Thursday, 2016-03-03

*** salmon_ has quit IRC01:22
*** openstackgerrit has quit IRC01:47
*** openstackgerrit has joined #openstack-solar01:48
*** ogelbukh has quit IRC03:26
*** ogelbukh has joined #openstack-solar05:27
*** dshulyak has joined #openstack-solar06:42
openstackgerritDmitry Shulyak proposed openstack/solar: Insert missing events during build of staged graph
*** salmon_ has joined #openstack-solar08:31
salmon_dshulyak: hi, we need to regenerate resources. This file moved from astute dir to roles
salmon_dshulyak: and maybe it would be better to generete it on the iso and not store in repo?08:43
dshulyaksalmon_: hi, i can remove it from repo, i guess you can use f2s t2r to generate it even now08:44
salmon_dshulyak: yeah I can, but I wanted to know your opinion :)08:45
dshulyaksalmon_: so was your deployment successful ?08:54
salmon_dshulyak: controller+compute+cinder, I had to update action path in db for this one resource, and I still have problem with permissions for /var/lib/glance/node. I fixed this two problems manually and it succeed :)08:56
dshulyaksalmon_: graph is connected?08:57
salmon_it seems so08:57
salmon_dshulyak: I tested it with your patch08:57
openstackgerritMaciej Kwiek proposed openstack/solar: Solar orch report doesn't fail for empty graph
pigmejhey guys, what's up ?09:19
openstackgerritJedrzej Nowak proposed openstack/solar: Add /etc/solar/config.yaml as first file in config
*** openstackgerrit_ has joined #openstack-solar09:53
salmon_+3 :)09:54
pigmejI'm reviewing stuff that we need for packages so... ;)09:56
openstackgerritMaciej Kwiek proposed openstack/solar: Solar orch report doesn't fail for empty graph
*** openstackgerrit_ has quit IRC10:02
openstackgerritMerged openstack/solar: Solar orch report doesn't fail for empty graph
openstackgerritMerged openstack/solar: Insert missing events during build of staged graph
pigmejdshulyak: take look at this very complex PR :)10:39
pigmejthx ;)10:45
openstackgerritMerged openstack/solar: Add /etc/solar/config.yaml as first file in config
dshulyaksalmon_: btw i think i will drop vrs directory10:57
salmon_dshulyak: +110:57
pigmejespecially that it should say crs :P10:57
pigmejdshulyak: when we can merge: ?11:17
dshulyakpigmej: actually there was ‘8.0’ for all resources, i think we had some misunderstanding when u asked this11:18
dshulyaki made it ‘8.0.0’, but thats another story11:19
openstackgerritAleksey Zvyagintsev proposed openstack/solar: Add default solar.yaml config
pigmejdshulyak: you mean?11:22
pigmej8.0.0 will be ok11:23
dshulyakit was 8.0 already11:23
dshulyakas a string11:23
azvyagintsevdshulyak: you drop recources and vrs, but forgot to drop them from spec
azvyagintsevrpm build will fail11:56
azvyagintsevwill you fix it? or should i ?11:56
dshulyakazvyagintsev: i will fix11:57
dshulyakazvyagintsev: thanks11:57
openstackgerritJedrzej Nowak proposed openstack/solar: Added systemd solar-worker.service
*** openstackgerrit has quit IRC12:03
*** openstackgerrit has joined #openstack-solar12:03
openstackgerritMerged openstack/solar-resources: Add spec for rpmbuild solar-resources.spec
openstackgerritMaciej Kwiek proposed openstack/solar: Remove repository if adding contents raised error
mkwiekpigmej: can you check out if above commit is a proper implementation for the issue with creating repositories?12:07
pigmejmkwiek: no12:09
pigmejthat was not the idea12:09
pigmejyou should operate on tmp stuff12:09
pigmejand then do atomic operation that will bring it back12:09
pigmejyou have no guarantee that rmtree will succeed12:09
pigmejyou should add repo to temp directory which will be ignored by repository module, and then at the end rename it12:10
pigmejI mean that rm could say but it could clean only these 'temp'12:11
mkwiekI see, I thought that tmp won't be necessary, because if rmtree on the directory that we created a moment ago failed, we have bigger problems than broken repo, but I guess it will add another layer of security there12:14
mkwiekI will add this stuff, thanks!12:14
pigmejwell, you could crtl+C12:15
pigmejor just process died12:15
pigmejalso bp is named "atomic"12:16
dshulyakpigmej: do you know if we have simple api to check if resource exists?12:31
dshulyaki mean without DBLayerNoFound12:31
pigmejwell, why do you want to do such check ?12:33
openstackgerritAleksey Zvyagintsev proposed openstack/solar: Add default solar.yaml config
dshulyakpigmej: to create resource if it doesnt exist12:35
dshulyakotherwise update it12:35
pigmejrace conditions everywhere then...12:35
dshulyakthats one of the most basic operations...12:36
dshulyakwhy do i, as a user, care about those conditons12:36
dshulyakalso i am working with single process12:36
pigmejI know12:36
pigmejjust wanted to mention it12:36
dshulyakok, we dont have :)12:36
pigmejyup we don't have any "exists" just because it's "unsafe"12:37
dshulyakit basically means that i will have this in client applications12:37
pigmejhmm but also we have some12:37
pigmejModel.get_or_create which is thing that you added12:38
pigmejwell, you should avoid if exists: update else create() you could assume always "update"12:38
pigmejadding any if exists: update else create() is huge place for race conditions and is usually dangerous in any concurrent / distributed env12:39
dshulyaksalmon_: why do u need id of env in fuel_data ?12:53
salmon_dshulyak: just in case :) If we add manager there we wiill need it12:57
pigmejguys what about this: ? have you time to chcek it ?12:59
pigmej this one :D12:59
openstackgerritJedrzej Nowak proposed openstack/solar: Support permissions, owner and group in naive_sync
pigmejmkwiek: ^13:37
dshulyaksalmon_: take a look please, if it makes sense i will update README14:00
dshulyakazvyagintsev: pleae take a look if i fixed spec correctly -
salmon_dshulyak: hiera patches may be useful later14:05
dshulyaksalmon_: we can restore them from history, but also i am not sure that it is viable approach to run tests on master node to generate those inputs14:06
salmon_dshulyak: I see, ok14:07
dshulyakfirst of all fixtures are not even in library package14:07
dshulyakpigmej: to merge sources patch - we need to adjust fsclient first14:13
dshulyaki will do it later14:13
salmon_dshulyak: I like it
salmon_non fatal error
pigmejdshulyak: k14:18
pigmejguys: "- wrap solar states (error or sucess) so they can be used in Nailgun in context of nodes" what about that ?14:21
salmon_not even started14:25
pigmejguys, can you craete bugs/issues whatever with todo items that exists for experimental f2s "fully" integrated ?14:26
pigmejby "fully" I mean minimal set of functions that enables it.14:26
pigmejor update
openstackgerritMaciej Kwiek proposed openstack/solar: Remove repository if adding contents raised error
pigmejshit I started to write test suite for handlers/transports and well....14:57
azvyagintsevdshulyak: please get back "space" from line 51
azvyagintsevsome builder will fail w\o it..15:08
dshulyaki am not responsible for this, it was my editor decision15:08
openstackgerritMaciej Kwiek proposed openstack/solar: Remove repository if adding contents raised error
*** angdraug has joined #openstack-solar15:31
dshulyaksalmon_: i also added a fix to 20 pull request - in some cases it was possible that dependencies between nodes were absent15:38
salmon_dshulyak: is it ready to be merged?15:41
dshulyaksalmon_: yes, everything looks good15:46
dshulyaksolar-worker consumes too much - 31956 root      20   0  734292  89056  11032 R  94.0  5.8   0:12.47 solar-worker15:48
salmon_maybe because of pigmej's locks implementation :)15:50
pigmejno it should be not15:50
dshulyakazvyagintsev: are u around? confirm please that it is ok
pigmejOR if you deploy the same stuff all over again15:51
pigmejthen it may be possible15:51
dshulyakit mights be database, i know that it is not used there in optimal way15:51
pigmejbecause locks are never cleared, so full locking history is preserved15:51
dshulyaknah, it is first run15:51
pigmejwhat db backend ?15:52
pigmejsqlite ?15:52
azvyagintsev<azvyagintsev> dshulyak: please get back "space" from line 51
azvyagintsevalready :)15:52
azvyagintsevall ok, just move back "space"15:52
dshulyakyes, i added it back15:53
dshulyakpigmej: yes, i am using sqlite, but the problem is that i am fetching all tasks every time, we can avoid that15:54
dshulyakwe need current + childs + parents of those childs15:54
pigmejyaeh for big graph that may be the case15:54
pigmejwe already discussed it I think15:55
pigmejor we could use cache in better way15:55
dshulyakit is not big though… maybe ~150 items15:55
pigmejdepends how many times we're fething it ;D15:56
pigmejI can play with it tomorrow15:57
pigmejI like profilling if so :D15:57
dshulyakatleast 150 + 115:57
dshulyakactually exectly 150 + 115:58
pigmejso 150 * 150 ?15:58
pigmej+ inputs ?15:58
dshulyakwhich inputs?15:58
pigmejah no15:58
pigmejit's graph :)15:58
pigmejI will create big enough graph with noops15:58
pigmejbut I have interview right now ;/15:58
pigmejand we have weekly in 2 minutes...15:58
dshulyaksalmon_: so whats your testing results? any success with /var/lib/glance ?16:01
salmon_dshulyak: still debuging, I also added ceilometer and it also fails :(16:03
dshulyaklooks like we will need to get bogdando tomorrow16:03
*** dshulyak has left #openstack-solar19:01
*** angdraug has quit IRC19:57
*** angdraug has joined #openstack-solar21:21
*** dshulyak has joined #openstack-solar21:52
*** dshulyak has quit IRC22:06
*** salmon_ has quit IRC22:45
*** salmon_ has joined #openstack-solar23:05
*** salmon_ has quit IRC23:09

Generated by 2.14.0 by Marius Gedminas - find it at!