Wednesday, 2020-02-12

lileMaking sure I'm starting on the right track, hacluster_pcs is continually restarting.  Inspecting the container shows it can't start because it can't find ssl.rb in /usr/share/pcsd, it should be /usr/lib/pcsd.  Loading ssl.rb from there results in a failure to find the rack rubygem, so I'm installing that in the kolla hacluster_pcs Dockerfile via rubygem-rack.01:40
goldyfruitlile, which base ?01:51
goldyfruitUbuntu or Centos?01:51
lileCentos01:54
lileDo I need to manifest the path based on Ubuntu v. Centos?01:54
goldyfruitI did all my tests on Ubuntu01:54
goldyfruitTry to build on Ubuntu01:54
lileAh, ok, I'll look at how to properly address that01:55
lileSorry, I don't have Ubuntu here and our target is CentOS02:16
lileI looked at the panko j2 files, I see how to properly address it02:16
lilehttps://www.irccloud.com/pastebin/0vPdFVO2/02:50
*** vishalmanchanda has joined #openstack-masakari04:13
lilegoldyfruit, is ruby >= 2.2 on Ubuntu?04:33
lilegoldyfruit, I got things running in the container, but it's going to require an scl rh-ruby-2.5 for CentOS 7 base.05:15
*** goldyfruit has quit IRC05:58
*** goldyfruit has joined #openstack-masakari05:58
*** goldyfruit has quit IRC06:30
*** goldyfruit has joined #openstack-masakari06:30
*** vishalmanchanda has quit IRC11:12
goldyfruitlile, https://github.com/openstack/kolla/tree/master/docker/hacluster14:03
goldyfruitThis is the commit https://github.com/openstack/kolla/commit/046158b1494938a61f8a9b98d8b0bba88295f2df14:03
goldyfruitThe repository should be there already14:04
goldyfruit"Stable High Availability/Clustering packages (CentOS_CentOS-7)"14:04
goldyfruitBut they changed something: https://github.com/openstack/kolla/commit/cf11cd6f3d1f0f0d9e1bb948672be12171a4099b14:05
goldyfruitYou will have to enable the repo14:05
goldyfruit"    The way of handling external repositories has changed. All such repos are14:06
goldyfruit    now disabled by default. There is "enable_extra_repos" macro to enable them14:06
goldyfruit    by name. Repositories names are defined in the "etc/repos.yaml" file."14:06
*** vishalmanchanda has joined #openstack-masakari14:11
lilegoldyfruit, thanks for the pointer.  I'll pull that thread and see where I can get14:31
lileIf I can't get there, I have a workaround in my back pocket14:32
lilekolla/train hasn't merged that change.  The repos are enabled by default in the version I'm using.14:47
goldyfruitI checked master only14:50
lileNo problem14:52
lileWhat I ran into, is the fact that CentOS is way behind on ruby (2.0.0).  There isn't a path to get all the gem dependencies in alignment, it falls apart at sinatra-contrib, I can't get an old enough version for sinatra/cookies to work with ruby 2.0.014:55
lileWhat version of ruby where you using on Ubuntu?14:55
goldyfruitFrom Bionic14:57
lileOk, that's ruby 2.5.1, I'll get the centos container to use that via scl.  If the kolla folks don't want it pulled in dynamically, I'll add it as a proper repo14:58
*** openstackgerrit has joined #openstack-masakari16:07
openstackgerritShilpa Devharakar proposed openstack/masakari master: Add support for multiple filters  https://review.opendev.org/70481016:07
openstackgerritShilpa Devharakar proposed openstack/masakari master: Add host query parameter support  https://review.opendev.org/70741516:07
openstackgerritShilpa Devharakar proposed openstack/python-masakariclient master: Add new filter `host_name` to segment list  https://review.opendev.org/70481116:09
openstackgerritShilpa Devharakar proposed openstack/python-masakariclient master: Add support for multiple filters  https://review.opendev.org/70741716:09
*** vishalmanchanda has quit IRC17:51
goldyfruitlile, https://review.opendev.org/697712 has been merged few minutes ago19:09
lilegoldyfruit, will that be merged back to train?19:36
lilei've got hacluster_pcs working on centos19:36
goldyfruitI can do the cherry pick19:36
goldyfruitGood !19:36
lilethe cluster isn't connecting though, it appears to be a nodeid issue19:36
lilehttps://www.irccloud.com/pastebin/2c1UCgx9/19:37
lilehttps://www.irccloud.com/pastebin/zvD5ZNcv/19:38
lileI've manually fixed the nodeid's and it seems to be starting now19:41
lileI think this might be better?19:47
lilehttps://www.irccloud.com/pastebin/VSnnhf6U/19:47
lileok, with my changes, I seem to have a working pacemaker/corosync and hostmonitor cluster on Centos19:50
liledeploying the full cluster now and will test hv node ha shortly19:51
*** ChanServ has quit IRC20:50
*** ChanServ has joined #openstack-masakari21:13
*** orwell.freenode.net sets mode: +o ChanServ21:13
lilegoldyfruit, the hostmonitor is hacluster_pacemaker_remote correct?22:48
goldyfruitpacemaker_remote and masakari-hostmonitor need to be on the compute22:49
lilewhere do pull masakari-hostmonitor on train?22:52
lileI don't see it in kolla or kolla-ansible for train22:52
goldyfruitThere don't exist22:55
goldyfruitI only build masakari-instancemonitor22:55
lileso I just need to build out a container for hostmonitor from masakari-monitors?22:57
goldyfruityeah22:58
liledid you ever get the pcsd ui working over the web?23:05
lileI get nothing but Unable to connect to node 10.232.194.237, no token available in the logs23:06
lileI think authkey needs to get pushed to the container in kolla-ansible23:08
lilelooks like it might need both the pacemaker and corosync authkeys23:15
lileah, regarding hostmonitor, it's already in masakari-monitors, I just need to redeploy it with kolla-ansible for hostmonitor.  That's a lot easier.23:35
*** goldyfruit has quit IRC23:45
*** goldyfruit has joined #openstack-masakari23:45

Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!