*** vnaboychenko has quit IRC | 00:36 | |
*** vnaboychenko has joined #magnetodb | 00:36 | |
*** vnaboychenko has quit IRC | 00:41 | |
*** vnaboychenko has joined #magnetodb | 01:16 | |
*** vnaboychenko has quit IRC | 02:10 | |
*** setho has joined #magnetodb | 04:24 | |
*** setho2 has quit IRC | 04:26 | |
*** vnaboychenko has joined #magnetodb | 04:50 | |
*** crxz0193 has joined #magnetodb | 04:51 | |
*** vnaboychenko has quit IRC | 05:07 | |
*** vnaboychenko has joined #magnetodb | 05:08 | |
*** vnaboychenko has quit IRC | 05:12 | |
*** vnaboychenko has joined #magnetodb | 05:27 | |
*** vnaboychenko has quit IRC | 06:19 | |
*** vnaboychenko has joined #magnetodb | 06:19 | |
*** vnaboychenko has quit IRC | 06:23 | |
*** achuprin has quit IRC | 06:59 | |
*** achuprin has joined #magnetodb | 07:11 | |
*** vnaboychenko has joined #magnetodb | 07:17 | |
*** jeromatron has quit IRC | 07:31 | |
*** ominakov has joined #magnetodb | 07:40 | |
ominakov | hello guys | 07:41 |
---|---|---|
ominakov | hello guys | 07:41 |
ikhudoshyn | o/ | 07:43 |
*** aostapenko has left #magnetodb | 08:26 | |
isviridov_ | hey ominakov ikhudoshyn | 08:40 |
*** jeromatron has joined #magnetodb | 08:41 | |
*** jeromatron has quit IRC | 08:43 | |
*** jeromatron has joined #magnetodb | 08:46 | |
*** jeromatron has quit IRC | 08:47 | |
*** aostapenko has joined #magnetodb | 08:52 | |
*** jeromatron has joined #magnetodb | 08:54 | |
*** istegantsev has joined #magnetodb | 08:55 | |
*** denis_makogon has joined #magnetodb | 09:03 | |
*** idegtiarov has joined #magnetodb | 09:10 | |
*** denis_makogon has quit IRC | 09:10 | |
ominakov | isviridov_, what do you mean by this comment "For schema we have documentation" ? (https://review.openstack.org/#/c/98408/1) | 09:11 |
*** isviridov_ is now known as isviridov | 09:12 | |
isviridov | ominakov, just wanted to say, that samples should be more specific | 09:12 |
*** denis_makogon has joined #magnetodb | 09:13 | |
isviridov | For example, there is general schema description here https://review.openstack.org/#/c/98408/1/doc/api-samples/batch_write_item_request_syntax.json,cm, but not a request sample. | 09:14 |
isviridov | From other side in API documentation we have such description. I mean this wiki https://wiki.openstack.org/wiki/MagnetoDB/api | 09:15 |
*** jeromatron has quit IRC | 09:38 | |
*** vnaboychenko has quit IRC | 09:53 | |
*** vnaboychenko has joined #magnetodb | 09:53 | |
openstackgerrit | Andrei V. Ostapenko proposed a change to stackforge/magnetodb: Adds cassandra keyspace creation on first table creation https://review.openstack.org/98169 | 09:57 |
*** vnaboychenko has quit IRC | 09:58 | |
openstackgerrit | Andrei V. Ostapenko proposed a change to stackforge/magnetodb: Adds cassandra keyspace creation on first table creation https://review.openstack.org/98169 | 10:22 |
openstackgerrit | Dmitriy Ukhlov proposed a change to stackforge/magnetodb: Fix expected condition bug https://review.openstack.org/99003 | 10:33 |
*** crxz0193 has quit IRC | 11:40 | |
openstackgerrit | Andrei V. Ostapenko proposed a change to stackforge/magnetodb: Adds cassandra keyspace creation on first table creation https://review.openstack.org/98169 | 11:52 |
*** crxz0193 has joined #magnetodb | 11:58 | |
*** crxz0193 has quit IRC | 12:26 | |
openstackgerrit | Ilya Sviridov proposed a change to stackforge/magnetodb: Added openstack theme for document generation https://review.openstack.org/99034 | 12:29 |
*** jeromatron has joined #magnetodb | 12:40 | |
isviridov | ikhudoshyn, dukhlov please take a look at https://review.openstack.org/99034 it is all for http://magnetodb.readthedocs.org/en/latest/ | 12:51 |
isviridov | Still troubleshooting, but changes are really isolated | 12:51 |
*** idegtiarov has quit IRC | 13:09 | |
openstackgerrit | Illia Khudoshyn proposed a change to stackforge/magnetodb: Add gunicorn support to streaming API https://review.openstack.org/99049 | 13:21 |
openstackgerrit | Illia Khudoshyn proposed a change to stackforge/magnetodb: Add gunicorn support to streaming API https://review.openstack.org/99049 | 13:22 |
isviridov | ikhudoshyn, here is documentation we are generating with my patch https://review.openstack.org/#/c/99034/ http://docs-draft.openstack.org/34/99034/1/check/gate-magnetodb-docs/1b9e6e8/doc/build/html/ | 13:25 |
openstackgerrit | Illia Khudoshyn proposed a change to stackforge/magnetodb: Add gunicorn support to streaming API https://review.openstack.org/99049 | 13:27 |
ikhudoshyn | isviridov, thanx, good to know, i've seen yr patch, will +2 it when jenkins job will be done | 13:28 |
isviridov | ikhudoshyn, already | 13:29 |
ikhudoshyn | ok, give me a sec | 13:29 |
ikhudoshyn | done | 13:31 |
ikhudoshyn | does anybody know how to cancel jenkins job | 13:32 |
*** dukhlov has quit IRC | 13:36 | |
isviridov | ikhudoshyn, no ideas. It can be only infra team | 13:37 |
isviridov | ikhudoshyn, are you sure? | 13:37 |
isviridov | ikhudoshyn, see three patches from you, and the latest is in queue | 13:38 |
ikhudoshyn | isviridov, great, that's what I meant - just didn't want to wait for (obsolete) patches #1, #2 | 13:45 |
*** CharlesW_ has joined #magnetodb | 13:52 | |
*** dukhlov has joined #magnetodb | 13:54 | |
CharlesW_ | With introduction of Oslo Notifier, we now have a dependency on evenlet (actually all OpenStack projects use eventlet). So far it seems compatible to have double dependencies. Most of the eventlet usage is just using it as library functions. Only one case is to act as an event receiver (zmq-receiver.py), and it will invoke eventlet’s monkey patch. All unit-tests/tempest tests/POSTMAN tests work fine. Both log notifier and rabi | 13:55 |
CharlesW_ | It would work since there's no eventlet.monkey_patch() call if you just send notification. If you want to receive notification using zmq, then eventlet.monkey_patch() will be called. | 13:55 |
CharlesW_ | With regard to the gevent vs. evenlet issue, I have tried to contact the primary developer (Tyler Hobbs) and haven’t heard back from him yet. I posted the question on datastax-drivers IRC channel. There are some interests and suggestions. | 13:55 |
CharlesW_ | Some other guys are looking for answers for the exactly same question. One suggestion was using asyncoreactor as the connection class. I tried that with eventlet monkey_patching. It wouldn’t work. Python driver just hang. | 13:55 |
CharlesW_ | Besides python driver, the other place we use gevent is the api wsgi server. It can be easily replaced with eventlet's version. I tried that and the wsgi server works fine. | 13:56 |
CharlesW_ | So at this point, to make our MVP date, I would think it’s safe to add the eventlet dependency. | 13:56 |
CharlesW_ | Thoughts? | 13:56 |
CharlesW_ | So, my point is: let's keep gevent as the monkey_patching module, and use eventlet as library functions in notifier | 13:58 |
dukhlov | Hello Charles | 13:58 |
CharlesW_ | Hi Dima | 13:58 |
dukhlov | It looks like that it is possible to have gevent and eventlet for now, but in this case we will have unexpected green threads scheduling. | 14:00 |
dukhlov | each of gevent and eventlet have its own thread proceccing loop | 14:02 |
CharlesW_ | can you expand on that? If we keep gevent's monkey patching, oslo notifier seems happy. | 14:02 |
CharlesW_ | We are not using eventlet for monkey patching | 14:02 |
dukhlov | so if you start gevent thread there is gevent loop for handeling threads is starting | 14:03 |
dukhlov | then when you spawn next gevent theread it is added to queue and will be handled by this loop | 14:04 |
dukhlov | then if you start eventlet thread from gevent thread - there is eventlet loop is started | 14:05 |
dukhlov | but it started inside logical gevent thread | 14:06 |
dukhlov | then when some eventlet thread call sleep() method or wait() then main gevent loop will be switched to next gevent green thread | 14:07 |
dukhlov | because we are using gevent monkey patching | 14:07 |
openstackgerrit | A change was merged to stackforge/magnetodb: Added openstack theme for document generation https://review.openstack.org/99034 | 14:08 |
dukhlov | and next eventlet thread will be executed only when the first eventlet green thread will be finished | 14:08 |
dukhlov | summarize, I want to say that coworking of eventlet and gevent is possible, but we can have some strange thread scheduling | 14:09 |
dukhlov | and also it is point of potential problems which we don't expect now | 14:10 |
CharlesW_ | I see. | 14:11 |
CharlesW_ | The remedy is then: add evenlet flavor in C* python driver | 14:12 |
dukhlov | yes it is the best solution and I think that it is easy to do | 14:12 |
dukhlov | because gevent is very similar to gevent | 14:13 |
dukhlov | and we can try to implement eventlet connection class based on GeventConnection code | 14:14 |
CharlesW_ | Yes, Cluster class does take a parameter: connection_class. We can pass in our own implementation | 14:16 |
CharlesW_ | I tried that quickly yesterday. It didn't work. Even though gevent and eventlet have similar APIs, they have different signatures many times. | 14:18 |
dukhlov | ok I will try today too | 14:20 |
CharlesW_ | thanks | 14:21 |
CharlesW_ | It might be a deployment issue | 14:23 |
CharlesW_ | I'll try again | 14:23 |
*** isviridov is now known as isviridov|away | 14:24 | |
*** crxz0193 has joined #magnetodb | 14:29 | |
*** ominakov has quit IRC | 14:47 | |
*** ominakov has joined #magnetodb | 14:54 | |
*** ominakov has quit IRC | 14:54 | |
*** isviridov|away is now known as isviridov | 14:56 | |
*** idegtiarov has joined #magnetodb | 15:02 | |
isviridov | ikhudoshyn, great job. Just +2ed | 15:02 |
ikhudoshyn | isviridov, tnx | 15:02 |
*** ominakov has joined #magnetodb | 15:14 | |
*** idegtiarov has quit IRC | 15:36 | |
*** crxz0193 has quit IRC | 16:24 | |
*** [o__o] has quit IRC | 16:29 | |
*** [o__o] has joined #magnetodb | 16:31 | |
isviridov | bye bye guys | 16:40 |
*** isviridov is now known as isviridov|away | 16:40 | |
*** denis_makogon has quit IRC | 17:00 | |
*** dmakogon_ has joined #magnetodb | 17:00 | |
CharlesW_ | Tyler replied and agreed to create a bug in cassandra python driver: https://datastax-oss.atlassian.net/browse/PYTHON-79 | 17:38 |
openstackgerrit | A change was merged to stackforge/magnetodb: Add gunicorn support to streaming API https://review.openstack.org/99049 | 18:00 |
*** [o__o] has quit IRC | 18:18 | |
*** [o__o] has joined #magnetodb | 18:21 | |
openstackgerrit | Alexander Chudnovets proposed a change to stackforge/magnetodb: Added DELETE actions for sets https://review.openstack.org/99183 | 18:35 |
*** vnaboychenko has joined #magnetodb | 18:36 | |
*** CharlesW_ has quit IRC | 18:54 | |
*** CharlesW_ has joined #magnetodb | 18:55 | |
*** vnaboychenko has quit IRC | 18:55 | |
*** jeromatron has quit IRC | 18:56 | |
*** vnaboychenko has joined #magnetodb | 18:56 | |
CharlesW_ | I finally got Cassnadra driver to work with evenlet. | 18:57 |
*** vnaboychenko has quit IRC | 19:00 | |
isviridov|away | CharlesW_, great news. Would you share it as a patch? | 19:08 |
*** CharlesW_ has quit IRC | 19:12 | |
*** vnaboychenko has joined #magnetodb | 19:24 | |
*** istegantsev has quit IRC | 20:54 | |
*** achuprin has quit IRC | 20:54 | |
*** openstackgerrit has quit IRC | 20:54 | |
*** istegantsev has joined #magnetodb | 21:08 | |
*** achuprin has joined #magnetodb | 21:08 | |
*** openstackgerrit has joined #magnetodb | 21:08 | |
*** jeromatron has joined #magnetodb | 22:16 |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!