| *** jamielennox|away is now known as jamielennox | 00:02 | |
| *** toure is now known as toure|gone | 00:26 | |
| *** FL1SK has joined #openstack-mistral | 00:33 | |
| *** FL1SK has quit IRC | 00:47 | |
| *** zhurong has joined #openstack-mistral | 00:49 | |
| *** jamielennox is now known as jamielennox|away | 01:00 | |
| *** jamielennox|away is now known as jamielennox | 01:10 | |
| *** bobh has joined #openstack-mistral | 01:25 | |
| *** zhurong has quit IRC | 01:29 | |
| *** bobh has quit IRC | 01:34 | |
| *** zhurong has joined #openstack-mistral | 02:04 | |
| *** FL1SK has joined #openstack-mistral | 03:02 | |
| *** bobh has joined #openstack-mistral | 03:29 | |
| *** gkadam has joined #openstack-mistral | 03:38 | |
| *** bobh has quit IRC | 03:43 | |
| *** bobh has joined #openstack-mistral | 03:45 | |
| *** bobh has quit IRC | 03:45 | |
| *** bobh has joined #openstack-mistral | 03:47 | |
| *** bobh has quit IRC | 03:50 | |
| *** zhurong has quit IRC | 04:25 | |
| *** zhurong has joined #openstack-mistral | 04:33 | |
| *** jaosorior has joined #openstack-mistral | 06:18 | |
| *** zhurong has quit IRC | 06:55 | |
| *** zhurong has joined #openstack-mistral | 06:56 | |
| *** shardy has joined #openstack-mistral | 06:56 | |
| *** zhurong has quit IRC | 06:57 | |
| *** amoralej|off is now known as amoralej | 07:07 | |
| openstackgerrit | JeffreyGuan proposed openstack/mistral master: This is only a minor defect in README.rst. https://review.openstack.org/472552 | 07:28 |
|---|---|---|
| *** jpich has joined #openstack-mistral | 07:55 | |
| *** dtantsur|sick is now known as dtantsur | 08:48 | |
| openstackgerrit | Van Hung Pham proposed openstack/mistral master: Optimize the link address https://review.openstack.org/472601 | 09:21 |
| *** zhurong has joined #openstack-mistral | 09:54 | |
| *** zhurong has quit IRC | 10:06 | |
| *** jkilpatr has joined #openstack-mistral | 10:59 | |
| *** shardy has quit IRC | 12:08 | |
| *** rbrady-afk is now known as rbrady | 12:08 | |
| *** shardy has joined #openstack-mistral | 12:17 | |
| *** bobh has joined #openstack-mistral | 12:18 | |
| *** bobh has quit IRC | 12:27 | |
| *** amoralej is now known as amoralej|lunch | 12:29 | |
| *** jaosorior has quit IRC | 12:37 | |
| *** catintheroof has joined #openstack-mistral | 12:40 | |
| *** shardy has quit IRC | 12:43 | |
| *** catintheroof has quit IRC | 12:45 | |
| *** catintheroof has joined #openstack-mistral | 12:46 | |
| fultonj | rbrady: d0ugal wanted to run something by you regarding the mistral user's home dir and its impact on mistral-ansible-actions . | 12:54 |
| fultonj | when you have a minute | 12:54 |
| rbrady | fultonj: ack | 12:55 |
| rbrady | fultonj: I'm free for about 30 min | 12:55 |
| fultonj | rbrady: ok, thanks | 12:55 |
| *** shardy has joined #openstack-mistral | 12:56 | |
| rbrady | fultonj: am I supposed to contact d0ugal? | 13:06 |
| fultonj | rbrady: no, if he was there i just wanted to share it with him | 13:06 |
| rbrady | fultonj: ahh...the lack of a comma made it sound like you were delivering a message that he wanted to talk to me | 13:07 |
| rbrady | fultonj: he's got an appt this morning and won't likely be around until later | 13:07 |
| *** bobh has joined #openstack-mistral | 13:11 | |
| fultonj | rbrady: sorry someone called me; if you want to ping me when you're free i can show you my idea regarding the home dir. I can see how my message wasn't clear (i want to run it by you, not d0ugal) | 13:17 |
| fultonj | (or d0ugal doesn't have somthing to tell you regarding it (/me having ambiguity of language issues)) | 13:18 |
| rbrady | fultonj: I've got a call in a few minutes, will ping you afterwards | 13:21 |
| fultonj | thanks | 13:21 |
| *** toure|gone is now known as toure | 13:27 | |
| openstackgerrit | Thomas Herve proposed openstack/mistral master: Always retrieve endpoint in mistral action https://review.openstack.org/472682 | 13:37 |
| *** amoralej|lunch is now known as amoralej | 13:38 | |
| *** chlong has joined #openstack-mistral | 13:51 | |
| *** jaosorior has joined #openstack-mistral | 13:56 | |
| *** gkadam_ has joined #openstack-mistral | 14:26 | |
| *** gkadam has quit IRC | 14:26 | |
| *** gkadam_ has quit IRC | 14:31 | |
| d0ugal | fultonj, rbrady - hey, I can help now. | 14:32 |
| * d0ugal reads up | 14:32 | |
| fultonj | hi d0ugal | 14:33 |
| fultonj | Mistral's home directory is set to /home/mistral [1], as per the RDO package I suspect, but that directory does not exist. It should be set to an existing directory and doing so fixes a problem with Mistral ansible actions. | 14:33 |
| fultonj | i have the details here | 14:34 |
| fultonj | https://github.com/fultonj/tripleo-ceph-ansible/issues/12 | 14:34 |
| d0ugal | fultonj: any idea what directory it should be set to? | 14:35 |
| fultonj | i don't know the back story with Mistral's home dir | 14:35 |
| d0ugal | fultonj: I suspect it was just a decision made in the packaging? | 14:36 |
| fultonj | yeah, i think so | 14:36 |
| fultonj | but i wasn't sure if there was a story behind it | 14:36 |
| d0ugal | fultonj: honestly this stuff is over my head, but I fairly sure that Mistral doesn't need that directory - since it doesn't exist I guess that is the case :) | 14:36 |
| fultonj | right it certainly doesn't need it | 14:36 |
| d0ugal | Should we have that directory created? if not, what should it be changed to? | 14:37 |
| d0ugal | fultonj: is the issue that ansible wants to create temp files in the home directory? | 14:37 |
| fultonj | ansible likes to write temp files SOMEWHERE | 14:37 |
| d0ugal | right | 14:37 |
| fultonj | exactly | 14:37 |
| fultonj | if i configure ansible on the server to always use a central place that runs into issues | 14:37 |
| fultonj | suppose i globally tell ansible write temp files to /tmp/foo | 14:38 |
| fultonj | the first user to use ansible then gets their permissions attached to /tmp/foo/.ansible | 14:38 |
| fultonj | so the mitral user is happy but not the stack user | 14:38 |
| fultonj | point is | 14:38 |
| fultonj | so far all mistral actions have managed their temp files in their own way OR not needed them | 14:39 |
| fultonj | it seems ^ | 14:39 |
| d0ugal | right | 14:40 |
| d0ugal | I would probably ask for help in #rdo to ask if we should create the directory or change it. | 14:40 |
| d0ugal | To pick a random project, Heat seems to use a shared directory: https://github.com/rdo-packages/heat-distgit/blob/rpm-master/openstack-heat.spec#L324 | 14:40 |
| fultonj | i could they will probably want mistrallers opinions | 14:40 |
| d0ugal | That would be fine for us too I think | 14:40 |
| rbrady | I concur with d0ugal on the shared dir | 14:41 |
| fultonj | /var/lib/heat | 14:41 |
| fultonj | is what heat gets | 14:41 |
| fultonj | i could request of the RDO packager to make | 14:41 |
| fultonj | /var/lib/mistral and set the msitral user's home to that | 14:41 |
| fultonj | wanted to check here if that's OK | 14:42 |
| rbrady | fultonj: can't we just submit a change to the specfile in the distgit ? | 14:42 |
| fultonj | rbrady: yeah | 14:42 |
| fultonj | in theory | 14:42 |
| fultonj | :) | 14:42 |
| d0ugal | fultonj: that sounds good to me. | 14:42 |
| fultonj | ok, thanks | 14:43 |
| fultonj | if you didn't want that done for some mistrally reason | 14:43 |
| fultonj | then i was tweaking the playbook action | 14:43 |
| fultonj | to temporarily set HOME for the playbook run | 14:43 |
| *** Dinesh_Bhor has quit IRC | 14:44 | |
| fultonj | i gather you prefer the RDO pkg approach? | 14:44 |
| rbrady | fultonj: I do | 14:44 |
| fultonj | here's a hack http://paste.openstack.org/raw/612140/ | 14:45 |
| fultonj | but i think i can use ansible's environment https://docs.ansible.com/ansible/playbooks_environment.html | 14:45 |
| d0ugal | the pkg change seems cleaner :) | 14:45 |
| fultonj | cool | 14:45 |
| fultonj | i like it too | 14:45 |
| fultonj | thanks guys | 14:45 |
| *** jaosorior has quit IRC | 15:09 | |
| *** dtantsur is now known as dtantsur|afk | 15:19 | |
| * flaper87 back | 15:30 | |
| flaper87 | flaper87: do I have to read all the messages or is there a tl;dr ? | 15:30 |
| flaper87 | :D | 15:30 |
| flaper87 | fultonj: ^ | 15:30 |
| * flaper87 is talking to himself | 15:30 | |
| flaper87 | it's friday after all | 15:30 |
| fultonj | flaper87: i can tl;dr it | 15:31 |
| fultonj | one sec | 15:32 |
| fultonj | flaper87: the mistral user's home dir is set to /home/mistral as per the RDO package | 15:35 |
| fultonj | https://github.com/rdo-packages/mistral-distgit/blob/rpm-master/openstack-mistral.spec#L281-L289 | 15:35 |
| fultonj | this has never been a problem, but that dir doesn't exist | 15:36 |
| fultonj | ansible wants to write temp files somewhere | 15:36 |
| fultonj | so i am going to propose the mistral package gets upated to make a real home dir | 15:37 |
| fultonj | else mistral-ansible-actions won't work | 15:37 |
| fultonj | flaper87: you listed 3 workarounds you did in your kargo/ansible change | 15:38 |
| fultonj | i'm trying to make it so you don't need them | 15:38 |
| flaper87 | fultonj: mmh, but, I think we could tell ansible where this files should be written | 15:39 |
| flaper87 | I'd rather configure ansible.cfg than changing the package, tbh | 15:39 |
| fultonj | flaper87: i tried that but it wasn't sufficient | 15:40 |
| flaper87 | my current problems are mostly related to the ssh_keys and playbooks having to be readable by the mistral user | 15:40 |
| flaper87 | fultonj: oh, really? | 15:40 |
| flaper87 | mmh | 15:40 |
| flaper87 | interesting | 15:40 |
| flaper87 | perhaps something we should fix in ansible? | 15:40 |
| fultonj | flaper87: i will gladly explain the issues | 15:40 |
| flaper87 | fultonj: please do | 15:40 |
| fultonj | two problems with just updating ansible.cfg | 15:41 |
| fultonj | 1. setting local temp to /tmp/.ansible works on the first run by Mistral: | 15:42 |
| fultonj | local_tmp= /tmp/.ansible/tmp | 15:42 |
| fultonj | but then the stack user cannot use ansible: | 15:42 |
| fultonj | ansible.errors.AnsibleError: Unable to create local directories(/tmp/.ansible/tmp): [Errno 13] Permission denied: '/tmp/.ansible/tmp' | 15:42 |
| fultonj | Commenting out "local_tmp=/tmp/.ansible/tmp" so that it goes back to it's default | 15:42 |
| fultonj | of ~/.ansible is better for whichver user ran it. | 15:42 |
| fultonj | 2. ansible, and it's modules, use $HOME quite a bit | 15:43 |
| fultonj | even with local_tmp set | 15:43 |
| fultonj | i still ran into problems | 15:43 |
| fultonj | because of $HOME expnsion to /home/mistral | 15:44 |
| flaper87 | I think both could be solved by making the execution localized | 15:45 |
| flaper87 | for example | 15:45 |
| flaper87 | it's possible to set the local_tmp dir in the inventory, IIRC | 15:45 |
| flaper87 | if not, we could create an ansible.cfg *per* mistral-action execution | 15:45 |
| flaper87 | like copy the one from `/etc/ansible`/ to some dir in `/tmp` and use that | 15:45 |
| flaper87 | just like we are doing for playbooks and inventories | 15:46 |
| flaper87 | for 2 we can set the home directory to `/tmp/mistral` or something | 15:46 |
| flaper87 | actually, tbh, I think we should. I don't really like the idea of mistral-ansible-actions depending on anything global | 15:46 |
| flaper87 | or creating files that are not temp | 15:46 |
| flaper87 | that sounds like a mess for whenever we'll have parallel executions of some sort | 15:47 |
| flaper87 | thoughts ? | 15:47 |
| flaper87 | I can help with both | 15:47 |
| flaper87 | d0ugal: ^ you may be interested in this too | 15:47 |
| fultonj | rbrady: ^ ? | 15:47 |
| * d0ugal reads | 15:48 | |
| d0ugal | flaper87: parallel actions is a good point. What happens if ansible is ran multiple times? | 15:49 |
| d0ugal | flaper87: I'm fine with the execution being localized... but I have no idea how you do that :) | 15:50 |
| flaper87 | d0ugal: exactly, well, right now, it'll just use global stuff | 15:50 |
| flaper87 | d0ugal: oh, I mean, we create a "home" for each execution | 15:50 |
| d0ugal | Wouldn't that require it being it its own process etc? | 15:50 |
| flaper87 | instead of creating a temp file for playbook/inventory, what we can do is create a temp dir | 15:50 |
| flaper87 | put stuff there | 15:50 |
| d0ugal | right | 15:50 |
| flaper87 | and we create an ansible.cfg file there | 15:51 |
| flaper87 | etc etc | 15:51 |
| fultonj | I don't think you can define HOME in ansible.cfg | 15:51 |
| fultonj | basically if we just set $HOME for the execution, things are happy | 15:52 |
| flaper87 | you can define it when you run the process | 15:52 |
| flaper87 | :) | 15:52 |
| fultonj | yes | 15:52 |
| fultonj | http://paste.openstack.org/raw/612146/ | 15:52 |
| flaper87 | that's what I meant | 15:52 |
| flaper87 | set it for the execution not ansible.cfg | 15:52 |
| flaper87 | this shouldn't be too hard to do | 15:52 |
| flaper87 | I can help out | 15:52 |
| flaper87 | did the patch land in tripleo_common ? | 15:53 |
| fultonj | mistral ansible actions hasn't landed yet | 15:53 |
| fultonj | ansible has a notion of enviornemnts that might help us here | 15:53 |
| *** jpich has quit IRC | 15:53 | |
| * fultonj looks for the odc | 15:54 | |
| fultonj | doc | 15:54 |
| flaper87 | yeah, that too | 15:54 |
| fultonj | http://docs.ansible.com/ansible/playbooks_environment.html | 15:54 |
| fultonj | flaper87: i conjecture that my setting $HOME alone for the life of the playbook run to something that the local mistral user can write to that it will solve the problem and not even require a change to ansible.cfg | 15:56 |
| fultonj | that "my setting" but "by setting" | 15:56 |
| fultonj | s/that/not | 15:56 |
| flaper87 | :) | 15:56 |
| flaper87 | I like that solution, TBH | 15:56 |
| d0ugal | regardless of how we solve this for ansible, the mistral user should have a home directory that exists | 15:57 |
| flaper87 | d0ugal: curious to know why you think that ? | 15:57 |
| d0ugal | flaper87: well, why have one otherwise? | 15:57 |
| flaper87 | s/?// | 15:57 |
| flaper87 | we can remove it | 15:57 |
| flaper87 | :) | 15:57 |
| flaper87 | instead of setting it | 15:58 |
| d0ugal | okay, that is also fine | 15:58 |
| flaper87 | the fact that it has one might be a bug | 15:58 |
| d0ugal | yeah | 15:58 |
| d0ugal | This stuff is mostly a level too far down for me, I have no strong opinion | 15:59 |
| flaper87 | cool, my question was more like: "Is there something in mistral that requires this dir to exist?" | 15:59 |
| flaper87 | if not then let's get rid of it | 15:59 |
| flaper87 | we don't need it | 15:59 |
| flaper87 | and I'm not a fan of leaving files hanging around the FS | 16:00 |
| flaper87 | especially tripleo files | 16:00 |
| flaper87 | or openstack deployment files | 16:00 |
| flaper87 | fultonj: so, do you want to do the home dir thing or do you want me to do it ? | 16:00 |
| d0ugal | I don't think it is required | 16:00 |
| flaper87 | fultonj: I can do it on top of rbrady's patch | 16:00 |
| fultonj | flaper87: do you know how you want to solve it? | 16:01 |
| flaper87 | I've an idea, yes! | 16:01 |
| * rbrady scrolls back up | 16:01 | |
| *** amoralej is now known as amoralej|brb | 16:02 | |
| fultonj | i was going to test run this http://paste.openstack.org/raw/612146/ to test my theory | 16:02 |
| fultonj | and then see what i can do with ansible environments | 16:02 |
| flaper87 | I'm not sure ansible envs will help much here | 16:03 |
| *** fultonj has quit IRC | 16:07 | |
| *** fultonj has joined #openstack-mistral | 16:09 | |
| fultonj | hello? | 16:09 |
| *** rbrady has quit IRC | 16:10 | |
| fultonj | flaper87: sorry i bounced out for a bit ; did you already sove the problem :) ? | 16:10 |
| fultonj | solve* | 16:10 |
| fultonj | comcast-- | 16:10 |
| flaper87 | fultonj: hahaha, yeah, almost | 16:11 |
| flaper87 | should have a patch soon | 16:11 |
| flaper87 | I'm confident this will work | 16:11 |
| fultonj | flaper87++ cool | 16:11 |
| flaper87 | (last famous words) | 16:12 |
| fultonj | :) | 16:12 |
| *** rbrady has joined #openstack-mistral | 16:18 | |
| *** rbrady has joined #openstack-mistral | 16:18 | |
| *** fultonj has quit IRC | 16:33 | |
| *** fultonj has joined #openstack-mistral | 16:41 | |
| *** fultonj has quit IRC | 16:50 | |
| *** fultonj has joined #openstack-mistral | 16:52 | |
| *** amoralej|brb is now known as amoralej | 17:20 | |
| *** rbrady is now known as rbrady-afk | 18:00 | |
| *** amoralej is now known as amoralej|off | 18:40 | |
| *** fultonj has quit IRC | 18:51 | |
| *** fultonj has joined #openstack-mistral | 18:52 | |
| *** amoralej|off is now known as amoralej | 19:36 | |
| *** rbrady-afk has quit IRC | 20:29 | |
| fultonj | flaper87: i found a way to get the home dir thing working for ansible | 20:35 |
| fultonj | stderr, stdout = processutils.execute( | 20:35 |
| fultonj | *command, log_errors=processutils.LogErrors.ALL, | 20:35 |
| fultonj | env_variables={'HOME':'/tmp/mistral-ansible-home'}) | 20:35 |
| fultonj | flaper87: but maybe you had something different in mind | 20:36 |
| fultonj | ^ i guess that argument for HOME could be something created temporarily and then cleaned up after the execution | 20:37 |
| *** rbrady-afk has joined #openstack-mistral | 20:45 | |
| *** bobh has quit IRC | 20:56 | |
| *** shardy has quit IRC | 21:13 | |
| flaper87 | fultonj: that's kinda what I did, yes. | 21:21 |
| fultonj | cool | 21:21 |
| flaper87 | Just passing env_variables to the CLI command | 21:21 |
| *** amoralej is now known as amoralej|off | 21:21 | |
| fultonj | flaper87: is your change only local ? | 21:21 |
| flaper87 | fultonj: for now yes, I'm about to push it but it might have to wait till monday | 21:24 |
| flaper87 | pls, don't work on this to avoid duplicating efforts | 21:24 |
| flaper87 | I know you need it | 21:24 |
| fultonj | ok, thanks | 21:25 |
| fultonj | i have my own hack so i can keep moving on my other stuff | 21:25 |
| fultonj | i'll switch to the real thing when it lands | 21:25 |
| flaper87 | ok | 21:25 |
| fultonj | this ^ and https://review.openstack.org/#/c/469644/8/tripleo_common/actions/validations.py | 21:26 |
| fultonj | should take care of 1 and 3 in the commit message of https://review.openstack.org/#/c/471759/ | 21:26 |
| fultonj | for 2. make access to the `kargo` roles/playbooks, were you going to get an RPM for kargo in RDO? | 21:27 |
| *** jkilpatr has quit IRC | 21:28 | |
| flaper87 | there's an RPM. What's broken there is the ansible search path. by including `kargo/roles/...` it'll try to include it from the same dir the playbook is instead of looking in the global dir | 21:33 |
| fultonj | is it looking for anything you can explicitly tell it about? | 21:36 |
| fultonj | e.g. i add action_plugins = /usr/share/ceph-ansible/plugins/actions to /etc/ansible/ansible.cfg | 21:36 |
| *** chlong has quit IRC | 21:58 | |
| *** catintheroof has quit IRC | 22:42 | |
| *** bobh has joined #openstack-mistral | 22:49 | |
| *** bobh has quit IRC | 22:49 | |
| *** dougshelley66 has joined #openstack-mistral | 22:59 | |
| *** amoralej|off is now known as amoralej | 23:00 | |
| *** toure is now known as toure|gone | 23:10 | |
| *** rbrady-afk has quit IRC | 23:56 | |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!