcev ^^ | 11:35 |
| mahatic_ | acoles: agree, it looks like it | 11:37 |
| *** hogepodge has joined #openstack-swift | 11:38 | |
| *** psachin has quit IRC | 11:42 | |
| *** Jeffrey4l has joined #openstack-swift | 11:43 | |
| *** hogepodge has quit IRC | 11:48 | |
| *** vint_bra has joined #openstack-swift | 11:54 | |
| *** Wzze has joined #openstack-swift | 12:00 | |
| *** hogepodge has joined #openstack-swift | 12:00 | |
| *** psachin has joined #openstack-swift | 12:01 | |
| *** joeljwright has joined #openstack-swift | 12:08 | |
| *** ChanServ sets mode: +v joeljwright | 12:08 | |
| *** psachin has quit IRC | 12:08 | |
| *** psachin has joined #openstack-swift | 12:10 | |
| joeljwright | acoles: you about? we're having some 'interesting' experiences with encryption | 12:10 |
| joeljwright | acoles: the etag returned for an uploaded object seems to be inconsistent between requests | 12:11 |
| joeljwright | acoles: same proxy each time | 12:12 |
| joeljwright | acoles: just wondering if you had any thoughts? | 12:12 |
| *** bkopilov has quit IRC | 12:15 | |
| *** Wzze has quit IRC | 12:16 | |
| *** McMurlock1 has quit IRC | 12:21 | |
| *** jordanP has quit IRC | 12:23 | |
| acoles | joeljwright: hi. | 12:24 |
| acoles | joeljwright: I read what david described in scrollback | 12:25 |
| acoles | joeljwright: It would be consistent with not all object servers being upgraded when objects are PUT | 12:25 |
| acoles | joeljwright: if that were the case then some container updates would have the etag of the ecnrypted object content, which is what david sees in listing some timez | 12:26 |
| acoles | joeljwright: but some ocntainer updates (from upgraded object servers) would get the encrypted etag of the original object. | 12:27 |
| joeljwright | acoles: we believe that everything is upgraded at the moment and we're seeing this with new uploads | 12:27 |
| acoles | joeljwright: oh. then I don't have a ready explanation :/ | 12:28 |
| joeljwright | acoles: I guess it's time to go checking the object servers :) | 12:28 |
| joeljwright | thanks for the suggestions | 12:28 |
| *** McMurlock1 has joined #openstack-swift | 12:37 | |
| acoles | joeljwright: let us know | 12:39 |
| acoles | tdasilva: did you have a patch for swift-object-info to show transient sysmeta? or did I dream that? | 12:39 |
| openstackgerrit | Christopher Bartz proposed openstack/python-swiftclient: prefix-based tempurls support https://review.openstack.org/408596 | 12:45 |
| *** vinsh has quit IRC | 12:48 | |
| *** psachin has quit IRC | 12:50 | |
| *** psachin has joined #openstack-swift | 12:52 | |
| *** links has quit IRC | 12:53 | |
| *** psachin has quit IRC | 12:59 | |
| *** psachin has joined #openstack-swift | 13:03 | |
| *** links has joined #openstack-swift | 13:05 | |
| *** AndyWojo has quit IRC | 13:09 | |
| *** AndyWojo has joined #openstack-swift | 13:10 | |
| *** psachin has quit IRC | 13:15 | |
| tdasilva | good morning | 13:31 |
| tdasilva | acoles: the patch was to pring the metadata checksum | 13:32 |
| tdasilva | acoles: but that does sound like a good idea | 13:32 |
| acoles | tdasilva: right, thanks. maybe someone else did it. | 13:59 |
| *** klamath has joined #openstack-swift | 14:03 | |
| *** klamath has quit IRC | 14:03 | |
| *** klamath has joined #openstack-swift | 14:03 | |
| *** psachin has joined #openstack-swift | 14:07 | |
| *** _JZ_ has joined #openstack-swift | 14:11 | |
| *** dmorita has joined #openstack-swift | 14:12 | |
| *** dmorita has quit IRC | 14:17 | |
| *** bkopilov has joined #openstack-swift | 14:17 | |
| *** StraubTW has joined #openstack-swift | 14:26 | |
| *** jordanP has joined #openstack-swift | 14:33 | |
| *** si1ver has joined #openstack-swift | 14:44 | |
| openstackgerrit | Shashirekha Gundur proposed openstack/python-swiftclient: modify 'swift <sub_command> —help' display https://review.openstack.org/390119 | 14:47 |
| *** daemontool has joined #openstack-swift | 14:49 | |
| *** si1ver has quit IRC | 14:49 | |
| *** psachin has quit IRC | 15:04 | |
| *** klrmn has joined #openstack-swift | 15:05 | |
| *** chlong has joined #openstack-swift | 15:13 | |
| *** si1ver has joined #openstack-swift | 15:19 | |
| *** links has quit IRC | 15:20 | |
| *** sams-gleb has joined #openstack-swift | 15:29 | |
| jrichli | Is there a Denis Makogon here? | 15:41 |
| *** cdelatte has joined #openstack-swift | 15:43 | |
| *** delattec has joined #openstack-swift | 15:43 | |
| *** delattec has quit IRC | 15:44 | |
| *** cdelatte has quit IRC | 15:44 | |
| *** cdelatte has joined #openstack-swift | 15:44 | |
| *** arch-nemesis has quit IRC | 15:46 | |
| *** ppai has quit IRC | 15:59 | |
| *** chsc has joined #openstack-swift | 16:02 | |
| *** daemontool has quit IRC | 16:05 | |
| *** SkyRocknRoll has quit IRC | 16:05 | |
| notmyname | good morning | 16:12 |
| notmyname | tdasilva: looks like there was some progress on ec stuff yesterday. what do you need me to do to help with the releases? | 16:14 |
| notmyname | also, is the get_version() patch going to be part of it? | 16:14 |
| notmyname | also, I support merging pyeclib and liberasurecode :-) | 16:14 |
| *** SkyRocknRoll has joined #openstack-swift | 16:15 | |
| notmyname | also, I had a conversation here yesterday with someone who wants to add another EC backend and get involved | 16:15 |
| tdasilva | notmyname: could you help review https://review.openstack.org/#/c/408257/ | 16:15 |
| patchbot | patch 408257 - liberasurecode - 1.4.0 release | 16:15 |
| tdasilva | and https://review.openstack.org/#/c/408280/ | 16:15 |
| patchbot | patch 408280 - pyeclib - release 1.4.0 | 16:15 |
| tdasilva | notmyname: was it https://memoscale.com/ ? | 16:16 |
| notmyname | no | 16:16 |
| tdasilva | ok | 16:16 |
| notmyname | but that's cool too :-) | 16:17 |
| tdasilva | heh, i think they were in barcelona or austin | 16:17 |
| tdasilva | can't remember | 16:17 |
| tdasilva | notmyname: re: get_version() IIRC it looks like kota_ would like to have it in, but clayg and timburke were fine if it landed after release | 16:18 |
| tdasilva | so i guess it would be nice to get consensus | 16:18 |
| *** pcaruana has quit IRC | 16:19 | |
| *** mingyu has joined #openstack-swift | 16:23 | |
| notmyname | it's already got a +2 on it (https://review.openstack.org/#/c/405926/) | 16:25 |
| patchbot | patch 405926 - liberasurecode - Add get_version functionality to liberasurecode lib | 16:25 |
| notmyname | unless there's a reason we need the release *TODAY*RIGHT*NOW* instead of tomorrow, then let's land p 405926 and do the release including it | 16:27 |
| patchbot | https://review.openstack.org/#/c/405926/ - liberasurecode - Add get_version functionality to liberasurecode lib | 16:27 |
| *** mvk has quit IRC | 16:30 | |
| *** rcernin has quit IRC | 16:31 | |
| *** oshritf has quit IRC | 16:32 | |
| *** hseipp has joined #openstack-swift | 16:34 | |
| *** vinsh has joined #openstack-swift | 16:35 | |
| *** cbartz has left #openstack-swift | 16:39 | |
| *** SkyRocknRoll has quit IRC | 16:41 | |
| *** mingyu has quit IRC | 16:56 | |
| openstackgerrit | Alistair Coles proposed openstack/swift: line wrap overview_expiring_objects.rst https://review.openstack.org/408717 | 16:58 |
| *** rcernin has joined #openstack-swift | 17:03 | |
| acoles | tdasilva: found it, it was timburke's patch 341663 :) | 17:10 |
| patchbot | https://review.openstack.org/#/c/341663/ - swift - Call out transient sysmeta in swift-object-info | 17:10 |
| tdasilva | acoles: nice! we should try to land that, it will probably be useful for debugging encrypted data files?? | 17:11 |
| *** hseipp has quit IRC | 17:11 | |
| acoles | yes! this ^^ might help joeljwright diagnose his mixed up etags. if you PUT an object with some x-object-meta-foo and have encryption in proxy then only upgraded object servers will store the associated x-object-transient-sysmeta. | 17:12 |
| acoles | So using patched swift-object-info on "dodgy" objects would indicate upgrade or not. | 17:13 |
| jrichli | I'll take a look at the patch as well. and, btw, i wasn't aware of joeljwright having the above mentioned issue. | 17:16 |
| jrichli | ah, i see i missed some scrollback. reading now | 17:19 |
| joeljwright | acoles, jrichli: thanks, we're still digging, it really it the worst kind of transient failure | 17:21 |
| acoles | joeljwright: fyi I am not here tomorrow but back next week | 17:22 |
| joeljwright | acoles: now I'm confused **and** jealous! | 17:22 |
| pdardeau | i put together some basic docs about swift ec pieces/parts (for my own understanding) | 17:25 |
| pdardeau | http://goo.gl/rlhA0M | 17:25 |
| pdardeau | will continue filling in some of the missing pieces | 17:26 |
| clayg | good morning | 17:31 |
| clayg | i still see no tarballs on liberasure releases - so i'm guessing i'm still working on that today | 17:32 |
| acoles | joeljwright: if you can PUT an object with user metadata (x-object-meta-*) and then use swift-get-nodes to find the data file for one of the misbehaving objects, then do python -c "import swift.obj.diskfile as df;print(df.read_metadata('path_to_.data_file'))" | 17:32 |
| *** vinaypotluri has joined #openstack-swift | 17:33 | |
| acoles | joeljwright: you should see a key(s) starting 'X-Object-Transient-Sysmeta-Crypto-Meta- | 17:33 |
| acoles | joeljwright: if you don't, then the object server was not upgraded | 17:33 |
| joeljwright | acoles: that's a great test, thanks | 17:34 |
| clayg | notmyname: i need a release of liberasurecode and pyeclib today - also patch 405926 doesn't really solve anything without the pyeclib change to expose it - and pyeclib can't merge that because of dependencies and distros | 17:34 |
| patchbot | https://review.openstack.org/#/c/405926/ - liberasurecode - Add get_version functionality to liberasurecode lib | 17:34 |
| acoles | if we'd merge timburke's patch then swift-object-info would have done that for you :/ | 17:34 |
| tdasilva | acoles: just got started with this: https://etherpad.openstack.org/p/swift_tiering_foundational_work | 17:34 |
| tdasilva | acoles: looking forward to your (and others) input | 17:34 |
| acoles | tdasilva: great, thanks | 17:35 |
| *** rledisez has quit IRC | 17:35 | |
| tdasilva | going out to get some lunch, brb | 17:36 |
| acoles | joeljwright: note, you need to have some X-Object-Meta-Blah=blah with the PUT for there to be any 'X-Object-Transient-Sysmeta-Crypto-Meta expected | 17:36 |
| *** vinaypotluri has left #openstack-swift | 17:36 | |
| acoles | joeljwright: are you using EC policy or replica? | 17:37 |
| joeljwright | acoles: replica | 17:38 |
| clayg | well what's wrong with patch 341663 !? | 17:38 |
| patchbot | https://review.openstack.org/#/c/341663/ - swift - Call out transient sysmeta in swift-object-info | 17:38 |
| acoles | clayg: us its whats wrong, we're so bad at reviewing stuff | 17:38 |
| joeljwright | acoles: but it's not consistent either, we've managed to get 3 different etags for a single object | 17:39 |
| joeljwright | acoles: it's hard to identify a single 'problem' object | 17:39 |
| clayg | acoles: I think we're bad at finding patches that should merge | 17:39 |
| clayg | acoles: that one looks like it should merge | 17:40 |
| joeljwright | acoles: we think we've found one, then it behaves consistently and correctly for ages :( | 17:42 |
| acoles | joeljwright: single object or same content? | 17:44 |
| acoles | i.e. same path or same content? | 17:45 |
| *** dmorita has joined #openstack-swift | 17:45 | |
| joeljwright | acoles: same path | 17:46 |
| acoles | clayg: thanks for introducing some sanity to reclaim_age conf, patch 374419, always drive me crazy | 17:46 |
| patchbot | https://review.openstack.org/#/c/374419/ - swift - Move documented reclaim_age option to correct loca... | 17:46 |
| joeljwright | acoles: same path, same content | 17:47 |
| joeljwright | 3 different etags | 17:47 |
| *** dmorita has quit IRC | 17:47 | |
| joeljwright | but mostly the correct one | 17:47 |
| *** dmorita has joined #openstack-swift | 17:47 | |
| acoles | so what's not good enough about "mostly" :P | 17:47 |
| acoles | 3 ?!? | 17:47 |
| acoles | 2 makes sense | 17:48 |
| clayg | acoles: dunno how i missed the doc build thing - i normally like to have my autodoc going when I'm editing in that tree just cause i'm curious - very strange | 17:48 |
| acoles | joeljwright: unless of course you also have some proxies upgraded and others not | 17:48 |
| acoles | joeljwright: but did you say you're hitting just one proxy? | 17:48 |
| clayg | is this something with made up etags on *LO's? | 17:49 |
| acoles | joeljwright: same path, same content, same PUT request? or repeating same request gives etag variations? | 17:50 |
| acoles | clayg: rst tables are pretty brittle it seems | 17:51 |
| *** mingyu has joined #openstack-swift | 17:51 | |
| acoles | clayg: I learnt that the hard way so I am alert to changes in those tables | 17:51 |
| clayg | acoles: yeah i learned that one time too - good catch | 17:52 |
| clayg | acoles: reviews work! | 17:52 |
| clayg | i'm mad at encryption right now - i have both mw in my pipeline; disable_encryption = false (worst double negative) - but i'm not getting encrypted stuff! | 17:53 |
| clayg | is there another step i'm forgetting? my "modified pipeline" log lines show they mw is in there... | 17:54 |
| clayg | acpahpcahp! i just rebuilt everything - werx now | 17:54 |
| acoles | joeljwright: assume you put the same content multiple times - every path gets encrypted differently, and every unique request to same path gets encrypted differently, hence the Etag of the data on disk is different in every case. And old object server will send that Etag to container and that would explain varying etags in listings. Upgraded object servers should send the encrypted etag of the plaintext which should be co | 17:55 |
| acoles | nstant. | 17:55 |
| joeljwright | acoles: same content, same path - only one PUT (keep etag) then just HEADs | 17:55 |
| joeljwright | acoles: result is sometimes inconsistent, but with no pattern we have been able to identify | 17:56 |
| acoles | joeljwright: huh. I was working off David's pastebin from earlier, thought you were seeing the issue with container GETs http://paste.openstack.org/show/591755/ | 17:57 |
| acoles | so object HEADs give back varying Etags? | 17:57 |
| clayg | ok, i'ma assume you guys got this and go look at liberasurecode releases | 17:58 |
| joeljwright | acoles: yes, we've been trying to refine the problem since David's pastebin | 17:58 |
| acoles | joeljwright: so like clayg mentioned, asre these slo's or just plain old objects? | 17:58 |
| *** McMurlock1 has quit IRC | 17:59 | |
| *** jordanP has quit IRC | 17:59 | |
| joeljwright | acoles, clayg: just simple objects | 17:59 |
| joeljwright | acoles, clayg: but we're unable to reproduce this consistently | 18:00 |
| joeljwright | acoles: yeah, the 3 really confused me | 18:00 |
| joeljwright | acoles: it was this morning, but we're pretty sure we didn't overwrite the object at any point | 18:02 |
| *** _david_cole_soho has joined #openstack-swift | 18:04 | |
| clayg | patch 405450 looks really close to merged to me - it's mostly just a logging and test change - me, acoles and kota seem on board - i'd like it landed because it's a fixup for another change that i'm packaging | 18:04 |
| patchbot | https://review.openstack.org/#/c/405450/ - swift - Fixups for ghost listing fix | 18:04 |
| acoles | joeljwright: any chance the HEADs use different proxies and one wasn't upgraded? | 18:07 |
| joeljwright | acoles: sadly not | 18:07 |
| joeljwright | I forced the request through individual proxies multiple times | 18:08 |
| acoles | hmm. ok well if you were able to do that read_metadata on one of these objects and capture the on disk metadata of all three replicas that would be very interesting | 18:09 |
| joeljwright | I'll do my best | 18:09 |
| *** StraubTW has quit IRC | 18:10 | |
| *** StraubTW has joined #openstack-swift | 18:10 | |
| timburke | good morning | 18:11 |
| *** silor1 has joined #openstack-swift | 18:14 | |
| clayg | timburke: good morning! | 18:20 |
| openstackgerrit | Clay Gerrard proposed openstack/liberasurecode: 1.4.0 release https://review.openstack.org/408257 | 18:21 |
| *** silor1 has quit IRC | 18:25 | |
| *** zul_ has joined #openstack-swift | 18:31 | |
| * acoles away til Monday | 18:33 | |
| *** tqtran has joined #openstack-swift | 18:33 | |
| *** acoles is now known as acoles_ | 18:34 | |
| openstackgerrit | Clay Gerrard proposed openstack/pyeclib: release 1.4.0 https://review.openstack.org/408280 | 18:36 |
| *** clu_ has joined #openstack-swift | 18:39 | |
| openstackgerrit | Clay Gerrard proposed openstack/pyeclib: release 1.4.0 https://review.openstack.org/408280 | 18:39 |
| clayg | ok, I think the liberasurecode/pyeclib release is ready to go (cc: timburke tdasilva notmyname ) | 18:41 |
| clayg | please let me know if that is blocked or looks like it won't happen this afternoon if there is anything I can do | 18:41 |
| clayg | also, gah, the error logs please - patch 405450 | 18:42 |
| patchbot | https://review.openstack.org/#/c/405450/ - swift - Fixups for ghost listing fix | 18:42 |
| timburke | joeljwright: maybe i missed it; what do you see when you do swift-object-info for all of the replicas? | 18:42 |
| timburke | and it works *fine* without my patch, right? it just lumps all that stuff under "Other Metadata" | 18:43 |
| timburke | clayg: tdasilva: i'm more and more partial toward merging the get_Version thing, at least for liberasurecode. that shit works, we can totally get a pyeclib patch in building against distro packages, it's gonna be great | 18:44 |
| timburke | (whether we get that patch in ahead of the pyeclib release is a separate matter) | 18:45 |
| *** zul_ has quit IRC | 18:46 | |
| *** zul_ has joined #openstack-swift | 18:48 | |
| *** siva_krish has quit IRC | 18:53 | |
| *** ChubYann has joined #openstack-swift | 18:59 | |
| tdasilva | timburke: whether we get that patch in ahead of the pyeclib release is *THE* matter | 19:03 |
| tdasilva | clayg: ^ | 19:03 |
| tdasilva | timburke: or are you saying merge the libec patch ahead of the libec release | 19:03 |
| tdasilva | and then it might be ok to wait for the pyeclib patch for after the pyeclib release | 19:04 |
| tdasilva | ? | 19:04 |
| timburke | tdasilva: the latter. if we end up wanting to do a pyeclib 1.4.1 or something to pick up better version handling, w/e | 19:04 |
| timburke | or i could finish putting out a separate small fire, and help get the pyeclib patch in shape to merge today. both are good options. | 19:06 |
| timburke | but i see no reason that the libec code can't merge as is | 19:06 |
| *** hseipp has joined #openstack-swift | 19:22 | |
| *** joeljwright has quit IRC | 19:25 | |
| clayg | pdardeau: ntata: do you know Sachin Patil - i'm not sure why i associate him with osic - he's doing a bang up job on patch 406012 | 19:33 |
| patchbot | https://review.openstack.org/#/c/406012/ - swift - Fix swift-get-nodes arg parsing for missing ring | 19:33 |
| clayg | I really hope he sticks with it - it's turning into a great cleanup effort! so helpful! | 19:34 |
| tdasilva | clayg: he is from RH | 19:34 |
| clayg | pdardeau: ntata: NM! | 19:34 |
| tdasilva | but i don't know him personally | 19:34 |
| clayg | tdasilva: hrm... wonder what group he's with - or if he's on irc? | 19:34 |
| tdasilva | psachin i thik | 19:35 |
| tdasilva | think | 19:35 |
| tdasilva | but he is India, so different TZ | 19:35 |
| clayg | timburke: i disagree that it matters at all if the version code thing is added to liberasure in 1.4.0 | 19:37 |
| *** mingyu has quit IRC | 19:37 | |
| clayg | timburke: IMHO the whole situation with liberasurecode and pyeclib dependencies and gate tests is %^&*ed and I don't think having a gate job that builds against liberasurecode master so we can skip tests that run against distro pacakges is the right target | 19:37 |
| timburke | clayg: one fewer things that need to be released after landing the pyeclib change to actually use it | 19:37 |
| clayg | timburke: I *do* agree that the liberasurecode function can get merged without breaking the pyeclib gate - but I do not agree that provides any value to anyone because pyeclib can't use it because our whole situation is screwed | 19:38 |
| clayg | meanwhile - i really need to shipt isa_l_rs_cauchy and i have few braincells nerves to spare for how to un^&* liberasurecode and pyeclib | 19:39 |
| timburke | clayg: pyeclib can *totally* use it. working on updating kota_'s patch now | 19:39 |
| timburke | and that can still be *after* a 1.4.0 release | 19:39 |
| clayg | but I appreciate your thinking about it - not trying to deminish that effort - you are a rockstart | 19:39 |
| clayg | ... but I disagree that this change should hold up liberasurecode 1.4.0 - it's more complicated than that - if we have to cut a 1.4.1 next week I don't care at all | 19:39 |
| clayg | production production production | 19:39 |
| timburke | what was that earlier about the trouble being finding code that should merge? | 19:40 |
| clayg | timburke: ok i don't understand how you can make unitests that are using distro packaged .so's that don't include the function you want to expose exercise the *unexposed in that version* function - but i'm sure you can educate me if we had a few days to kill | 19:41 |
| openstackgerrit | Merged openstack/swift: Call out transient sysmeta in swift-object-info https://review.openstack.org/341663 | 19:42 |
| clayg | timburke: exactly - in this case it's not obvious to me that it should merge! I obviously don't care if it's obvious to tdasilva that it *should* - as long as 1.4.0 is not blocked any longer than it has to be | 19:42 |
| timburke | clayg: oh yeah, the unit tests are gonna be worthless. it's going to rely *entirely* on people actually testing against a more-recent liberasurecode | 19:42 |
| clayg | gvlop[oftuyiuop-oiugyhol | 19:42 |
| patchbot | Error: Missing "]". You may want to quote your arguments with double quotes in order to prevent extra brackets from being evaluated as nested commands. | 19:42 |
| clayg | [;l'jbkhj | 19:42 |
| patchbot | Error: Missing "]". You may want to quote your arguments with double quotes in order to prevent extra brackets from being evaluated as nested commands. | 19:42 |
| *** rcernin has quit IRC | 19:43 | |
| clayg | timburke: ok, fair to say that we disagree on both the correctness and value of this approach - quite possible i'm wrong | 19:43 |
| timburke | do you want to have code that works or tests that are meaningful? if we wait for the tests to be meaningful, you'll forever have this chicken-and-egg problem | 19:43 |
| clayg | timburke: but I'm acctually choosing not pass judgement since I don't know enough to evaluate the correctness - frustrated it's being raised as a blocker against 1.4.0 tho :'( | 19:44 |
| timburke | "no point in merging libec change, pyeclib can't use it" "no point in even *thinking* about pyeclib change; we'll never be able to test it" | 19:44 |
| timburke | i don't think it's a blocker, i think it's a nice-to-have that already has a +2 | 19:45 |
| clayg | timburke: i *just* want to talk about this after I get out my release :'( | 19:45 |
| tdasilva | timburke, clayg so i just looked at the libec change | 19:45 |
| clayg | i also think that merging pyeclib into liberasure code is what we should be doing | 19:45 |
| clayg | yay - great! | 19:45 |
| tdasilva | it looks straight forward and I tested as timburke suggested | 19:45 |
| tdasilva | looks great! | 19:45 |
| clayg | tdasilva: awesome! go guys! | 19:45 |
| tdasilva | i just don't know if i'm missing something | 19:45 |
| tdasilva | but I'm ready to +2 +A | 19:46 |
| clayg | tdasilva: did you figure out how it's going to be used in pyeclib!? | 19:46 |
| tdasilva | and get this over with | 19:46 |
| clayg | tdasilva: good call! | 19:46 |
| tdasilva | no, not thinking about that right now | 19:46 |
| clayg | ROFL | 19:46 |
| tdasilva | just want to merge this and move on | 19:46 |
| tdasilva | clayg, timburke thoughts? | 19:47 |
| timburke | i'm totally on board | 19:48 |
| * clayg hugs timburke | 19:48 | |
| clayg | thanks tdasilva ! | 19:48 |
| openstackgerrit | OpenStack Proposal Bot proposed openstack/swift: Updated from global requirements https://review.openstack.org/88736 | 19:54 |
| openstackgerrit | Merged openstack/liberasurecode: Add get_version functionality to liberasurecode lib https://review.openstack.org/405926 | 19:54 |
| *** mingyu has joined #openstack-swift | 19:55 | |
| openstackgerrit | Thiago da Silva proposed openstack/liberasurecode: 1.4.0 release https://review.openstack.org/408257 | 19:57 |
| *** rcernin has joined #openstack-swift | 19:58 | |
| tdasilva | clayg, timburke: can you +2 p 408257 again? | 19:58 |
| patchbot | https://review.openstack.org/#/c/408257/ - liberasurecode - 1.4.0 release | 19:58 |
| tdasilva | timburke: also did you have a chance to look at p 408280 ? | 19:58 |
| patchbot | https://review.openstack.org/#/c/408280/ - pyeclib - release 1.4.0 | 19:58 |
| timburke | i've not looked at either yet; sorry :-( | 19:59 |
| timburke | will do today | 19:59 |
| *** mingyu has quit IRC | 20:00 | |
| *** mingyu has joined #openstack-swift | 20:04 | |
| *** jamielennox|away is now known as jamielennox | 20:13 | |
| clayg | tdasilva: thanks! | 20:15 |
| openstackgerrit | Tim Burke proposed openstack/pyeclib: WIP: Change the version reference https://review.openstack.org/405929 | 20:17 |
| timburke | kota_: ^^^ hope you don't mind | 20:17 |
| openstackgerrit | Thiago da Silva proposed openstack/pyeclib: release 1.4.0 https://review.openstack.org/408280 | 20:20 |
| timburke | "Verified+1" see that, we can totally get this through the gate when we want | 20:21 |
| clayg | tdasilva: gah! nice catch :P | 20:21 |
| *** mingyu has quit IRC | 20:21 | |
| clayg | timburke: :) | 20:23 |
| tdasilva | notmyname: around? | 20:26 |
| clayg | tdasilva: I think he's traveling (SLC? OpenStack Midwest?) - so he may be spotty | 20:27 |
| tdasilva | ah ok! | 20:27 |
| tdasilva | cool, np | 20:27 |
| timburke | "mountain west." different. | 20:27 |
| clayg | i mostly piece it together from twitter - i have no idea - he said something at some point in this channel this am I think? | 20:27 |
| clayg | timburke: ah, thank you -> https://twitter.com/notmyname/status/806508214615412736 | 20:28 |
| *** d0ugal has quit IRC | 20:29 | |
| openstackgerrit | Merged openstack/liberasurecode: 1.4.0 release https://review.openstack.org/408257 | 20:31 |
| *** zul_ has quit IRC | 20:33 | |
| clayg | ^ wooooooo! | 20:33 |
| *** d0ugal has joined #openstack-swift | 20:35 | |
| openstackgerrit | Merged openstack/pyeclib: release 1.4.0 https://review.openstack.org/408280 | 20:47 |
| *** rcernin has quit IRC | 20:51 | |
| *** rcernin has joined #openstack-swift | 20:51 | |
| zaitcev | waaah, no 1.3.2 | 20:52 |
| tdasilva | zaitcev: decided to go with 1.4.0 since we added support for a new EC lib | 20:55 |
| zaitcev | which one is that? | 20:55 |
| zaitcev | I thought both shss and isa-l were in 1.3.1 | 20:56 |
| timburke | zaitcev: isa-l cauchy; 1.3.1 has isa-l vand | 20:57 |
| zaitcev | does it allow for more parities? | 20:58 |
| zaitcev | (than 4) | 20:58 |
| timburke | yup :-) at least, that's what clayg and kota_ keep telling me | 20:58 |
| clayg | zaitcev: yeah it do! isa_l_rs_cauchy is where it's *at* | 21:02 |
| *** rcernin has quit IRC | 21:09 | |
| tdasilva | clayg: release tags have been pushed | 21:09 |
| *** joeljwright has joined #openstack-swift | 21:10 | |
| *** ChanServ sets mode: +v joeljwright | 21:10 | |
| *** chlong has quit IRC | 21:10 | |
| *** rcernin has joined #openstack-swift | 21:12 | |
| clayg | tdasilva: yeah it has! | 21:13 |
| clayg | https://github.com/openstack/liberasurecode/releases | 21:13 |
| *** siva_krish has joined #openstack-swift | 21:14 | |
| tdasilva | my changes to pyeclib did not work as expected https://pypi.python.org/pypi/PyECLib/1.4.0 :( | 21:14 |
| *** siva_krish has quit IRC | 21:24 | |
| *** chsc has quit IRC | 21:25 | |
| clayg | tdasilva: my pypi username is clayg - if you want to add me on the project and I can poke around? | 21:27 |
| *** dmorita has quit IRC | 21:27 | |
| clayg | tdasilva: maybe only kgreen has the ability to change that particular piece of metadata? | 21:28 |
| clayg | tdasilva: as timburke pointed out - it was kind of a log shot that the metadata would help | 21:28 |
| * clayg shrugs | 21:28 | |
| *** dmorita has joined #openstack-swift | 21:28 | |
| *** mvk has joined #openstack-swift | 21:33 | |
| *** StraubTW has quit IRC | 21:33 | |
| *** Jeffrey4l has quit IRC | 21:35 | |
| *** Jeffrey4l has joined #openstack-swift | 21:36 | |
| tdasilva | clayg: added you the project | 21:37 |
| *** vint_bra has quit IRC | 21:40 | |
| *** sams-gleb has quit IRC | 21:47 | |
| *** sams-gleb has joined #openstack-swift | 21:47 | |
| *** sams-gleb has quit IRC | 21:52 | |
| openstackgerrit | Merged openstack/swift: Use assertGreater(len(x), 0) instead of assertTrue(len(x) > 0) https://review.openstack.org/408488 | 21:52 |
| *** hseipp has quit IRC | 21:54 | |
| *** hseipp has joined #openstack-swift | 21:55 | |
| clayg | tdasilva: yeah i couldnt see anything - i'm in #pypa see if anyone there knows how it works | 22:00 |
| tdasilva | clayg: yeah, looks like they changed their web interface, so it's not very clear | 22:01 |
| clayg | zaitcev: timburke: tdasilva: notmyname: torgomatic: making good progress on packages for liberasure/pyeclib now - only hold out for swift is patch #405450 | 22:02 |
| patchbot | https://review.openstack.org/#/c/405450/ - swift - Fixups for ghost listing fix | 22:02 |
| clayg | I would love to see that go in before my updaters loose their mind on async pendings | 22:02 |
| zaitcev | clayg: wait, what packages | 22:03 |
| zaitcev | clayg: I've already started Fedora build for 1.4.0 that I got with git archive -o ../liberasurecode-1.4.0.tar.gz --prefix=liberasurecode-1.4.0/ 1.4.0 | 22:03 |
| clayg | pyeclib liberasurecode 1.4.0 packages! | 22:04 |
| clayg | zaitcev: nice sounds like we're on the right track | 22:04 |
| zaitcev | ok | 22:04 |
| clayg | zaitcev: after i finish with pyeclib & liberasurecode i'm *also* building packages for swift with a bunch of >2.11 bug fixes that have gone in | 22:04 |
| tdasilva | zaitcev: should i ping apevec about centos packages? | 22:05 |
| clayg | i'm sure notmyname will be thinking about cutting a swift 2.12 pretty soon for the same reason | 22:05 |
| zaitcev | tdasilva: too soon | 22:05 |
| clayg | tdasilva: is that how that works? yeah we should send it to liberasure/pyeclib-annouce or something? | 22:05 |
| zaitcev | clayg: I think CentOS use DLRN, which is a script that automatically builds whatever is the latest tag. | 22:06 |
| clayg | zaitcev: yeah now *that's* what i'm talking about! | 22:06 |
| tdasilva | zaitcev: not sure it's automatic? for the last time, i had to ask apevec to kickoff a build, sounded like it was a more manual process? | 22:07 |
| zaitcev | tdasilva: well, maybe they realized that it was wasteful | 22:07 |
| tdasilva | hehehe | 22:07 |
| tdasilva | maybe i'm looking at the wrong thing | 22:07 |
| tdasilva | i've been looking at this: https://cbs.centos.org/koji/packageinfo?packageID=643 | 22:08 |
| zaitcev | looks fine to me. I was confused about what was good in 1.3.x cycle, so I skipped it | 22:10 |
| zaitcev | we had this issue about some more silly messages flooding with 1.3.0 or such | 22:10 |
| clayg | i think 1.3.1 included a fix for that maybe? | 22:12 |
| zaitcev | maybe | 22:12 |
| zaitcev | But I have good hopes for 1.4.0, so let's go with that. | 22:12 |
| clayg | yeah i think i packaged 1.3.1 *for* that fix - oh maybe that was to make isa_l_rs_vand return an error instead of garbage on erase lists with parity > 4 | 22:12 |
| clayg | zaitcev: i kid you not - 1.4.0 is the best liberasurecode/pyeclib evar | 22:13 |
| tdasilva | lol | 22:13 |
| clayg | i have it on good authority liberasurecode even has a function that can tell you waht it's version is | 22:13 |
| clayg | ;) | 22:13 |
| zaitcev | not like anyone calls it though | 22:13 |
| * clayg snickers | 22:13 | |
| tdasilva | ROFL | 22:14 |
| clayg | i'm so so so just kidding - it was for a laugh - god i'm such a jerk :'( | 22:14 |
| tdasilva | timburke: i'm a little pessimistic about that prefix tempurl api | 22:15 |
| tdasilva | re path 274048 | 22:15 |
| tdasilva | patch 274048 | 22:15 |
| patchbot | https://review.openstack.org/#/c/274048/ - swift - tempurls with a prefix-based scope | 22:15 |
| timburke | tdasilva: the implementation, or the whole premise? | 22:16 |
| tdasilva | it seems awkward to have to provide the patch to a whole object like https://swift-cluster.example.com/v1/my_account/container/my_prefix/object only to then say pprefix temp_url_prefix=my_prefix | 22:16 |
| tdasilva | s/patch/path | 22:18 |
| tdasilva | can't spell today | 22:18 |
| *** jamielennox is now known as jamielennox|away | 22:29 | |
| mattoliverau | morning, I probably wont be in much today. At the airport, flying to Sydney for a day in the Sydney office and Christmas party. | 22:31 |
| tdasilva | timburke: nevermind, i understand better now on second read | 22:33 |
| *** jamielennox|away is now known as jamielennox | 22:34 | |
| *** dmorita has quit IRC | 22:36 | |
| *** dmorita has joined #openstack-swift | 22:37 | |
| timburke | hrm. although you still can't create SLOs... | 22:39 |
| *** hseipp has quit IRC | 22:40 | |
| *** hseipp has joined #openstack-swift | 22:40 | |
| clayg | mattoliverau: party time! | 22:40 |
| mattoliverau | you know it! | 22:40 |
| *** cschwede has quit IRC | 22:43 | |
| *** tmoreira has quit IRC | 22:44 | |
| *** joeljwright has quit IRC | 22:52 | |
| *** david-lyle_ has joined #openstack-swift | 23:05 | |
| *** david-lyle has quit IRC | 23:07 | |
| *** rcernin has quit IRC | 23:11 | |
| notmyname | hello | 23:18 |
| notmyname | online at the airport (and flight was just delayed) | 23:18 |
| notmyname | clayg: tdasilva | 23:18 |
| clayg | notmyname: afaik everything is awesome tdasilva and timburke are hero - zaitcev and I are building packages | 23:19 |
| notmyname | awesome! | 23:19 |
| notmyname | tdasilva: did you pushed the signed tags? | 23:19 |
| *** chsc has joined #openstack-swift | 23:21 | |
| clayg | when I fetch them they sure *look* signed - PGP some such - release tarballs were published, pypi was updated, i think tdasilva knows what he's doing | 23:21 |
| notmyname | yay | 23:21 |
| clayg | nice | 23:21 |
| notmyname | oh, sorry, I meant that in "was it tdasilva who did it" instead of "did tdasilva know what to do" | 23:21 |
| notmyname | cool, yeah, my local fetch finally finished. I see the tags | 23:22 |
| clayg | oh interesting... i have know idea if tdasilva had to do something by hand or magic - if anything good happened there i'm pretty sure it was tdasilva | 23:22 |
| clayg | as opposed to kota_ or timburke (who are also both high on my list of blame when something good happens related to ec libararies) | 23:23 |
| clayg | *praise | 23:23 |
| notmyname | indeed | 23:26 |
| notmyname | clayg: the "magic" for a release (of non-governance openstack projects like pyeclib and libec) is simply to make a signed tag and push it to the gerrit remote | 23:26 |
| notmyname | for governance projects, you have to create/update some yaml in the openstack-releases repo and get that +A'd by the release team | 23:27 |
| clayg | notmyname: gtk | 23:27 |
| *** rcernin has joined #openstack-swift | 23:33 | |
| *** david-lyle_ is now known as david-lyle | 23:35 | |
| *** kei_yama has joined #openstack-swift | 23:37 | |
| clayg | EVERYTHING IS AWESOME! | 23:48 |
| *** klamath has quit IRC | 23:49 | |
| timburke | best release yet | 23:49 |
| openstackgerrit | Merged openstack/swift: line wrap overview_expiring_objects.rst https://review.openstack.org/408717 | 23:49 |
| *** rcernin has quit IRC | 23:49 | |
| *** jamielennox is now known as jamielennox|away | 23:49 | |
| openstackgerrit | Merged openstack/swift: adding unit tests for if-none-match fix https://review.openstack.org/401044 | 23:50 |
| notmyname | timburke: clayg: tdasilva: is there something pressing I should be looking at while sitting here? | 23:58 |
| notmyname | (if not, I'll be working on the 2.12.0 release notes) | 23:58 |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!