Monday, 2016-04-11

*** hoangcx_ has joined #openstack-ha00:48
*** hoangcx has quit IRC00:48
*** hoangcx_ is now known as hoangcx00:48
*** masahito has joined #openstack-ha01:06
*** hoangcx has quit IRC03:22
*** hoangcx has joined #openstack-ha03:22
*** moiz has joined #openstack-ha03:57
*** moizarif has joined #openstack-ha03:58
*** moiz has quit IRC04:02
*** hoangcx has quit IRC04:48
*** hoangcx has joined #openstack-ha04:49
*** pcaruana has joined #openstack-ha05:39
*** dimsum__ has quit IRC05:40
*** dims has joined #openstack-ha05:47
*** nkrinner has joined #openstack-ha05:54
*** mjura has joined #openstack-ha06:11
*** dims has quit IRC06:31
*** dims has joined #openstack-ha06:54
*** dims has quit IRC06:58
*** markvoelker has joined #openstack-ha07:01
*** dgurtner has joined #openstack-ha07:03
*** dgurtner has joined #openstack-ha07:03
*** markvoelker has quit IRC07:06
*** dgurtner has quit IRC07:13
*** dgurtner has joined #openstack-ha07:15
*** jpena|off is now known as jpena07:37
*** dims has joined #openstack-ha07:45
aspiershi all07:57
aspiers-> #openstack-meeting07:57
aspiersif anyone's around08:01
aspiershmm, maybe people are sticking to 9am UTC08:07
aspiersnot 9am Europe/London08:07
aspiersI should change the schedule08:07
ddejaaspiers: I think so08:12
masahitoI'm sticking to 9am UTC, b/c Japan don't have summertime :)08:37
masahitoDid the meeting end?08:38
*** masahito has quit IRC08:39
*** masahito has joined #openstack-ha08:39
aspiersmasahito: it didn't start yet08:40
ddejamasahito: I didn't start08:41
ddejaheh08:41
masahitook08:41
*** haukebruno has joined #openstack-ha08:48
*** dims has quit IRC08:48
*** dims has joined #openstack-ha08:55
*** pcaruana has quit IRC09:00
aspiersanyone else here now?09:01
ddejaI am09:01
aspiersok09:02
aspiersif masahito is too we could have the meeting, even if only 3 of us09:02
*** markvoelker has joined #openstack-ha09:02
*** rossella_s has quit IRC09:03
*** rossella_s has joined #openstack-ha09:03
*** markvoelker has quit IRC09:07
*** masahito has quit IRC10:10
moizarifhi, my canonical POD is up with Controller HA. just added remote nodes in the pacemaker cluster. with the command "crm configure primitive compute0-B3 ocf:pacemaker:remote   op monitor interval=20"10:33
moizarifcrm status was working earlier. but after remote node addition it throws "Segmentation fault (core dumped)10:33
moizarifERROR: crm_mon exited with code 139 and said:"10:33
moizarifsince PCS is not present in the system. i manually cloned it from github10:34
moizarifand when i run pcs status nodes i can see remote nodes as ONLINE10:35
moizarifwhy crm status is throwing the above errors?10:35
moizarifpcs status shows "Error: cluster is not currently running on this node"10:37
ddejamoizarif: Hi. For the last error - try 'sudo' ;)10:38
moizarifi am in root :(10:38
ddejahm10:38
moizarifp.s. when i remove the remote node with pcs resource delete Compute0-B3 --force ... the crm status and pcs status start working again10:39
ddejaI'm not expert in crm - I always use pcs... can you try to add this node using pcs and see if this helps?10:40
moizarifi did that too. pcs resource create Compute0-B3 ocf:pacemaker:remote op monitor interval=2010:41
moizarifsame issue..10:41
moizarifcrm_mon throws: 6 Nodes configured10:43
moizarif17 Resources configured10:43
moizarifbut at the end says: Segmentation fault (core dumped)10:43
ddejamoizarif: the only thing that I have in mind right now is that you may have not share the aythenitcation key beetwen controllers and remote10:43
moizarifi did that too... /etc/pacemaker/authkey with all controllers and remote ndoes10:44
moizarifgenerated using: dd if=/dev/urandom of=/etc/pacemaker/authkey bs=4096 count=110:44
aspiersmoizarif: why use pcs instead of crmsh? IIUC, pcs makes more sense on RH where it is heavily tested10:45
aspiersoh I see, you got a crash10:45
aspiersmoizarif: please report the crash at https://github.com/ClusterLabs/crmsh/issues10:46
aspiersmoizarif: you can expect a lot of these kinds of problems when you build the stack yourself10:46
aspiersmoizarif: it took us months to get it right10:47
moizarif1. canonical doesn't come with "pcs", also not available upstream. so the only choice is crm. but i am very used to pcs as i have used it extensively on RHOS10:47
moizarif2. Okay let me report it.10:47
moizarif3. months.. i gave myself 3 days :( considering it worked on RHEL rather easily and took me 5 days along with extensive debugging10:48
haukebrunointeresting. for me it is vice versa :p worked with crmsh for years, so switching to pcs is kinda like _difficult_10:50
moizarif@haukebruno: then you are my goto guy for crm :D10:51
aspiershaha10:52
aspierswell I can assure you that the crmsh developers are superb when it comes to support10:52
aspiersthey often fix bugs I find within hours or even minutes :)10:52
aspiersI think #clusterlabs is the right channel10:53
haukebruno:p10:53
aspiersmoizarif: if you are using the OCF agents approach, yes it would work on RHEL rather easily since that's where it was originally developed10:53
aspiersit's never been tested on Ubuntu before, AFAIK10:54
aspiersbut you are very welcome to try, of course :)10:54
aspiersit depends on what your goals are10:54
*** pece has joined #openstack-ha10:57
moizarifMy goal here is to test VM evacuations on Canonical (just as i did with RHEL). and yes i am trying all this on canonical, lets see how well it goes10:58
moizarif@aspiers: https://github.com/ClusterLabs/crmsh/issues/13110:59
aspiersoh wow10:59
aspiersyour pacemaker is ANCIENT10:59
aspiersyou can completely forget about using remotes with 1.1.10 :)10:59
moizariflatest upstream from canonical :(10:59
moizarifoh.. upgrade may be?11:00
aspiersyou'll need 1.1.13 absolute minimum I think11:00
aspiersbut ideally 1.1.1411:00
moizarifon RHEL i had pacemaker-1.1.13-10.el7.x86_6411:01
haukebrunoyep 1.1.10 is the lates in trusty :/11:01
aspiersthat's terrible11:01
aspiersthere must be newer packages somewhere?11:01
*** markvoelker has joined #openstack-ha11:03
haukebrunonot in 14.04 IIRC11:03
aspiersI think even 1.1.13 is not enough11:04
aspiersor 1.1.1411:04
moizarif@aspiers: my evacs worked perfectly with 1.1.13 on RHEL11:04
aspierswe found some critical issues which are now fixed in master and then backported them11:04
aspiersmoizarif: yes, but your remotes would not have been stable11:04
haukebrunoIIRC we/you need 16.04 for pacemaker 1.1.1411:05
aspiersmoizarif: we found they could be spontaneously fenced for no good reason11:05
aspiersand we fixed the issues upstream11:06
moizarif@aspiers: yes actually that happened to me a couple of times while i was setting up the cluster. but on my 2nd install the setup was stable.11:06
*** markvoelker has quit IRC11:07
moizarifoh right... thats why my 2nd install was stable.. wow.. you guys did a perfect job :)11:08
aspiersmoizarif: well, or maybe you just got lucky :)11:16
aspiersmoizarif: I don't even know whether RH packages include those fixes yet11:16
aspiersmoizarif: I only know they were backported to SLE HA11:16
moizarifi see11:20
moizarifokay so, the issue i am facing is more related to Ubuntu with the obsolute pacemaker packages. and i need to ask ubuntu for support on this issue.11:37
haukebrunomoizarif, do you have any canonical subscription?11:40
moizarifsupport subscription ?11:40
haukebrunoyes11:43
haukebrunocould be helpful here :P we don't have it, therefore I ask11:44
*** masahito has joined #openstack-ha11:50
*** masahito has quit IRC11:55
*** markvoelker has joined #openstack-ha12:19
moizarif@mashito: which pacemaker version are you using with your masakari deploy ? and which ubuntu release ?12:25
moizarif@haukebruno: unfortunately no :(12:25
moizarif@haukebruno: can i compile pacemaker 1.1.14 from source on my 14.04 install ? would it work?12:32
*** jpena is now known as jpena|lunch12:37
haukebrunomoizarif, never tried that :/12:43
aspiersmoizarif: do you have an estimate on when you would like to try SOC?12:47
moizarifhhmm i can may be give it a try and see what happens. would it be worth the effort in your opinion?12:49
moizarif@aspiers: its down the pipeline of my To Do's. would have to put the canonical experiment to some conclusion before i move on.12:50
aspierssure12:51
*** jpena|lunch is now known as jpena13:49
*** FL1SK has quit IRC13:49
*** kgaillot has joined #openstack-ha14:04
hoonetorghi14:11
hoonetorganybody here uses zmq instead of rabbitmq/qpid.14:11
hoonetorgis zmq feasible for HA, if yes how?14:12
hoonetorg(any link to a documentation would be nice :) )14:12
*** mjura has quit IRC14:49
*** dgurtner has quit IRC15:15
*** dgurtner has joined #openstack-ha15:15
*** dgurtner has joined #openstack-ha15:15
*** openstackgerrit has quit IRC15:18
*** openstackgerrit has joined #openstack-ha15:19
*** dgurtner has quit IRC15:23
*** dgurtner has joined #openstack-ha15:24
*** dgurtner has joined #openstack-ha15:24
*** FL1SK has joined #openstack-ha15:48
*** haukebruno has quit IRC16:05
*** nkrinner has quit IRC16:10
dimshoonetorg : please ping ozamiatin on #openstack-oslo17:04
hoonetorgdims: thx17:04
*** dgurtner has quit IRC17:07
*** haukebruno has joined #openstack-ha17:18
*** jpena is now known as jpena|off17:27
*** kgaillot has quit IRC19:48
*** haukebruno has quit IRC19:55
*** kgaillot has joined #openstack-ha19:58
*** openstackgerrit has quit IRC20:05
*** dgurtner has joined #openstack-ha20:20
*** openstackgerrit has joined #openstack-ha20:28
*** openstackgerrit has quit IRC20:41
*** openstackgerrit has joined #openstack-ha20:55
*** rossella_s has quit IRC21:03
*** rossella_s has joined #openstack-ha21:03
*** yan-gao has quit IRC21:06
*** yan-gao has joined #openstack-ha21:10
*** pece has quit IRC21:28
*** dgurtner has quit IRC22:06
*** dgurtner has joined #openstack-ha22:30
*** markvoelker has quit IRC22:33
*** dgurtner has quit IRC23:09
*** kgaillot has quit IRC23:30
*** markvoelker has joined #openstack-ha23:34
*** markvoelker has quit IRC23:39

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