Tuesday, 2016-03-01

*** dshulyak has joined #openstack-solar00:28
*** dshulyak has quit IRC00:33
*** angdraug has quit IRC01:20
*** salmon_ has quit IRC02:34
*** dshulyak has joined #openstack-solar07:02
*** salmon_ has joined #openstack-solar08:22
mkwiekpigmej: any tips for reproducing the timestamp in total time?08:47
mkwiekI have problems making it work like you described08:47
*** openstackgerrit has quit IRC08:48
*** openstackgerrit has joined #openstack-solar08:48
mkwiekor rather - not work :D08:52
mkwiekI mean your comment in https://review.openstack.org/#/c/286040/108:52
pigmejmkwiek: just use test script09:21
mkwiekpigmej: which test script?09:21
pigmejbash utils/jenkins/run_hosts_example.sh09:22
pigmejit happens always there09:23
salmon_dshulyak: hmm, your graph is missing rsync_core_puppet09:23
mkwiekoh, I haven't gotten to running examples yet, I had problems with packer and images09:24
mkwiekneed to investigate what's going on there09:24
pigmejdshulyak: https://review.openstack.org/#/c/285336/09:24
dshulyaksalmon_: sure, rsync should be used from created part09:24
pigmejwhen we will merge it version 8.0 will be "invalid" how hard is to fix your scripts to generate 8.0.0 instead ?09:24
salmon_dshulyak: why? I thought it will be in a graph from nailgun09:25
dshulyaksalmon_: you r right, it would be better to create rsync resource from graph09:29
dshulyakand also genkeys09:29
dshulyakpigmej: lets wait a bit, it is not hard, but better to do it later09:30
pigmejdshulyak: ok, but we need to merge it before releas09:31
salmon_dshulyak: did you try to deploy changes with your patch? For me astute fails with strange error: https://bpaste.net/raw/0bc8a103362b09:37
salmon_without you patch it's working ok09:38
dshulyaksalmon_: my patch rebased on top of some changes in nailgun, and if u want to deploy with it u also need astute part09:38
salmon_dshulyak: ah, ok09:38
dshulyaksalmon_: this one https://review.openstack.org/#/c/281448/09:39
salmon_dshulyak: ok09:39
openstackgerritMerged openstack/solar: Update rpm spec dependency  https://review.openstack.org/28605709:39
salmon_dshulyak: btw. this is what astutes does with shell tasks: https://bpaste.net/show/85ccaa89201209:39
salmon_it creates puppet munifest from it :/09:40
dshulyaki know :)09:40
dshulyakit is because bash was blocking, and puppet is always spawned as daemon process09:40
salmon_I wonder if we can use astute to execute it...09:40
dshulyakwhat do you mean?09:41
salmon_dshulyak: not only that, they also use some variables in commands like CLUSTER_ID09:41
salmon_dshulyak: just thinking...09:41
dshulyaknope, this variable is resolved on the side of nailgun09:41
salmon_ah, one problem less then09:41
salmon_ok, then I need it in a graph09:42
dshulyakthe problematic task is update_nodes_info, because it needs to be called for each new added now on every node09:42
dshulyaksalmon_: can you show me that shell resource u was talking about09:43
salmon_dshulyak: update or upload?09:43
dshulyaksalmon_: i think we will need to trigger both somehow09:45
salmon_dshulyak: I don't see  update_nodes_info task09:46
salmon_dshulyak: my resource is simple as this: https://bpaste.net/show/318fe2108a0909:47
dshulyaksalmon_: https://github.com/openstack/fuel-library/blob/master/deployment/puppet/osnailyfacter/modular/astute/tasks.yaml#L2409:47
salmon_dshulyak: it's upload not update09:47
dshulyaksalmon_: can we merge it?09:48
salmon_dshulyak: this resource?09:48
salmon_one moment09:48
salmon_dshulyak: btw. you need to rebase this one https://github.com/Mirantis/f2s/pull/609:52
salmon_dshulyak: https://github.com/Mirantis/f2s/pull/809:53
salmon_dshulyak: what's tasks_directory?10:16
dshulyaksalmon_: tasks connections and metadata are stored in different data structures to reduce size of data10:17
salmon_dshulyak: rsync_core_puppet is in  tasks_directory, but it's not in tasks_graph10:21
dshulyakthats bad10:22
salmon_yup, and this task is crucial :)10:22
dshulyaksalmon_: are u sure? because i have it in graph10:23
dshulyak{u'type': u'sync', u'requires': [{u'node_id': None, u'name': u'pre_deployment_start'}], u'id': u'rsync_core_puppet', u'required_for': [{u'node_id': None, u'name': u'pre_deployment_end'}]}10:23
salmon_dshulyak: which role did you assigned?10:23
dshulyakit shouldnt depend on role10:24
dshulyakthat would be even bigger problem :)10:24
salmon_dshulyak: hmm, I have base-os role assigned and this is what I have https://bpaste.net/raw/0f1ed5a6b66210:25
dshulyakdo we have noop resource?10:27
salmon_noop? With handler: none?10:28
dshulyakwith noop we can avoid generation of stages10:28
salmon_you can create one10:29
-openstackstatus- NOTICE: Gerrit is going to be restarted due to poor performance10:38
*** ChanServ changes topic to "Gerrit is going to be restarted due to poor performance"10:38
openstackgerritMaciej Kwiek proposed openstack/solar: Timestamp comparison in calculating time fixed  https://review.openstack.org/28604010:44
*** ChanServ changes topic to "#openstack-solar"10:44
-openstackstatus- NOTICE: gerrit finished restartign10:44
pigmejmkwiek: this time works ? :D10:45
mkwiekyeah, it should ;)10:45
mkwieksame issue as with calculating delta10:46
dshulyakalso we are missing override_configuration resource10:46
dshulyaki believe it was added for reexecute_on feature10:46
pigmejmkwiek: I don't get this comparsion with <110:50
mkwiekI wanted to compare these float timestamps to 010:51
mkwiekbut you can't really test floats for equality reliably with "==" or "is"10:52
mkwiekso sometimes "0.0" was not equal to "0.0" and it broke10:52
pigmejwhy not int(float(.. ?10:53
pigmejI mean, this float < 1 looks weird ;D10:53
pigmejit's pretty much the same as int(float) == 010:53
pigmejbut... it looks more strange to me10:53
mkwiekit's another option that came to my mind when salmon_ asked for it in review and I asked which one would you guys prefer :)10:53
mkwiekso sure, I can cast it to int10:54
mkwiekI think it may be more readable anyway10:54
mkwiekwill submit new patchset10:54
pigmejbtw what when there will be very fast task?10:55
pigmejlike noops ?10:55
salmon_dshulyak: is it required for now?10:55
pigmejmkwiek: I think it will break then, isn't it?10:55
mkwiekpigmej: not really10:55
pigmejah timestamp10:56
mkwiekthese are timestamps, so they won't be anywhere near 010:56
mkwiekunless we GO BACK IN TIME!!!10:56
pigmej2037 :P10:56
dshulyaksalmon_: i guess no10:56
openstackgerritMaciej Kwiek proposed openstack/solar: Timestamp comparison in calculating time fixed  https://review.openstack.org/28604010:58
openstackgerritBogdan Dobrelya proposed openstack/solar: Fix riak container nested docker mounts  https://review.openstack.org/28650111:04
dshulyaksalmon_: take a look https://github.com/Mirantis/f2s/pull/6/files, i will alse upload generated tasks from fresh fuel library11:08
dshulyaksalmon_: together with this https://github.com/Mirantis/f2s/pull/6/files - graph should be ok, but it is still disconnected from node11:10
salmon_dshulyak: it's the same link11:11
dshulyaksalmon_: https://github.com/Mirantis/f2s/pull/1011:12
salmon_dshulyak: is it ready to be merged?11:15
dshulyaksalmon_: yes, please merge, it is not intergrated with nodes and fuel_data11:15
salmon_ok, merging11:15
openstackgerritBogdan Dobrelya proposed openstack/solar: Fix riak container nested docker mounts  https://review.openstack.org/28650111:20
dshulyaki will update readme after some work in fuel, so - *fsclient alloc 2 null* will create stage anchors in solar, i think the best would be to connect fuel_data before pre_deployment_start11:21
dshulyak2 is cluster id11:21
dshulyakfsclient assoc 2 2 - wil generate full graph based on fuel tasks, but they are not yet connedted to nodes11:22
dshulyaksalmon_: btw fsclient nodes is working?11:22
dshulyakyeah, it seems to work11:24
*** salmon_1 has joined #openstack-solar11:24
mkwiekcan I take this one now: https://bugs.launchpad.net/solar/+bug/1547537 ?11:24
openstackLaunchpad bug 1547537 in Solar "`solar orch report` yields an error when there are no resources" [Undecided,New]11:24
*** salmon_ has quit IRC11:28
*** salmon_1 is now known as salmon_11:28
pigmejmkwiek: you can take any bug that is assigned to 0.3.011:37
pigmejBUT if you take bug you need to fix it :P so choose wisely :P11:37
mkwieksure, will do :)11:38
openstackgerritBogdan Dobrelya proposed openstack/solar: Describe how to create/use the builder container  https://review.openstack.org/28651711:39
pigmejmkwiek: you could also check: https://review.openstack.org/#/c/286517/11:40
pigmejjust say that it works for you, before build step, then I will merge that :)11:41
pigmejbecause it works for me (which is sadly rare recently) :D11:41
mkwiekok, give me a minute, I will check it out11:42
openstackgerritBogdan Dobrelya proposed openstack/solar: Describe how to create/use the builder container  https://review.openstack.org/28651711:47
pigmejsalmon_: dshulyak Maybe I will release new solar vagrant stuff today ?11:57
pigmejand then centos7 + docker stuff + f2s in next release ?11:58
pigmejmkwiek: ^11:58
pigmejbecause, well, current image doesn't work with master11:58
dshulyakpigmej: i dont mind11:58
salmon_pigmej: go for it!11:59
openstackgerritMerged openstack/solar: Describe how to create/use the builder container  https://review.openstack.org/28651712:05
pigmejyeah always blame others... :P12:05
dshulyakwho knows what is this error - AttributeError: 'CreatedResources' object has no attribute 'db_obj' ?12:58
pigmejoh, you wanted to do something bad with CreatedResources12:58
pigmejyou probably passed it to connect12:58
pigmejor something12:58
pigmejvr.create(...) returns CreatedResources12:58
pigmejit's not single resource12:58
pigmejyou can [int] or [name]12:58
pigmejthe most frequent fuckup I would say d;12:59
salmon_dshulyak: do you know how refresh_on is working? They are not providing tasks there13:04
dshulyaksalmon_: there is configs files in nailgun, and reexecute_on acts like subscription to those config files13:06
openstackgerritMaciej Kwiek proposed openstack/solar: Solar orch report doesn't fail for empty graph  https://review.openstack.org/28657113:07
salmon_dshulyak: I see, thx13:07
dshulyakpigmej: i think i had to update zeromq13:37
pigmejdshulyak: anyway, alexz has some questions about packaging f2s13:37
pigmejmkwiek: vagrant@solar-dev:/vagrant$ solar o report13:50
pigmejNothing to report13:50
pigmejTotal Delta: 0.013:50
pigmejTotal Time: -inf13:50
pigmejrly ?13:50
dshulyaksalmon_: i’ve got this TemplateAssertionError: no filter named 'to_nice_yaml’, do i need to update smth?13:52
salmon_dshulyak: solar13:53
salmon_dshulyak: https://review.openstack.org/#/c/285449/13:53
pigmejdshulyak: that's my awesome hack13:57
pigmejsalmon_: I don't think taht's this one13:57
dshulyaknow it works13:57
salmon_pigmej: https://review.openstack.org/#/c/285449/2/solar/utils.py13:57
dshulyak:( we have problems with anchors13:57
pigmejanchors ?13:58
dshulyaknoop resources13:58
salmon_dshulyak: why?13:58
dshulyakthey should be added to each deployment13:58
dshulyakbecause this is like backbone of workflow in fuel14:00
dshulyakand we cannot stage and commit them :) because each next time graph will be disjoined14:00
mkwiekpigmej: I wasn't sure if we want to leave it, but I guess not. You are right, I should have removed it.14:04
pigmejoh cmon we shouldn't display wrong information :D14:04
pigmejif it would say 0 maybe :P14:04
pigmejbut -inf ?:D14:04
openstackgerritMaciej Kwiek proposed openstack/solar: Solar orch report doesn't fail for empty graph  https://review.openstack.org/28657114:10
dshulyakand also we dont have a way to pass timeouts and retries from resource to orchestration :(14:20
salmon_dshulyak: https://github.com/Mirantis/f2s/pull/1114:25
salmon_dshulyak: why do you want to pass it to orchestration? Can not we use it in resource?14:25
salmon_or, handler14:25
mkwiekHey guys, can I mark this bug https://bugs.launchpad.net/solar/+bug/1546497 as "Fix committed" as this patch was merged in https://review.openstack.org/#/c/283054/3 ?14:28
openstackLaunchpad bug 1546497 in Solar "Relax solar requirements" [High,Confirmed] - Assigned to Maciej Kwiek (maciej-iai)14:28
dshulyaksalmon_: yes, maybe it would be better to use them in handler, but still we dont have any interface for that14:29
pigmejmkwiek: yeah14:30
salmon_dshulyak: not yet!14:30
mkwieksalmon_: can you shed some light on exact scenario to reproduce this one: https://bugs.launchpad.net/solar/+bug/1549312 ?14:45
openstackLaunchpad bug 1549312 in Solar "AttributeError: PENDING_RETRY" [Undecided,New]14:45
dshulyaksalmon_: i added nodes and role_data integration https://github.com/Mirantis/f2s/pull/12/files14:46
salmon_mkwiek: "When running solar orch stop"14:49
mkwieksalmon_: do you get it every time you run this, without even providing uid?14:49
pigmejmkwiek: when you not provide uid then you will get 'last' from file14:50
salmon_mkwiek: I tried it once, maybe with uid. I don't remember :(14:50
salmon_mkwiek: variable is undefined there, I checked the code14:50
mkwiekpigmej: I get this when I run just `solar orch stop`14:51
mkwiekUsage: solar orch stop [OPTIONS] UID14:51
mkwiekError: Missing argument "uid".14:51
pigmejthen you can submit PR that eanbles 'last' :)14:52
pigmej(you just need to make uid a special type)14:52
salmon_dshulyak: did you deploy something with it?14:52
dshulyaksalmon_: not yet14:53
dshulyaki also found bug in graph building14:54
mkwiekpigmej: 'last' from what file? I don't follow14:55
pigmejmkwiek: that 'last' is a bit hacky I would say, we inspect when you craete new uid via CLI (and only there), then we store it in file14:55
pigmejand when you use command that requires uid, we usually (!) default that argument to last, then we have some special click type which translates 'last' to correct uid14:56
salmon_mkwiek: so I got with provided uid14:56
salmon_dshulyak: I will play with it today later14:58
mkwiekpigmej: ok, I think I understand. I will check how it's implemented in other commands14:59
pigmejyeah but that's just start :) because then you need to switch to original bug; P14:59
mkwieksalmon_: is PENDING_RETRY a state that we want to have in code? or was it dropped?15:09
pigmejmkwiek: it should be a question to dshulyak15:09
mkwiekdshulyak then: is PENDING_RETRY a state that we want to have in code? or was it dropped?15:09
salmon_yup, dshulyak knows :)15:10
mkwiekI gotta go now guys, I will ask dshulyak tomorrow :)15:11
mkwiekhave a nice afternoon, see you tomorrow :)15:12
dshulyakmkwiek: i think it used in scheduler module15:16
pigmejdshulyak: it doesn't exists :D15:16
pigmejthere is only one occurence of PENDING_RETRY in whole code15:17
dshulyakit should be ERROR_RETRY, it is not fixed in soft_stop15:18
*** salmon_ has quit IRC15:18
openstackgerritDmitry Shulyak proposed openstack/solar: Insert missing events during build of staged graph  https://review.openstack.org/28666515:44
pigmejdshulyak: what are these 'missing events' ?15:44
dshulyaki dont how to anwer your question :)15:45
pigmejhow do you catch it, and where is bug for it ?15:45
dshulyaki will do that stuff later15:46
pigmejbtw dshulyak that's a neat bug.. ;/15:47
pigmejanyway, I need to test changes from bogdando I may be logged out, I will be back if so :D15:54
*** salmon_ has joined #openstack-solar17:42
*** openstackgerrit has quit IRC18:18
*** openstackgerrit has joined #openstack-solar18:18
openstackgerritArtem Silenkov proposed openstack/solar: solar.spec install path fixed  https://review.openstack.org/28676218:31
*** evgenyl_ has joined #openstack-solar18:31
*** evgenyl has quit IRC18:37
*** evgenyl_ is now known as evgenyl18:37
*** openstack has joined #openstack-solar19:15
*** openstackstatus has joined #openstack-solar19:17
*** ChanServ sets mode: +v openstackstatus19:17
*** openstackgerrit has joined #openstack-solar19:19
*** ikalnitsky has quit IRC19:25
*** ikalnitsky has joined #openstack-solar19:31
openstackgerritMerged openstack/solar: solar.spec install path fixed  https://review.openstack.org/28676220:12
*** dshulyak has quit IRC21:43
*** dshulyak has joined #openstack-solar21:44
*** dshulyak has quit IRC21:48
*** dshulyak has joined #openstack-solar22:14
*** dshulyak has quit IRC22:19
*** angdraug has quit IRC22:36

Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!