Tuesday, 2016-02-23

openstackgerritAnnie Lezil proposed openstack/solar: Fixed alignment issues in the text  https://review.openstack.org/28332600:18
*** tzn has joined #openstack-solar01:25
*** tzn has quit IRC01:30
*** tzn has joined #openstack-solar02:44
*** tzn has quit IRC02:48
*** tzn has joined #openstack-solar07:22
*** tzn has quit IRC07:27
*** dshulyak has joined #openstack-solar07:57
mkwiekhello guys! I have a question: where do we keep db config in solar? where does the "riak://" come from?08:11
mkwiekalso, a question about this bug: https://bugs.launchpad.net/solar/+bug/1545813 "When worker has different solar_db than cli". Does it mean that worker has correct db string, which it can use to connect to db, but it's different than the one that regular solar cli uses?08:15
openstackLaunchpad bug 1545813 in Solar "Better error reporting in solar-worker when database is wrongly configured" [Undecided,Confirmed] - Assigned to Maciej Kwiek (maciej-iai)08:15
mkwiekis this even an error? I mean, how can we know while starting solar-worker what db string will cli use?08:16
mkwiekpigmej: ^^08:20
dshulyakmkwiek: hey, take a look at L56 in config.py, there can be several overrides, in vagrant environment relevant config stored in /.solar_config_override08:27
mkwiekdshulyak: thanks!08:29
dshulyaki think the first thing we need to do is to log database configuration, i even think that i was adding this logging somewhere but it is not commited, the second problem is that worker doesnt fail when graph is not written to database08:30
mkwiekdshulyak: when should graph writing occur?08:37
dshulyakwriting will occur either when you type - solar or create or solar ch process, but if cli will use different config file (different db connection)…08:40
pigmejhey guys :)08:41
pigmejmkwiek: you can export SOLAR_DB=....08:41
pigmejand check yourself :)08:41
pigmejdshulyak: have you seen: https://review.openstack.org/#/c/283155/2/solar/dblayer/test/test_real.py ?08:44
dshulyakpigmej: why is it non-working?08:45
pigmejbecause we (I) added resolver for bucket08:46
pigmejso whole resolving is hidden08:46
dshulyaki forgot08:46
pigmejme too :D08:46
pigmejand riak client dislike how mock works so I had to do ugly workaround ;/08:46
dshulyak1. will we propagate inputs from higher-level composition to lower-level? 2. do we support AND lookups with tags?08:56
pigmejdshulyak: ?08:56
dshulyakwhich part?08:56
pigmejdshulyak: :>08:59
dshulyakyes, one sec09:00
dshulyakpigmej: http://paste.openstack.org/show/487855/ this is two separate vrs09:01
dshulyakcompositiions of whatever09:01
pigmejyeah and the question is about ?09:01
dshulyak1st is like collection09:01
dshulyakwill lower-level composition will see params that we will pass to higher-level one?09:02
pigmejafair no09:03
pigmej+ https://github.com/openstack/solar-resources/blob/master/templates/node/1.0.0/node.yaml09:03
pigmejwe have isolation of inputs passed to separate VRS09:04
pigmejComposer thigns*09:04
dshulyaki’ve seen examples with input declaration09:04
dshulyakok, i guess i will just declare them as well09:04
dshulyakwhat about 2nd ?09:04
pigmejafair the idea  was to separate namespaces, I mean each resource has it's own namespace there09:04
dshulyakwe do support AND lookups?09:04
pigmejwhat do you mean by that?09:04
pigmejLukasz improved Ply thingy for complicated queries09:05
dshulyakwith_tags: [resource=deploy_start, 'node={{node}}']09:05
dshulyakis it via ply or it can be done on a backend?09:06
dshulyakply means we need to load a lot of things09:06
pigmejthe good part is that it's just index scan, and it's fast enough it most cases09:06
pigmejwe return all candidates ids, then filter them, and then fetch only these which are needed09:07
pigmejwe can't do AND queries on backend level09:07
pigmejwe COULD provide optional functions per backend though09:07
dshulyakok, i just spotted that there will be a lot of them, we will try and then decide09:08
pigmejwe have that cache enabled there, so maybe it's not a big problem09:09
pigmejand if that cache is not enabled then we can enable it there :)09:09
*** salmon_ has joined #openstack-solar09:12
*** salmon_ has quit IRC09:15
*** salmon_ has joined #openstack-solar09:16
openstackgerritMerged openstack/solar: Fixed non working test on riak backend  https://review.openstack.org/28315509:36
salmon_pigmej: remove workflow from this https://review.openstack.org/#/c/283326/1 it's broken09:38
pigmejsalmon_: I tested it09:38
pigmejwhy it's broken you say?09:38
salmon_lines are too long now09:38
salmon_this is correct fix: https://bpaste.net/show/5c5843a2455a09:39
pigmejah I forgot about line length ...09:39
pigmejjenkins already gave +2 ;P09:42
pigmejbut it's not merged so there is still a chance :D09:44
pigmejsalmon_: "We are trying to have max 80 characters in a line." :D:D:D09:44
pigmejand well tox -edocs prints a lot of stuff right now;p09:44
salmon_well, there is no auto test for it and there are still some lines with more than 80 characters09:45
salmon_pigmej: yeah, but message about line length is at the end :)09:45
mkwiekcan't we add another gate check to OS infra for this?09:48
pigmejwell, we could but run tox -edocs :P09:50
salmon_dshulyak: we need to agree on repository path :)09:51
pigmejthe problem is that after bogdando added these checkniceness etc we have a lot of warnings there :D09:51
pigmejmaybe we should create bug for fixing these docs...09:56
*** tzn has joined #openstack-solar10:01
*** tzn has quit IRC10:01
*** tzn has joined #openstack-solar10:01
*** tzn has quit IRC10:03
*** azvyagintsev_ has joined #openstack-solar10:03
*** azvyagintsev_ has quit IRC10:05
*** alexz__ has joined #openstack-solar10:08
*** alexz__ has quit IRC10:21
pigmejdshulyak: salmon_ can you do some PoC of non connections based approach ?10:42
pigmejeven non working PoC, just some described steps.10:44
salmon_pigmej: for now nothing is working :P10:45
pigmejyeah i expected it10:48
pigmejbut can you describe a predicted workflow ?10:48
salmon_"same as in fuel"10:49
pigmejwe need it from solar side10:49
*** dshulyak has quit IRC11:06
openstackgerritBogdan Dobrelya proposed openstack/solar-resources: Do not use atomic file updates for hosts_file  https://review.openstack.org/28312011:06
*** dshulyak has joined #openstack-solar11:25
salmon_dshulyak: I merged your changes and will rebase main11:26
*** azvyagintsev_h__ has joined #openstack-solar11:42
salmon_dshulyak: why this is empty now https://github.com/loles/f2s/blob/master/vrs/role_base-os.yaml ? there are 3 tasks there https://github.com/openstack/fuel-library/blob/master/deployment/puppet/deployment_groups/tasks.yaml#L120-L13311:49
dshulyaksalmon_: thats because relationship from group to task is not parsed, only from task to group, this shouldnt affect any other role11:52
salmon_hmm, ok. We will see :)11:53
pigmejcan I help somehow you both ?11:54
dshulyakbogdando: where are fuel-library fixtures?11:57
dshulyakok i found them11:58
dshulyakpigmej: if you want you can get those f2s artifacts and some fixture, and try to deploy something11:59
dshulyakbut you will need to change network config11:59
pigmejnothing "more" general?12:01
dshulyakwe also need to preare 2 specs, one about exposing compiled graph, another that will describe general integration, even if it wont be complete12:05
pigmejanyway, we're going to deliver f2s as separate package, right?12:07
pigmejbecause we need to include 2 things in fuel, solar itself and fs2, isn't it?12:07
dshulyaki think yes, from f2s we need 2 parts - solar resources and compositions (repository) and client that will take data from fuel and pass it to solar12:09
pigmejso we need to package it too?12:09
pigmejresoruces and compositions => add to solar repository as "fuel", and the client is what, python program, right?12:10
salmon_dshulyak: why did you remove fuel_node?12:12
dshulyaksalmon_: because i forgot that there was also non-generated parts..12:13
salmon_ah :)12:13
salmon_dshulyak: will you revert it?12:13
salmon_I think genkeys also12:14
salmon_dshulyak: you can put it in separate directory12:14
dshulyakyeah, lets revert it, i didnt think that you will merge it12:14
salmon_merge fast, iterate faster!12:15
salmon_dshulyak: not revert everything. just not generated parts12:16
salmon_ah, there was WIP in the tittle. Sorry, I didn't notice :(12:17
salmon_Ok, I will revert not generated parts12:17
dshulyakthere is only 212:18
salmon_fuel_node and genkeys?12:18
salmon_prep.yaml also12:19
dshulyaksalmon_: yes, sorry - 3, btw can you describe how you are using inputs, because i removed them from every resource12:20
dshulyakand i removed role data12:20
salmon_why did you remove role_data?12:20
dshulyakbecause role data schema is sum of all inputs (astute.yaml + globals.yaml) and it is probably outdated12:24
dshulyakand also i thought that you want to feed data to globals resource12:24
dshulyakand then share it via hack in hiera12:24
salmon_dshulyak: I wanted to do it the same way as fuel is doing. By having data only in "role resource" which adds it to hiera12:25
dshulyaksalmon_: but then this resource is not what was needed, isnt it?12:32
dshulyakalso it seems we have problem with schema..12:32
salmon_dshulyak: please explain more :)12:33
dshulyakit can be dynamicly extended with plugins..12:33
salmon_role data?12:33
salmon_or global data?12:34
dshulyakwhat is global?12:35
salmon_dshulyak: I think we need a call12:35
dshulyakyeah, lets use syncup channel12:35
salmon_pigmej: if you want to join^12:35
pigmejI joined12:50
pigmejbut noone is there12:50
salmon_we were fast :)12:57
pigmejback :)13:21
salmon_afk, will be back later13:45
*** salmon_ has quit IRC13:49
pigmejhmm wtf13:53
pigmejdshulyak: did anyone fixed some things in pg backend or these 2 tests?13:54
dshulyakpigmej: i didnt13:54
pigmejbecause one of them started to work13:57
dshulyakyes, i was asking about it yesterday13:58
pigmejbut I also pasted: https://bpaste.net/show/73c651fdfbe513:58
pigmejalso yesterday13:58
pigmejok finally crashed it13:59
bogdandofolks, the docker experimantal support seems ok, this  https://review.openstack.org/#/c/281767 works for me with the patch https://review.openstack.org/#/c/283120/14:26
bogdandoI suggest you to test it, merge as is. Later we can address workarounds (then vagrant ready for docker nets)14:27
bogdandoAnd later we can add support for multi nets as well, to deploy the openstack example14:27
bogdandopigmej, ^^14:28
bogdandoI didn't test preprovisioned mode though, I have no idea how it should be working14:28
openstackgerritBogdan Dobrelya proposed openstack/solar: Add packer build for docker container  https://review.openstack.org/28176714:31
pigmejbogdando: Yeah I'm going to check it :)14:35
pigmejbogdando: can you also take care about migrating to centos from ubuntu ?14:35
pigmejor you prefer to wait from packaging stuff that azvyagintsev is doing?14:35
bogdandopigmej, I think we should add centos packer things, right near to existing ones for ubuntu.14:36
bogdandoI see no point in removing that we already have14:36
bogdandoI will take a look14:37
pigmejbut it will double maintenance time, we need to think about it. Maybe we will leave then the ubuntu as "old" stuff14:37
bogdandodo we have a bug for this created? :)14:37
pigmejnope, you can create it14:38
*** salmon_ has joined #openstack-solar14:40
pigmejbogdando: olar-master-docker.json where is that file?14:40
*** salmon_ has quit IRC14:44
mkwiekhey guys, do we have a generic way to retrieve an object from database? I can't find anything like this in tests14:44
mkwiekI would like to retrieve an object with given uid14:45
*** salmon_ has joined #openstack-solar14:45
bogdandopigmej,  https://bugs.launchpad.net/solar/+bug/154885114:45
openstackLaunchpad bug 1548851 in Solar "Run solar nodes as docker containers (Centos7 based)" [Medium,New] - Assigned to Bogdan Dobrelya (bogdando)14:45
pigmejbogdando: not like that14:46
pigmejwe need 'master' node to be centos :)14:46
bogdandofor now, there is only single image14:46
pigmejnot exactly docker, it has to be centos :)14:46
bogdandoI suggested to start with docker, and finish with libvirt/vbox providers/boxes14:47
bogdandoAnd let's make the split master/slaves images as a separate bug14:47
bogdandowe relly should not do all in one place14:47
mkwieknvm, I think I got it14:48
pigmejbogdando: sure, we need to split the images :)14:48
pigmejbut ok, let's first do docker + centos thingy14:49
pigmejbut we still will need vbox/libvirt stuff I'm afraid for compaitbility14:49
bogdandoI will update them as well14:49
bogdandoyou can check the patch for docker at ubuntu, the diff to atlas json is really small14:50
pigmejbogdando: cool :)14:51
pigmejmkwiek: dbmodel.get(id)14:51
pigmejall dblayer tests are in solar/dblayer/tests14:51
mkwiekpigmej: yeah, but I need to know what dbmodel the key belongs to, right?14:52
mkwiekso if I have system_log:1d38ec24-8042-4d62-ac10-b74a9179f60c I need to call system_log type model?14:53
pigmejsure you need to know it :)14:54
pigmejhmm this one has no model at all14:54
pigmejyeah it's a list of all log items with the same log, why do you ask mkwiek ?14:55
mkwiekI am trying to figure out if anything with uid that I have is available in the database14:56
mkwiekI want to prevent Locks from locking non-existent uids14:57
pigmejoh that's not a good idea I think14:57
pigmejbecause it will require additional lookups14:57
mkwiekyeah, I am aware of that14:57
pigmejwhich will be executed in all cases, even these which are good :)14:58
pigmejwhat exactly are you trying to solve now?14:58
mkwiekstill working on this one: https://bugs.launchpad.net/solar/+bug/154581315:00
openstackLaunchpad bug 1545813 in Solar "Better error reporting in solar-worker when database is wrongly configured" [Undecided,Confirmed] - Assigned to Maciej Kwiek (maciej-iai)15:00
mkwiekwhen there are two different databases15:00
mkwiekwhat would you suggest then?15:00
mkwieksending a message with current db config with zmq message and printing a warning if it doesn't match?15:01
mkwiekthis might be better15:02
pigmejmkwiek: well no why like that ?15:02
pigmejjust proceed, and when you try to fetch graph info you will have empty graph15:02
pigmejthen you can raise error that graph does not exists15:02
mkwiekhmm, that's a good point15:02
pigmejmkwiek: when you're fixing worst case scenario, you shouln't introduce new things in good cases (there are some exceptions though :D)15:07
pigmejbogdando: it sems that i need to configure my fedora first ..15:07
mkwiekpigmej: yeah, I guess I just wanted to fiddle with locking :D15:08
bogdandopigmej, there is docs update in the patch15:08
bogdandobut it doesn't cover docker install15:09
pigmejbogdando: well on fedora only root can use docker; p15:11
bogdandoaddgroup trick doesn't work?15:11
bogdandoby default only root can use it at ubuntu as well15:12
pigmejbogdando: adding user to docker is not recommended15:13
pigmejbut it should work15:13
bogdandosure, but it works. This is only for qa/dev/ci jobs15:13
bogdandothe docker provider I mean15:13
pigmejnope it doesn't solve it15:20
pigmejehs stupid selinux15:21
openstackgerritMaciej Kwiek proposed openstack/solar: Raise an error for empty plan  https://review.openstack.org/28363215:27
pigmejbogdando: is it expected that it provisions *very* slow?15:37
bogdandono, vagrant ups it for like 20 sec at my box15:37
pigmejyeah but just building images etc15:38
pigmejprobably ubuntu docker container installs more stuff than iso :)15:39
pigmejcrashes for me15:42
pigmejwtf is that error...15:43
pigmejmkwiek: I think jenkins will give you -1 :P15:45
*** dshulyak has quit IRC15:47
pigmejbogdando: I think it requires me to thave the same docker in my system that is in centos....15:51
pigmejbogdando: I commented :)15:54
*** dshulyak has joined #openstack-solar16:10
mkwiekpigmej: I am OK with that, wanted to commit sth before I step out of office in case my computer magically erases all the data :D16:19
pigmejmkwiek: good choice :p16:19
*** dshulyak has quit IRC19:17
*** angdraug has joined #openstack-solar19:18
*** dshulyak has joined #openstack-solar20:02
*** angdraug has quit IRC22:05
*** dshulyak has quit IRC22:24
*** salmon_ has quit IRC22:58
*** mihgen has quit IRC22:58
*** kun_huang has quit IRC22:58
*** nihilifer has quit IRC22:58
*** pigmej has quit IRC22:58
*** sylwesterB has quit IRC22:58
*** evgenyl has quit IRC22:58
*** mkwiek has quit IRC22:58
*** azvyagintsev_h__ has quit IRC22:58
*** azvyagintsev has quit IRC22:58
*** ikalnitsky has quit IRC22:58
*** bogdando has quit IRC22:58
*** ChanServ has quit IRC22:58
*** bogdando has joined #openstack-solar23:01
*** ikalnitsky has joined #openstack-solar23:01
*** kun_huang has joined #openstack-solar23:01
*** mihgen has joined #openstack-solar23:01
*** salmon_ has joined #openstack-solar23:01
*** mkwiek has joined #openstack-solar23:01
*** evgenyl has joined #openstack-solar23:01
*** sylwesterB has joined #openstack-solar23:01
*** nihilifer has joined #openstack-solar23:01
*** pigmej has joined #openstack-solar23:01
*** azvyagintsev_h__ has joined #openstack-solar23:03
*** azvyagintsev has joined #openstack-solar23:03
*** pigmej has quit IRC23:04
*** pigmej has joined #openstack-solar23:04
*** ChanServ has joined #openstack-solar23:06
*** asimov.freenode.net sets mode: +o ChanServ23:06

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