13:59:27 <slashme> #startmeeting freezer
13:59:28 <openstack> Meeting started Thu Sep 15 13:59:27 2016 UTC and is due to finish in 60 minutes.  The chair is slashme. Information about MeetBot at http://wiki.debian.org/MeetBot.
13:59:29 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
13:59:32 <openstack> The meeting name has been set to 'freezer'
13:59:44 <slashme> Hello everyone :)
14:00:19 <slashme> As usual, Meeting topics and recap is available here: https://etherpad.openstack.org/p/freezer_meetings
14:01:24 <szaher> slashme: Hi :)
14:01:40 <yangyape_> hello guys
14:03:18 <ddieterly> o/
14:03:49 <slashme> #topic Release.
14:03:49 <domhnallw> o/
14:04:26 <slashme> Just to inform you that we are going to be releasing newton RC1 today.
14:04:34 <domhnallw> Nice!
14:05:48 <yangyape_> slashme: Whether we should take a depth bug fix and test :)
14:05:50 <slashme> If you have fix for critical bugs to needs to be merged before the final release, we need to merge them quickly. If this is really required, we can have an RC2 next week
14:08:46 <slashme> Anything else about this ?
14:11:02 <slashme> Okay, Next topic
14:11:21 <slashme> Still on the administrative side
14:11:51 <slashme> #topic PTL
14:12:02 <slashme> PTL nomination for Ocata are this week
14:12:26 <slashme> Is there anyone that wish to nominate himself except ddieterly and I ?
14:12:37 <slashme> The votes will be held next week.
14:14:06 <yangyape_> whethe  you and ddieterly   have a plan in next  freezer version   teh  roadmap
14:15:38 <yangyape_> I think we should have a general plan about some BP some good idea,  such as ceph  support， tenant backup and so on.
14:15:52 <yangyape_> There should be a priority
14:16:29 <slashme> I agree
14:16:32 <vnogin> o/
14:19:39 <vnogin> slashme, regarding critical issue: https://review.openstack.org/#/c/367917/ due to some reason if we use nember of replicas 2 - shards in elasticsearch are failed. Do we need to investigate why?
14:20:57 <slashme> Is it a one node cluster ?
14:22:06 <vnogin> yes, there is only one node with elasticsearch. Actually we use freezer api outside the cloud infrastructure
14:22:57 <slashme> This is why.
14:23:25 <slashme> Ift you set replica 2 on a 1 node cluster, ES tries to find another node to store the replica.
14:23:41 <slashme> And can't find one.
14:23:55 <slashme> This is why is ends up with an error.
14:24:19 <vnogin> correct, so from my point of view it's good idea to use number of replicas = 1 as default value, because all new guys in freezer face this issue after first installation
14:25:14 <slashme> Well I'd say it should even be 0.
14:25:27 <slashme> https://www.elastic.co/guide/en/elasticsearch/guide/current/_index_settings.html
14:26:33 <vnogin> slashme, agree. I'll change to 0 value
14:26:45 <yangyape_> vnogin: slashme https://review.openstack.org/#/c/367917/  change this patch replica to 0, +w quikly
14:26:47 <slashme> I think (but not completly sure) that the good value is ( "number of ES node in the cluster" - 1 )
14:28:05 <vnogin> slashme, for first use I think it's enough. After that everyone can fine tune it
14:28:13 <vnogin> yangyape_, yep
14:29:11 <vnogin> slashme, actually my point now - it doesn't work "from the box" now
14:30:21 <slashme> Well ... In production an ES cluster is usualy bigger that one node. This is why we set a replica value of 2. Thinking the configuration should be written for production...
14:31:02 <slashme> And in case someone is using a smaller ES cluster, he can set a number_of_replica=0
14:31:36 <slashme> But I'm not against changing the default behaviour.
14:31:55 <slashme> Though we should also update the documentation to specify what the value should be
14:32:51 <vnogin> slashme, ok, it good idea to add this information to documentation
14:35:57 <slashme> Okay so let's update that patch and merge it.
14:36:04 <slashme> Next topic
14:36:39 <slashme> #topic backup of all vm/volume in a tenant/openstack
14:37:06 <slashme> This is something we should implement in Ocata.
14:37:21 <slashme> The first and obious step is to repair nova/cinder backups
14:38:19 <yangyape_> nova backup should be ok, but i have a confused，
14:39:42 <yangyape_> container freezer_backup and freezer-backup-segment  which one to store data and metadata,
14:40:05 <yangyape_> I put_project nova data to freezer-backup-segment
14:43:25 <yangyape_> backup nova have a validate for path-to-backup. this patch need to review and merge   https://review.openstack.org/#/c/368296/
14:44:13 <slashme> Indeed
14:47:10 <vnogin> do we have any blueprint for all vm/vol backups? As far I as remember we need to have admin rights for nova list --all-tenants
14:47:35 <slashme> Regarding the feature of "backup of all vm/volume in a tenant/openstack" we will add it to the roadmap.
14:48:14 <slashme> We are implementing a feature to allow for the freezer-scheduler to execute multiple freezer-agent at the same time.
14:48:21 <slashme> This will be needed.
14:48:41 <yangyape_> slashme: nice.
14:49:52 <yangyape_> slashme:  whether we have a repair nova/cinder backup before rc2, and  backup all nova and cinder that have a discuss in next week?
14:50:10 <yangyape_> because 10 minutes left
14:50:53 <vnogin> nevertheless how are we going to gather information about all VMs and VOLs?
14:51:04 <slashme> yangyape_: yes
14:51:15 <slashme> What do you mean by informations ?
14:51:46 <vnogin> IDs for freezer agent which we want to backup
14:51:48 <yangyape_> vnogin:  https://etherpad.openstack.org/p/freezer-tenant-backup this is the bp, we  view nova and cinder now :)
14:52:40 <vnogin> yangyape_, it's about single tenant. I'm talking about ALL_TENANTS backups :)
14:53:16 <vnogin> yangyape_, something like - "backup everything" button
14:54:12 <yangyape_> vnogin:  everything ? or (all nova and cinder)?
14:54:37 <vnogin> yangyape_, sorry, I mean nova and cinder for all tenants
14:55:22 <slashme> If connected with admin role cinder has a --all-tenant option. I suppose nova has the same and the apiclients do also.
14:55:50 <yangyape_> if all nova and cinder, we need define a mode or a simple parameters  to  gather nova and cinder with nova/cinder api.
14:56:06 <vnogin> slashme, yep, so we going to use it?
14:57:16 <slashme> I guess this is our best solution but nothing is definitive yet.
14:57:47 <vnogin> I see. I think that it's best option for now too
14:59:41 <yangyape_> we need more discuss in  freezer irc
14:59:48 <yangyape_> :)
14:59:54 <vnogin> :) agree
15:00:10 <vnogin> let's continue there
15:00:25 <yangyape_> slashme: :)
15:00:44 <slashme> #endmeeting