15:01:29 <m3m0> #startmeeting openstack-freezer 13-08-2015
15:01:29 <openstack> Meeting started Thu Aug 13 15:01:29 2015 UTC and is due to finish in 60 minutes.  The chair is m3m0. Information about MeetBot at http://wiki.debian.org/MeetBot.
15:01:29 <freezerBot`> Meeting started Thu Aug 13 15:01:29 2015 UTC and is due to finish in 60 minutes.  The chair is m3m0. Information about MeetBot at http://wiki.debian.org/MeetBot.
15:01:29 <freezerBot`> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
15:01:29 <freezerBot`> The meeting name has been set to 'openstack_freezer_13_08_2015'
15:01:31 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
15:01:33 <openstack> The meeting name has been set to 'openstack_freezer_13_08_2015'
15:01:43 <m3m0> All: meetings notes available in real time at: https://etherpad.openstack.org/p/freezer_meetings
15:02:05 <m3m0> fabriziov_ are you around?
15:02:31 <daemontool_> m3m0, I can go first
15:02:50 <m3m0> please daemontool_ go ahead
15:03:14 <daemontool_> so I've been working the lvm auto snap improvement, also the work_dir creation  is now under review
15:03:19 <daemontool_> I'm planning today and tromorrow
15:03:49 <daemontool_> to improve the freezer-agent usability by automatic guessing command line params as the lvm dir mount, snap size, etc
15:04:07 <daemontool_> progress on rsync are going slowly
15:04:15 <daemontool_> no blockers
15:04:22 <daemontool_> that's all from me :)
15:04:47 <m3m0> the auto_snap used to work only in ubuntu right? now it works with hlinux and other distros?
15:04:58 <m3m0> do you need help with rsync?
15:13:22 <m3m0> fabriziov_ you are next
15:13:28 <fabriziov_> ok
15:13:57 <fabriziov_> The integration tests are in review. Darragh added some valid comments. I'm checking them.
15:13:58 <fabriziov_> I think I'm going to have a look at the suggested fixtures library and update the code for the tests. But I think this can wait for another sprint, together with some other tests maybe.
15:14:12 <fabriziov_> The code for apiclient support of keystone has been rebased and Pierre is deploying it so we can test what is the failure
15:15:16 <fabriziov_> I've also to complete the support for falcon-0.1.10. the job.py has been converted, I just need to modify the code for sessions, clients etc
15:15:45 <fabriziov_> no big deal. Updating the unit tests takes the most time :)
15:16:21 <fabriziov_> only blocker is haveing the possibility to test K3 on HLM, but I have other things to do in the meantime
15:16:33 <m3m0> does that refactoring will impact the implementation on windows?
15:17:03 <fabriziov_> no. only the api code is involved
15:17:18 <fabriziov_> I've to modify the code for the middlware
15:17:42 <fabriziov_> but all is api side
15:17:50 <m3m0> which functionality is not present on that falcon version?
15:18:03 <daemontool_> m3m0,  sorry for the delay, yes the autosnap used to work only on ubuntu, yes rsync I have to find the time to work on it...
15:18:15 <daemontool_> I might need some help in a bit....
15:19:11 <fabriziov_> falcon>=0.2.0 support middleware objects that implement the methods:
15:19:18 <fabriziov_> process_request
15:19:23 <fabriziov_> process_response
15:19:36 <m3m0> daemontool_ just let us know what do you need :) I'm a little bit busy but we can have a mini sprint on the weekend for that :)
15:20:25 <daemontool_> haha! great :)
15:20:27 <daemontool_> ty
15:20:49 <fabriziov_> that's all for me.
15:22:15 <m3m0> thanks
15:22:21 <m3m0> reldan are you here?
15:22:30 <reldan> yes, I’m here
15:22:38 <reldan> I'm doing rsync integration. It requires some refactoring. And also I'm trying to unify approach of doing backups and use streams for local and ssh storages as well.
15:23:05 <reldan> After this I should take rsync and integrate to our code
15:23:44 <reldan> I am also trying to throw away “ssh” command invocation and allow to run it for windows as well
15:23:48 <reldan> It is all from my side
15:25:17 <m3m0> if you remove ssh command what will you use? scp?
15:25:56 <reldan> Paramiko. I’m not sure about speed, but Paramiko allows to open remote file throw sftp
15:26:23 <reldan> with self.ftp.open(filename, bufsize=chunk_size) as backup_file:
15:26:24 <reldan> data = backup_file.read(chunk_size)
15:26:25 <reldan> while data:
15:26:26 <reldan> queue.put(data)
15:26:27 <reldan> data = backup_file.read(chunk_size)
15:26:44 <m3m0> that sounds nice, there are ssh binaries for windows for this version
15:26:47 <openstackgerrit> Fabrizio Vanni proposed stackforge/freezer: apiclient uses openstackclient.api  https://review.openstack.org/206104
15:27:00 <m3m0> but is better if we can get rid of those
15:27:30 <reldan> m3m0: I will try to do it. It seems possible - but some guys are complaing about speed
15:27:50 <m3m0> to fast or to slow?
15:28:16 <reldan> m3m0: To slow, but I have idea that probably they have wrong size of buffer
15:28:20 <daemontool_> reldan, my advice would be to focus on performance / limits after the feature is ready
15:28:31 <reldan> Yes, yes sure
15:28:35 <daemontool_> I think the paramiko stuff is good
15:28:39 <daemontool_> it allows us to be more portable
15:28:43 <daemontool_> and to remove the binaries..
15:28:47 <daemontool_> so well done
15:28:48 <reldan> IYes, yes sure
15:29:08 <daemontool_> also
15:29:17 <daemontool_> we'll add the pypy support at some point
15:29:25 <daemontool_> we need that anyway for rsync...
15:29:43 <daemontool_> so with that we might get significan improvements on also the other bottlenecks
15:29:59 <openstackgerrit> Fabrizio Vanni proposed stackforge/freezer: basic integration tests for freezer-agent  https://review.openstack.org/210057
15:31:06 <m3m0> that sounds great, and pypy is available for windows as well
15:31:13 <m3m0> only in 32 bits
15:32:22 <m3m0> do you have anything more to add?
15:32:32 <reldan> I have nothing to add from my side
15:32:57 <m3m0> thanks reldan
15:33:08 <m3m0> Slashme, you're next
15:33:10 <reldan> Thank you m3m0
15:33:18 <Slashme> Not a lot on the public side. Just resolved a bug that was causing the freezer-scheduler not to find the freezerc binary if you were in a non-activated virtualenv. I hope I'll get some time soon to work on automatic devstack integration.
15:34:20 <m3m0> did you commit that right?
15:34:44 <Slashme> Yep
15:34:53 <m3m0> is it the one relate to the paths when you overwrite them ?
15:35:59 <Slashme> Yes, freezer-scheduler uses os.environ['path'], and not sys.path to find the binary
15:36:13 <Slashme> Not it falls back to sys.path
15:36:38 <m3m0> nice, great job, do you have anything more to add?
15:37:03 <m3m0> my turn
15:37:36 <m3m0> I'm writing a windows service in order to have the scheduler running as a "daemon" on windows
15:38:02 <m3m0> I had to modify the behaviour on how its executed on linux in order to make it work on windows
15:38:28 <m3m0> right now I'm debugging why the service stops as soon as I started
15:39:14 <m3m0> for the freezer-agent side, I still have to some tests on the ssh implementation, and I need to do it before the paramiko module gets merged
15:39:28 <m3m0> local and swift storages are working greate
15:39:32 <m3m0> great*
15:39:39 <m3m0> that's it for me
15:39:51 <m3m0> does anyone wants to say anything?
15:40:33 <samuelBartel> nop
15:41:21 <m3m0> sorry samuelBartel didn't see you :(
15:41:29 <m3m0> you are ok then?
15:41:47 <samuelBartel> m3m0, no problem
15:41:56 <samuelBartel> finish the compression algo bp
15:42:32 <samuelBartel> and will look in launchpad today for a new task
15:42:45 <m3m0> nice :)
15:43:26 <m3m0> thanks a lot everyone
15:43:35 <m3m0> #endmeeting