| *** openstackgerrit has quit IRC | 03:04 | |
| *** openstackgerrit has joined #magnetodb | 03:05 | |
| *** shakamunyi has quit IRC | 05:14 | |
| *** ominakov has joined #magnetodb | 05:42 | |
| *** shakamunyi has joined #magnetodb | 06:10 | |
| openstackgerrit | Dmitriy Ukhlov proposed a change to stackforge/magnetodb: Parallelize BatchWrite operation https://review.openstack.org/86644 | 06:11 |
|---|---|---|
| *** crxz0193 has joined #magnetodb | 06:13 | |
| *** shakamunyi has quit IRC | 06:15 | |
| ominakov | hi all | 06:19 |
| achuprin | Hi everyone | 06:22 |
| *** crxz0193 has quit IRC | 06:35 | |
| *** crxz0193 has joined #magnetodb | 06:37 | |
| *** achuprin has quit IRC | 06:55 | |
| *** idegtiarov has joined #magnetodb | 07:03 | |
| *** achuprin has joined #magnetodb | 07:10 | |
| *** shakamunyi has joined #magnetodb | 07:14 | |
| *** shakamunyi has quit IRC | 07:19 | |
| *** crxz0193 has quit IRC | 07:22 | |
| *** crxz0193 has joined #magnetodb | 07:23 | |
| *** jeromatron has joined #magnetodb | 07:45 | |
| isviridov | Good morning, achuprin | 07:50 |
| *** crxz0193 has quit IRC | 08:08 | |
| *** crxz0193 has joined #magnetodb | 08:09 | |
| *** crxz0193 has quit IRC | 08:13 | |
| *** jeromatron has quit IRC | 08:14 | |
| *** achuprin has quit IRC | 08:16 | |
| *** achuprin has joined #magnetodb | 08:16 | |
| miarmak | good morning guys! | 08:30 |
| *** keith_newstadt has quit IRC | 08:32 | |
| *** jeromatron has joined #magnetodb | 08:36 | |
| *** jeromatron has quit IRC | 09:09 | |
| *** jeromatron has joined #magnetodb | 09:09 | |
| *** jeromatron has quit IRC | 09:10 | |
| *** jeromatron has joined #magnetodb | 09:12 | |
| *** jeromatron has quit IRC | 09:16 | |
| *** jeromatron has joined #magnetodb | 09:16 | |
| *** jeromatron has quit IRC | 09:20 | |
| *** jeromatron has joined #magnetodb | 09:58 | |
| *** denis_makogon has joined #magnetodb | 10:08 | |
| openstackgerrit | Aleksey Chuprin proposed a change to stackforge/magnetodb: Changed cassandra version from 2.0.6 to 2.0.7 https://review.openstack.org/89810 | 12:16 |
| *** [o__o] has quit IRC | 12:43 | |
| *** [o__o] has joined #magnetodb | 12:45 | |
| ominakov | isviridov, hi | 13:29 |
| ominakov | isviridov, please verify this bug https://bugs.launchpad.net/magnetodb/+bug/1297247 | 13:40 |
| ikhudoshyn | guys, looks like there is a bug in C* that bothers me pretty much, https://issues.apache.org/jira/browse/CASSANDRA-5987 | 13:59 |
| ikhudoshyn | does anybody know anything about it aside of what can be read in jira? | 14:00 |
| ikhudoshyn | jeromatron, I just wonder could you be any closer to the ^ ^ then the rest of us? | 14:01 |
| jeromatron | hmm, sounds like there wasn't necessarily a problem in the Cassandra code, Yuki gave him some suggestions to go from. | 14:03 |
| jeromatron | the bulk loader is used for a variety of purposes including restoring from backup, so I would be surprised if there were any problems that don't get worked out quickly. | 14:04 |
| *** ominakov has quit IRC | 14:06 | |
| miarmak | Hey guys, take a look please on this bp, and join the discussion in it's whiteboard: https://blueprints.launchpad.net/magnetodb/+spec/configurable-consistency | 14:15 |
| jeromatron | ikhudoshyn do you have a reproducible case of that bug? | 14:21 |
| ikhudoshyn | jeromatron, well, I tried to follow Yuki's suggestion but failed with it | 14:39 |
| ikhudoshyn | yep, i do | 14:40 |
| isviridov | ominakov, done. Nice shot | 14:44 |
| jeromatron | @ikhudoshyn I was just asking Yuki and he said "hmm, now we have CQL SSTableWriter so he can try that instead." | 14:46 |
| ikhudoshyn | jeromatron, do you mean using SSTableWriter instead of SSTableSimpleUnsortedWriter? | 14:48 |
| jeromatron | ikhudoshyn https://github.com/apache/cassandra/blob/cassandra-2.0.7/src/java/org/apache/cassandra/io/sstable/CQLSSTableWriter.java | 14:49 |
| ikhudoshyn | jeromatron, thanks, gonna go and try | 14:51 |
| jeromatron | np - let us know on the ticket or in here if that doesn't work for you | 14:52 |
| *** CharlesW has joined #magnetodb | 14:54 | |
| *** ominakov has joined #magnetodb | 14:56 | |
| isviridov | Welcome back, CharlesW | 14:56 |
| isviridov | Can we congratulate you? | 14:57 |
| *** idegtiarov has quit IRC | 15:04 | |
| openstackgerrit | Charles Wang proposed a change to stackforge/magnetodb: Implement the RESTFul API of DeleteItem operation https://review.openstack.org/84889 | 15:09 |
| openstackgerrit | A change was merged to stackforge/magnetodb: Changed cassandra version from 2.0.6 to 2.0.7 https://review.openstack.org/89810 | 15:11 |
| ikhudoshyn | jeromatron, big thanks, it works for me | 15:13 |
| jeromatron | ikhudoshyn great to hear. do you mind updating the ticket indicating that for anyone who stumbles across it? | 15:14 |
| ikhudoshyn | sure, I can do that | 15:17 |
| isviridov | dukhlov, around? | 15:19 |
| CharlesW | I'm also looking for Dmitriy | 15:21 |
| CharlesW | Thanks isviridov. I have a wonderful new addition to my family. Very happy :) | 15:25 |
| isviridov | Is it boy or girl? | 15:27 |
| *** dukhlov_ has joined #magnetodb | 15:27 | |
| isviridov | dukhlov, or dukhlov_ | 15:28 |
| isviridov | ? | 15:28 |
| isviridov | )) | 15:28 |
| CharlesW | Boy. Now I have a girl and boy in my family. | 15:28 |
| dukhlov_ | hi guys | 15:28 |
| miarmak | dukhlov_: hello! | 15:28 |
| isviridov | CharlesW, it is great news! I'm just 2 months ahead of you, man :)) | 15:29 |
| CharlesW | boy or girl? | 15:31 |
| isviridov | CharlesW, it is a boy | 15:31 |
| CharlesW | Great. Will some of you guys go to Atlanta OS summit? | 15:35 |
| CharlesW | @dukhlov_, wrt async table creation, shall we continue our discussion about two weeks ago? | 15:43 |
| CharlesW | or let me know a more convenient time for you so we can chat? | 15:46 |
| dukhlov_ | CharlesW: unfortunately I haven't finished my plugability task and can't give you information to start your task | 15:47 |
| dukhlov_ | but is you have general questions we can to talk about it | 15:47 |
| dukhlov_ | *if | 15:48 |
| *** achuprin has quit IRC | 15:51 | |
| CharlesW | question: will schema disagreement happen and require manual intervention if schema changes come in in parallel? Or it just takes long time to reach schema agreement and disagreement will resolve on its own? | 16:02 |
| *** jeromatron has quit IRC | 16:03 | |
| dukhlov_ | 1 | 16:08 |
| dukhlov_ | 1) Cassandra handle parallel schema changes correct | 16:09 |
| dukhlov_ | and we shoudn't have schema disagreement on Cassandra level | 16:10 |
| dukhlov_ | eventually | 16:10 |
| dukhlov_ | and there we don't have problem with long time of that operation | 16:11 |
| dukhlov_ | but on MagnetoDB level we need to check that schema is the same on each node | 16:12 |
| dukhlov_ | if you use original python-driver you can meet some concurrency problem | 16:14 |
| dukhlov_ | it can return that table already distributed between all nodes but it actually isn't | 16:15 |
| dukhlov_ | to fix this we patched python-driver | 16:15 |
| CharlesW | so what's the problem if we use the execute_async to create table, and return CREATING to magnetodb client. Client needs to run describe table on their own to make sure it's in active state | 16:16 |
| dukhlov_ | and now we have only problem with possible long time of create table operation | 16:16 |
| dukhlov_ | 1) we need to use patched python-driver | 16:17 |
| dukhlov_ | 2) we can get more longer time of async_create table operation and describe table will return status CREATING for 5 min for example instead of 1 min | 16:19 |
| *** ominakov has quit IRC | 16:20 | |
| dukhlov_ | 3) it hard to recover after fails: we can have a lot of workers and we heve no information about them | 16:21 |
| dukhlov_ | if some table creation hang we have no information why | 16:22 |
| dukhlov_ | is it still in progress | 16:22 |
| dukhlov_ | or it was failed | 16:22 |
| dukhlov_ | It is possible problem which I expect | 16:23 |
| CharlesW | 1) python driver should be fixed on C* side to handle schema refresh | 16:25 |
| CharlesW | 2) depending on the Q on magnetodb side, it can also take long time > 1 min | 16:26 |
| CharlesW | 3) C* is supposed to handle this, shouldn't it? | 16:26 |
| dukhlov_ | 1) I think so, but we need to talk with its developers first | 16:27 |
| dukhlov_ | 2) yes, but this queue will be processed with predictabe time and we will change status to ACTIVE when table is distributed between nodes immediately | 16:31 |
| dukhlov_ | 3) No I mean problem on Magnetodb layer. For example our magnetodb instance crashed. And async work will never finished. We restart instance, but our async task is lost | 16:34 |
| dukhlov_ | we can send create_table again by client | 16:34 |
| dukhlov_ | but we don't know is that task was on that crashed node | 16:35 |
| miarmak | guys, pay attantion please to this bp: https://blueprints.launchpad.net/magnetodb/+spec/configurable-consistency and join discussion in it's whiteboard. Thanks! | 16:35 |
| miarmak | and bye all :) | 16:35 |
| dukhlov_ | or maybe it serves by another node and it stiil alive | 16:35 |
| dukhlov_ | miarmak: sure | 16:36 |
| dukhlov_ | miarmak: bye! | 16:36 |
| isviridov | dukhlov, bye bye | 17:22 |
| *** denis_makogon has quit IRC | 17:32 | |
| *** denis_makogon has joined #magnetodb | 17:33 | |
| *** denis_makogon has quit IRC | 17:36 | |
| *** ominakov has joined #magnetodb | 17:42 | |
| dukhlov_ | isviridov: bye! | 17:53 |
| *** ominakov has quit IRC | 18:29 | |
| *** ominakov has joined #magnetodb | 19:33 | |
| *** jeromatron has joined #magnetodb | 19:40 | |
| *** ominakov has quit IRC | 19:41 | |
| *** pas-ha has joined #magnetodb | 19:55 | |
| *** pas-ha has left #magnetodb | 19:56 | |
| *** jeromatron has quit IRC | 20:04 | |
| *** jeromatron has joined #magnetodb | 20:41 | |
| *** jeromatron has quit IRC | 21:39 | |
| *** CharlesW has quit IRC | 22:33 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!