14:02:37 <szaher> #startmeeting freezer
14:02:38 <openstack> Meeting started Thu Nov 10 14:02:37 2016 UTC and is due to finish in 60 minutes.  The chair is szaher. Information about MeetBot at http://wiki.debian.org/MeetBot.
14:02:39 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
14:02:42 <openstack> The meeting name has been set to 'freezer'
14:02:50 <szaher> Hello everyone
14:03:00 <szaher> let's wait 5 minutes for everyone to join the meeting
14:03:00 <yangyapeng> hello szaher
14:03:08 <szaher> Hi yangyapeng
14:03:19 <zhusz> Hi szaher
14:03:29 <szaher> The list of topics we have for today to be discussed are here https://etherpad.openstack.org/p/freezer_meetings
14:03:30 <neilus_> hi guys
14:03:36 <szaher> Hi zhusz
14:03:40 <szaher> Hi neilus_
14:04:39 <szaher> You can find here a list of useful links about freezer https://etherpad.openstack.org/p/freezer_usefull_links
14:09:01 <szaher> Ok guys, Let start
14:09:18 <szaher> #topic Stateless use of freezer
14:09:32 <szaher> saggi are you around ?
14:11:00 <szaher> anyone knows what the idea behind stateless freezer ? or what does it mean ?
14:12:01 <szaher> According to the description in freezer_meetings pad they want to run freezer without any permanent jobs or actions.
14:12:48 <szaher> when you install freezer you don't get any default job or action ... I think this should be Ok ?
14:12:54 <szaher> anyone has any ideas ?
14:13:37 <yangyapeng> karbor  commit  this topic ?
14:15:10 <szaher> yangyapeng: according to freezer_meetings yes, they want to have this feature
14:15:42 <yangyapeng> xingxinyong are you around?
14:16:30 <zhusz> I think it's OK.
14:17:40 <szaher> Ok ... I think we need more elaboration from the karbor guys around this topic
14:17:55 <zhusz> Maybe We need check more detailed about Karbor on this.
14:18:03 <szaher> anyone has any thoughts about this  or we move forward ?
14:18:21 <yangyapeng> hi saggi
14:18:27 <saggi> Hi, sorry, in my calendar it says the meeting is in an hour
14:18:47 <yangyapeng> saggi:  welcome
14:19:17 <szaher> saggi: Hi saggi no problem. We changed the meeting time to be 2 UTC after changing the time
14:19:48 <szaher> Can you please explain us what do you guys mean by stateless freezer ?
14:19:54 <saggi> In a nutshell, we would like to be able to run freezer without setting anything persistent in freezer. Since Karbor want's to do the book keeping.
14:20:14 <saggi> So we want to send all the options to what plugins we use on what VM on what node and have it just run
14:20:28 <saggi> without first setting up a plan in freezer
14:21:43 <saggi> Am I being clear?
14:22:06 <szaher> saggi: Well. I did get some but didn't get the full picture here.
14:22:23 <szaher> When we install freezer, we have 3 components at least
14:22:49 <szaher> freezer, freezer-api, web-ui
14:23:34 <szaher> freezer (agent, scheduler) we need only os credentials to run and take backups ...
14:24:02 <szaher> freezer-api we need os credentials, and elasticsearch info to store jobs and actions
14:24:24 <saggi> What about what plugins to use?
14:24:41 <szaher> which plugins ?
14:24:50 <saggi> Also from what I understand elasticsearch would stop being mandatory for next version
14:24:51 <szaher> do you mean fs, nova,cinder,mysql ?
14:24:52 <yangyapeng> saggi: you mean ,want to make  freezer   as a plugin in karbor ,do a packaging in karbor?
14:25:35 <saggi> From what I gathered from the roadmap in BCN you intend to allow more pluggability in the freezer workflow
14:25:47 <saggi> and remove the dependency on elasticsearch
14:25:54 <szaher> saggi: Yes, that's true
14:26:24 <szaher> we will have pluggable engines, apps, storages
14:26:31 <saggi> exactly
14:26:37 <saggi> Can I select what plugins to use when I start the backup itself
14:26:39 <szaher> we do have some kind of pluggable storage now and mods as well
14:26:46 <saggi> or do I need to do it before hand
14:26:48 <szaher> Yes, you can
14:27:41 <saggi> Also, if we don't want scheduling, can we not have a persistent running freezer agent on compute nodes
14:27:42 <szaher> you can use --mode or -m
14:28:40 <szaher> you can run freezer-agent without freezer-scheduler but in this case freezer-api will be useless and you won't get any info about the backups
14:29:24 <szaher> you can fire freezer-agent and do backup/restore you data but you wont be able to list backups/jobs/actions
14:29:26 <saggi> Isn't the state saved in the DB anyway?
14:29:44 <saggi> Can't I query the DB?
14:29:48 <szaher> The only component that communicates with db is freezer-api
14:30:07 <szaher> neither freezer-agent nor scheduler do query the db
14:30:43 <saggi> So the freezer agent send information to freezer-api through the message bus?
14:31:02 <saggi> about progrees?
14:31:20 <szaher> saggi: freezer-agent doesn't talk to freezer-api.
14:31:43 <saggi> How does the progress information get to the api
14:31:59 <szaher> saggi: freezer-scheduler is the one who communicates freezer-api using api-client and updates freezer-api with the status of job/action/backups
14:32:23 <szaher> saggi: freezer-scheduler is the intermediate point between freezer-api and freezer-agent
14:33:11 <saggi> szaher: I understand
14:34:18 <saggi> Can I live with only one scheduler for a cluster
14:34:59 <szaher> saggi: No, scheduler is fetching the agent by subprocess so they need to be on the same host
14:35:51 <saggi> szaher: It's a question of deployment
14:36:48 <szaher> to be clear if you want to schedule backups for a certain host, you need to have freezer-scheduler and freezer-agent running on this host
14:37:51 <saggi> Are there plans to not have this restriction
14:38:08 <saggi> So we can ssh and run the agent or something similar
14:39:02 <szaher> saggi: we have some thoughts about remote backups or agentless backups where you can ssh and fetch the backup remotely or using any different way to take the backup
14:39:36 <saggi> szaher: I like those thoughts :) How concrete are they?
14:40:59 <szaher> saggi: they are low priority for the time being as we need to finish the refactoring and plugin layer first then we can move forward with agentless backups as an agent with some modes or apps
14:41:30 <szaher> if you want we can have talk more about in freezer room
14:42:39 <saggi> szaher: OK, I'll just some up what I understand and then I'll stop for this meeting.
14:42:39 <saggi> I should be able to do any style of backup without scheduling. Running agentless is something that might be in the future.
14:43:08 <szaher> saggi: Yes
14:43:27 <saggi> szaher: Thanks!
14:43:46 <szaher> saggi: Just keep in mind if you take backups without scheduling you won't get any metadata about backups and you won't be able to see those backups in the db or api
14:44:34 <szaher> for agentless backup, yes it's in future
14:47:04 <saggi> szaher: What kind of metadata
14:50:05 <szaher> saggi: it backup information like which host, backup name, was it incremental, engine used and some other stuff that help
14:52:24 <szaher> saggi: If you use only freezer-agent without scheduler you will still be able to backup and restore your data, but you wont be able to list backups for example
14:52:55 <saggi> szaher: That's OK as long as it's information I already know so we can save it in the Karbor DB instead
14:54:18 <szaher> saggi: somehow you can do it, reach me if you need any help with that
14:54:36 <szaher> anymore questions ?
14:55:01 <szaher> we are running out of time. anyone wants to add anything ?
14:55:13 <zhusz> a question only.
14:55:37 <szaher> zhusz: go a head
14:55:44 <zhusz> If the backup process is interupt, we can resume from break point?
14:56:18 <szaher> zhusz: how did you stop the backup job ?
14:56:51 <zhusz> e.g. Network issue
14:56:53 <szaher> zhusz: anyway, No you need to start again even the cleanup is manual, so you have to cleanup yourself then start the backup job again
14:57:22 <yangyapeng> it is in memery
14:57:29 <zhusz> Ok, I understand.
14:57:49 <zhusz> Thanks szaher
14:57:56 <szaher> zhusz: welcome :)
14:58:02 <szaher> anymore questions ?
14:58:11 <yangyapeng> 2 minutes
14:58:36 <szaher> guys please we need to finish the pending reviews upstream. Please try to review all patches
14:58:55 <szaher> Thank you guys. Have a great day :)
14:59:03 <yangyapeng> thank you szaher
14:59:08 <zhusz> Thanks
14:59:13 <szaher> #endmeeting