Thursday, 2016-05-26

*** openstack has joined #congress00:04
openstackgerritEric K proposed openstack/congress-specs: WIP - HA spec  https://review.openstack.org/31838300:13
*** openstackstatus has quit IRC00:46
*** sballe_ has quit IRC00:48
*** openstack has joined #congress00:55
*** openstackgerrit has quit IRC01:49
*** openstackgerrit has joined #congress01:57
*** masahito has quit IRC02:13
*** openstack has joined #congress02:39
*** ramineni_ has joined #congress02:54
*** fawadkhaliq has joined #congress03:42
*** ramineni_ has quit IRC03:51
*** fawadkhaliq has quit IRC03:56
openstackgerritAnusha Ramineni proposed openstack/congress: Tempest for Push Type Driver  https://review.openstack.org/31917303:56
*** fawadkhaliq has joined #congress04:57
*** masahito has joined #congress04:58
*** masahito has quit IRC04:58
*** masahito has joined #congress05:49
openstackgerritMerged openstack/congress: Tempest for Push Type Driver  https://review.openstack.org/31917306:20
*** openstack has joined #congress06:55
*** ramineni_ has joined #congress07:27
evrardjp'morning everyone07:52
masahito:)07:55
masahitoevrardjp: I recently see your name here. Are you using or intereseted in Congress? :-)07:57
evrardjpnot using right now07:57
evrardjpwhen I'll have time I'll create an openstack-ansible role for it07:57
evrardjpthis way congress users can consume it when deploying with ansible07:58
evrardjpas the community is currently small in congress, I thought it was maybe good to be involved/help07:59
evrardjpnow it's just a question of time07:59
evrardjpI'll probably ask questions in the future when I'll start working on this08:00
evrardjpin the meantime I'm trying to understand the community :)08:00
masahitooh great!08:00
masahitowe are still small community, so that kind of works will extremely improve this project ;-)08:02
evrardjpthat's what I thought08:04
evrardjpit could be bad if it's not maintained 'though08:05
evrardjpso when I'll start on this, I'll analyse the situation with some of you, this way we know the proper way of action08:05
masahitoramineni_: you there?09:13
ramineni_masahito: hi09:16
masahitoI checked ha replica error in gate jobs.09:20
ramineni_masahito: ya, are you able to figure out ..what is causing it?09:21
masahitoIt looks like dsenode doesn't start well with rabbit.09:22
ramineni_i reproduced it .. its causing only when i change to rabbit driver09:22
ramineni_hmm..y?09:23
ramineni_Im just seeing DatasourceNameInUser error in my local env09:23
masahitobecause service-topic-prefix and control-top are same in original node and dse node09:23
ramineni_oh09:23
ramineni_masahito: but multilple services can listen on same topic?09:24
ramineni_canĀ“t **09:24
ramineni_and any idea why is happenig only using rabbit?09:27
masahitofor rabbit I think the rpc which creating new datasource only arrived at one dse node and another doesn't have it.09:28
masahitoThe rpc call is rpc not cast.09:29
masahitoI'm still wondering why this error happens in gate-job which uses deepsix .09:30
ramineni_masahito: the one on your patch?09:31
ramineni_masahito: push driver tempest got merged with recheck only .. some timeout issue09:31
masahitobecause it start synchronyzer as a service to synchronize with db.09:31
masahitoramineni_: yes, for my patch.09:31
ramineni_masahito: but your patch got merged with just recheck09:32
masahitoyup.09:32
masahitobefore it's merged I thought the root cause is in another place not related to dse and dse2.09:33
masahitoSo I wanted to hear your status :)09:34
ramineni_masahito: hmm.. might be .. im also not very clear on what is the issue .. i couldn reproduce it on my local env with your patch, so i did rebase and recheck on your patch09:34
ramineni_it merged :)09:35
ramineni_masahito: may be next time , we have to investigate more , if this happens in gate job using deepsix09:35
ramineni_masahito: but with rabbit driver, HA is not working :( any clue on how to resolve it?09:37
ramineni_masahito: https://review.openstack.org/#/c/320306/09:37
ramineni_masahito: one doubt , with current arch , can we start multiple congress services on diff nodes? like in case of pacemaker09:39
ramineni_masahito: curr arch doesn allow right09:39
masahitowe can start multiple congress with current (deepsix) architecture, but we can't start it with new (dse node on rabbit) architecture.09:41
masahitoand we can start it with new (dse node on kombu+memory) architecture.09:42
masahitoin the case of rabbit, I saw the following happens.09:43
ramineni_hmm .. to enable rabbit , we have to change that right , otherwise HA wont be possible09:43
masahito1. start main Congress (congress1 with PE1, novaDS1 and etc)09:44
masahito2. start replica Congress (congress2 with PE2, novaDS2 and etc)09:45
masahito3. novaDS2 calls RPC to create datasource policy with 'congress-service-engine' topic09:46
masahito4-1: if PE2 receives the rpc, the creation succeed.09:46
masahito4-2. if PE1 receives the rpc, the creation fails because the datasource policy has already exists in PE1.09:47
ramineni_masahito: but PE1 and PE2 synchronize with same databse right09:47
masahito5. if 4-2. happens replica Congress stops.09:48
ramineni_masahito: so it should fail anyway, if it queries DB for policy?09:48
masahitosorry, what timing do you mention?09:51
masahitoduring initializing replica node or during creating new datasource with API?09:52
ramineni_initializing replica node?09:55
ramineni_but i think we dont create persistent policies there , just we store in self.theory , so should be fine i guess09:56
ramineni_masahito: ^^09:56
*** fawadkhaliq has quit IRC09:57
*** fawadkhaliq has joined #congress09:58
ramineni_masahito: do you  think we should keep defaulting to rabbit driver patch on hold , until we figure out the arch to support multiple instances of congress starting up without conflict?09:58
masahitoI think we don't need to abandon it if we hope Congress works on rabbit.10:01
masahitowhen we figure out the root cause and its patch we can push it basing your patch ;-)10:01
ramineni_masahito: sounds good :)10:02
masahitooh, I think I resolve it.10:12
masahitoJust add a partition id for rpc.10:13
masahitoI'll pushed the patch.10:13
masahitooops, sorry. I didn't resolve it :-<10:20
openstackgerritMasahito Muroi proposed openstack/congress: Check root cause of tempest test fail  https://review.openstack.org/32145910:22
*** ramineni_ has quit IRC10:55
*** absubram has quit IRC13:27
openstackgerritMasahito Muroi proposed openstack/congress: Check root cause of tempest test fail  https://review.openstack.org/32145913:38
*** thinrichs has joined #congress15:20
*** ramineni_ has joined #congress15:30
*** ramineni_ has quit IRC15:51
*** thinrichs has quit IRC16:55
*** absubram has joined #congress16:55
*** absubram has quit IRC16:56
*** absubram has joined #congress16:59
*** fawadkhaliq has quit IRC17:07
*** thumpba has joined #congress17:08
*** thinrichs has joined #congress17:09
*** fawadkhaliq has joined #congress17:27
*** _xin has quit IRC17:56
*** _xin has joined #congress17:58
*** thumpba has quit IRC17:58
*** thumpba_ has joined #congress18:02
*** thinrichs has quit IRC18:04
bryan_atthi congress team - quick question for anyone. Is there a spec for fields in congress.conf file? As of Liberty this has to be tox-generated but I need to know which conf values are required vs optional; the generated output doesn't clarify that. This is for the Puppet Module and JuJu Charm I am developing.18:26
*** thumpba_ has quit IRC18:47
*** masahito has quit IRC18:51
*** thumpba has joined #congress18:59
*** thumpba has quit IRC19:11
*** fawadkhaliq has quit IRC19:16
*** thumpba has joined #congress19:18
*** thinrichs has joined #congress19:44
*** absubram has quit IRC19:55
*** absubram has joined #congress20:11
*** thinrichs has quit IRC20:39
*** thumpba has quit IRC20:51
*** ekcs has joined #congress21:13
*** absubram has quit IRC23:07
*** rena9067 has joined #congress23:18
*** rena9067 has quit IRC23:20

Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!