15:06:09 <vannif> #startmeeting freezer-meeting_2015-10-22 15:06:09 <freezerBot`> Meeting started Thu Oct 22 15:06:08 2015 UTC and is due to finish in 60 minutes. The chair is vannif. Information about MeetBot at http://wiki.debian.org/MeetBot. 15:06:09 <freezerBot`> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 15:06:09 <freezerBot`> The meeting name has been set to 'freezer_meeting_2015_10_22' 15:06:09 <openstack> Meeting started Thu Oct 22 15:06:09 2015 UTC and is due to finish in 60 minutes. The chair is vannif. Information about MeetBot at http://wiki.debian.org/MeetBot. 15:06:10 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 15:06:13 <openstack> The meeting name has been set to 'freezer_meeting_2015_10_22' 15:06:24 <vannif> hello everyone 15:08:18 <vannif> marzif, would you start ? 15:08:41 <marzif> ok 15:09:10 <marzif> I've been working on switching to testr from pytest, that still needs to be done on freezer-api and freezer-web-ui 15:09:19 <marzif> it's done for freezer repo 15:09:31 <marzif> then we need to convert our tests that uses pytest monkeyaptch 15:09:47 <marzif> to use mock or what is used in the other os projects 15:10:02 <marzif> also I'm helping vannif with the integration tests, doing code reviews 15:10:20 <marzif> if I can finish this asap, I'll restart to the block based incremental integration with the current code base 15:10:36 <marzif> soon we'll have a freezer logo also :) 15:10:47 <vannif> woohoo! :) 15:11:12 <marzif> we are not in the openstack big tent 15:11:39 <marzif> we probably should propose freezer as integrated project by end of the year, so we need to to many thinkgs like itnegrates testing 15:11:42 <marzif> oslo.log 15:12:00 <marzif> place the right code on a python-freezerclient and so on 15:12:07 <marzif> this is on the general roadmap... 15:12:13 <marzif> that's all from me :) 15:12:38 <vannif> yes. I think we should define clearly the various things needes and give priorities. 15:12:52 <vannif> *needed 15:13:03 <marzif> yes 15:13:29 <vannif> I'm not a huge expert on mock, but if if I can help ... maybe I can learn something new too :) 15:14:45 <vannif> good. 15:14:49 <marzif> I think we all have to use the same framerowk for testing 15:15:01 <marzif> and taht framework for unittesting has to be the one most osed in os 15:15:06 <marzif> s/osed/used/ 15:15:20 <vannif> yes definitely 15:15:48 <vannif> and everyone should care about writing tests and avoid lowering the coverage when he submits patches 15:15:58 <marzif> +1 15:16:13 <vannif> otherwise we delegate to someone the specific task to write the tests ... It's a solution .. 15:16:46 <vannif> anything more to say ? 15:17:38 <marzif> nope 15:17:48 <marzif> delegate to who ? :D 15:18:03 <marzif> each engineer has to write the unittests for the code he/she commit 15:18:04 <marzif> :) 15:18:26 <vannif> to a convict :) 15:18:30 <marzif> haha 15:18:40 <vannif> ok. thank you, marzif 15:18:46 <vannif> reldan ? 15:19:22 <reldan> Yes, I was doing coverage this week and refactoring of ssh and local - now they are using much more common codebase 15:19:43 <reldan> After switching to pbr and testr I have some problems with running coverage locally 15:20:19 <reldan> I have checked nova backup and it doesn’t work with hpcloud, but I have installed kilo with devstack and testing it against kilo 15:20:47 <marzif> reldan, does it work? 15:20:57 <reldan> Yesterday we have had a conversation about refactoring swift storage and usage swiftclient.put_object instead of uploading chunks 15:20:58 <marzif> or you didn't get the time to test it yet? 15:21:42 <reldan> It has some issues like race conditions, probably I will be able to fix it be inserting timeouts 15:22:29 <reldan> I also removed 300 lines of code from tests common 15:23:10 <marzif> mmhhh.... race conditions can be there even if we add timeouts... 15:23:15 <reldan> I also have fixed bug with swift chunk_size, and it should work now 15:23:54 <reldan> Sure, we can create integration tests probably 15:23:57 <reldan> for nova backup 15:24:05 <reldan> As we have it now for swift storage 15:25:38 <reldan> I have nothing to add what I’m doing now. I would like to have performance measurement tool and charts, real data continious tests and improved logging. 15:26:59 <reldan> It is all 15:27:35 <vannif> yes, the need for performance testing is increasing 15:28:10 <vannif> it will become the next big goal 15:28:17 <vannif> for testing 15:28:21 <reldan> Yes, I would like something like metrics in java. I would like to mark a block of code as critical and hava logging with time of executing in distributed log or chards 15:28:25 <reldan> charts 15:29:05 <vannif> regarding race condition, I'm dubious as well about solutions which involve "timing" 15:29:19 <reldan> And it would be great to have history of previous runs 15:29:26 <vannif> It's not a trivial task, I know 15:29:35 <reldan> As I am, but when I’m doing image.get(id) 15:29:50 <reldan> sometime it returns me Image: {Propreties:{}} 15:30:07 <reldan> And when I’m trying to update image.get - it throws exeption 15:30:48 <reldan> And it’s really tricky to catch things like that, because novaclient, cinderclient can return unexpected something 15:31:24 <reldan> The only way to have something more-less stable - it runs continiously nova/cinder backups against different installations of openstack 15:31:30 <reldan> It’s my opinion 15:32:06 <reldan> So for my particular case: while not image.properties: sleep; recheck; 15:32:53 <marzif> sounds reasonable 15:33:34 <reldan> but actually in case of nova we have a lot of different scenarious 15:33:51 <vannif> yes, at some point you have to deal with the unexpected :) 15:33:54 <reldan> let’s imageine - I would like to make an image, and I exceed images limit in my project 15:34:17 <reldan> I even don’t know what novaclient return me in that case 15:35:01 <reldan> yes, agree 15:36:02 <vannif> you said you also have problem with unit tests after the switch to testr ? 15:36:41 <reldan> Yes, locally It shows no coverage, so I am using review for checking coverage now 15:36:56 <reldan> Probably it is a Mac specific problem 15:39:39 <reldan> I have nothing to add now 15:40:06 <vannif> ok. thank you reldan 15:40:14 <reldan> Thank you vannif! 15:40:39 <vannif> On my side. 15:41:29 <vannif> the devstack plugin are merged and working. I need to finalize the execution of the integration tests on the CI toolchain 15:42:49 <vannif> I'm still not such a big expert on that. thanks marzif for the help :) 15:43:57 <marzif> vannif, do you recall what's the exact command to execute only the integration tests? 15:44:32 <vannif> no. when I run the tests locally I just run pytest 15:44:53 <vannif> the tests are run automatically if the correct environ vars are set 15:45:23 <marzif> in testr what would be the command? 15:45:39 <vannif> besides that, I've tested in a real hlm deployment, and it was a good thing :) 15:46:05 <daemontool_> ok 15:46:26 <vannif> not yet looked into it. I'll reach you asap with expanded knowledge on the topic 15:46:55 <vannif> the last part took more time than expected indeed 15:48:35 <marzif> ok as soon as you can please help me with the exact testr command 15:48:35 <vannif> I've also planned some "hidden" behaviors of the api. that is: it will extract the actions from a newly created job and add them to the action index. so that jobs created from the command line become editable from the web-ui 15:48:52 <marzif> ok 15:48:56 <marzif> that is important 15:49:02 <vannif> yes 15:49:41 <vannif> since the web-ui is improving, we need to keep it "in sync" with the rest 15:49:53 <marzif> yes 15:49:56 <vannif> that is all from me 15:50:02 <marzif> we need to include also the web ui 15:50:08 <marzif> in the integrated gate tests 15:50:22 <vannif> the devstack plugin is working for the web-ui as well 15:50:50 <vannif> I mean: you can have a devstack machine with a working freezer/freezer-api/freezer-web-ui environment 15:51:14 <marzif> yes, but we need the integration tests for the web ui 15:51:20 <vannif> but I don't know how to actually do integration tests on a weg gui 15:51:21 <marzif> and afaik we don't have them now 15:51:31 <marzif> that's m3m0 job :D 15:52:02 <vannif> I think they involve a tool do actually open a browser and "click" and enter text 15:52:10 <vannif> s/do/to 15:52:23 <marzif> like selenium 15:52:24 <vannif> yes :) 15:52:31 <vannif> exactly 15:52:36 <marzif> we need something that render java script 15:53:39 <vannif> I'll ping memo for a crash course on web ui testing :) 15:53:42 <marzif> ok 15:54:55 <vannif> any questions ? 15:55:22 <vannif> szaher, are you there ? 15:55:39 <szaher> Yes 15:55:45 <sc> vannif: get horizon code and look at it, I bet they are doing unit test 15:56:25 <szaher> I was working on the oslo.log and oslo.config in freezer-api 15:56:35 <vannif> sc, yes. that's a good starting point. 15:56:44 <vannif> yes szaher, sorry 15:56:47 <vannif> go on 15:56:50 <szaher> np 15:57:20 <szaher> I did it with freezer-api and changed most of the files to use oslo.log instead of pythong logging 15:57:40 <szaher> I was working today on freezer-db-init script to move it to oslo_config and oslo_log 15:58:04 <daemontool_> sc, brilliant 15:58:51 <szaher> I need to change some files for freezer-db-init script and will commit all together 15:59:45 <szaher> I did a quick commit to fix the requirements of freezer-api find it here https://review.openstack.org/#/c/237514/ 16:00:17 <szaher> to separate it from logging and config because it may take sometime for logging and config to be merged 16:03:39 <vannif> are you going to send something for review ? 16:04:15 <szaher> Yes, if I finished freezer-db-init tomorrow will send everything together if not will be early next week 16:06:11 <vannif> you removed the freezer-db-init script and embedded everything in the freezer-api command ? 16:07:39 <szaher> No, I changed freezer-db-init to use oslo.config and oslo.log 16:08:02 <szaher> before it wasn't using any kind of logging, so now will be using oslo.log 16:08:20 <szaher> and oslo.config to get configs from either file or cli 16:08:58 <szaher> I was thinking of using one config file for api and db-init like all os projects 16:09:25 <szaher> we could have a database or connection section in the configuration file and we use this section to store information related to backend 16:09:29 <vannif> one for each ? 16:10:59 <szaher> whatever we decide we can use one file with 2 diff sections, for ex. DEFAULT will be for freezer-api and [connection] for freezer-db-init 16:12:22 <vannif> what are the settings which are specific of each program and are not shared ? 16:13:28 <szaher> the same existing one 16:16:49 <vannif> sorry. I don't get it. what are the specific setting needed by the freezer-db-init which are not needed by the freezer-api ? 16:23:52 <szaher> something like mappings, confirmation and so on 16:24:42 <openstackgerrit> Memo Garcia proposed openstack/freezer-web-ui: Refactoring api.py https://review.openstack.org/236175 16:24:51 <vannif> ok 16:25:41 <vannif> remember to update the unit tests :) 16:27:16 <vannif> I'll take some time to review it and we'll got through it together 16:27:23 <vannif> in case 16:27:30 <vannif> anything else to say ? 16:28:13 <szaher> No Thanks 16:28:22 <vannif> thank you szaher 16:28:24 <vannif> I think no one else is around. m3mo is busy 16:28:51 <vannif> if anybody wants to add something ... 16:31:28 <vannif> ok. thank you all 16:31:38 <vannif> #endmeeting