| *** salv-orlando has joined #kolla | 00:00 | |
| *** salv-orl_ has quit IRC | 00:03 | |
| *** sdake has quit IRC | 00:07 | |
| *** cloudnautique has quit IRC | 00:08 | |
| *** coolsvap|away has quit IRC | 00:12 | |
| *** sdake has joined #kolla | 00:13 | |
| *** coolsvap|away has joined #kolla | 00:14 | |
| *** tzn has joined #kolla | 00:27 | |
| *** mbound_ has quit IRC | 00:29 | |
| *** dims has quit IRC | 00:42 | |
| *** salv-orlando has quit IRC | 00:51 | |
| *** tzn has quit IRC | 01:01 | |
| *** sdake has quit IRC | 01:10 | |
| *** dims has joined #kolla | 01:16 | |
| *** ssurana has joined #kolla | 01:18 | |
| openstackgerrit | Sam Yaple proposed openstack/kolla: pep8 the ansible modules https://review.openstack.org/260101 | 01:25 |
|---|---|---|
| SamYaple | nihilifer: the "# noqa" thing, i didnt realize that was a _per_line_ change | 01:25 |
| SamYaple | thats my mistake | 01:25 |
| SamYaple | im all for that! | 01:25 |
| SamYaple | implemented above | 01:25 |
| SamYaple | nihilifer: also there were two pep8 errors like i remember, but both were related to the 'import *' stuff, so that should be all we need to proceed | 01:28 |
| *** mbound_ has joined #kolla | 01:29 | |
| *** ssurana has quit IRC | 01:30 | |
| openstackgerrit | Merged openstack/kolla: Use c-style string interpolation for log messages https://review.openstack.org/259325 | 01:32 |
| *** mbound_ has quit IRC | 01:34 | |
| openstackgerrit | David Moreau Simard proposed openstack/kolla: Make the yum repositories configurable in the base image https://review.openstack.org/258017 | 01:37 |
| *** dims has quit IRC | 01:40 | |
| dmsimard | SamYaple: oh so should I put back the %s's then ? :p | 01:41 |
| openstackgerrit | Jeffrey Zhang proposed openstack/kolla: Add the rhel in the base_distro condition in swift dockerfile https://review.openstack.org/260294 | 01:44 |
| SamYaple | dmsimard: no you should be using .format() still | 01:46 |
| SamYaple | notice nihilifer comment in the logging patch, thats where I was tripped up. the logging function is using ('something %s %s', arg1, arg2) and not ('something %s %s' % arg1, arg2) | 01:47 |
| SamYaple | that was the key difference, its a change specific to the logging stuff only | 01:47 |
| dmsimard | ohhhhhhhhh | 01:48 |
| dmsimard | make sense | 01:48 |
| SamYaple | that said, you could change 'cmd = "RUN yum -y install {file}".format(file=file)' to 'cmd = "RUN yum -y install {}".format(file)' | 01:50 |
| *** dims has joined #kolla | 01:50 | |
| SamYaple | but its not a big thing | 01:50 |
| dmsimard | Will do, in fact I didn't know you could put just {}, I thought indexes were required (i.e, {0} {1}) | 01:55 |
| openstackgerrit | Jeffrey Zhang proposed openstack/kolla: Fix the mysql connection string in doc https://review.openstack.org/260295 | 01:56 |
| dmsimard | Running the deploy test with both the pymysql patch and my patch to see if it works now | 01:57 |
| dmsimard | I did a mini script to test out different ways of passing things and getting the expected result (i.e, http://paste.openstack.org/show/482463/) | 01:58 |
| dmsimard | turns out when you do a split on an empty string, it puts the empty string as is in the list .. so it would iterate over it and just put nothing.. like "yum -y install " | 01:59 |
| dmsimard | different flavors of using None didn't work out too well either | 02:00 |
| SamYaple | dmsimard: thats not what actually happens thoguht. you can test on the command line with python | 02:00 |
| SamYaple | for _ in '': | 02:00 |
| SamYaple | print('nothing prints') | 02:00 |
| dmsimard | '' and [''] is not the same | 02:00 |
| dmsimard | try for _ in [''] :P | 02:01 |
| SamYaple | correct, it is not | 02:01 |
| dmsimard | split yields a list | 02:01 |
| SamYaple | thats not an empty tlist | 02:01 |
| *** dave-mccowan has quit IRC | 02:01 | |
| SamYaple | oh i see yo uare generating a list | 02:01 |
| dmsimard | yeah, but I need to split the string by commas since configparser cannot yield a list, it looks like the best way to do it is to put a csv | 02:01 |
| dmsimard | each parameter can have no value, or n values | 02:02 |
| SamYaple | well you could do a 'for file in repo_blah: if file: **rest of code**' | 02:02 |
| SamYaple | dont know if thats cleaner | 02:02 |
| SamYaple | all i know is the code right now can be broken in many ways | 02:02 |
| dmsimard | it looks like filter does the job of getting the empty string out if necessary | 02:02 |
| dmsimard | broken where ? | 02:03 |
| SamYaple | if not any((self.rpm_file, self.rpm_repo_file, self.rpm_repo_url)): | 02:03 |
| dmsimard | yeah, that's the only way I found that provides what I expect :( | 02:03 |
| SamYaple | let me throw up a gist of what i think works with my understanding | 02:04 |
| *** dims has quit IRC | 02:06 | |
| *** Guest9750 has quit IRC | 02:10 | |
| SamYaple | dmsimard: http://paste.openstack.org/show/482465/ | 02:13 |
| SamYaple | that will go through all three options and only use the default if no option has any values | 02:14 |
| dmsimard | SamYaple: recursing is cheating | 02:15 |
| dmsimard | jk, I like it | 02:15 |
| SamYaple | dmsimard: we can do it without recursion, that was for DRY reasons | 02:15 |
| dmsimard | SamYaple: I wanted to do something like that but it was DRY | 02:16 |
| dmsimard | hence why I handled rpm_url at the end | 02:16 |
| SamYaple | if we do better checking on the rpm_repo_url variable _first_ that would also work | 02:16 |
| dmsimard | and setting the default first | 02:16 |
| SamYaple | better checking first would avoid the recursion i mena | 02:16 |
| SamYaple | and still be DRY | 02:16 |
| *** sdake has joined #kolla | 02:17 | |
| dmsimard | We don't know if we need to set rpm_repo_url until we check the other two, so it'd be kind of like the any | 02:17 |
| SamYaple | thats true, but any() didnt so what you wanted | 02:17 |
| SamYaple | because it was a list full of a string (even empty) it was retuning true | 02:17 |
| SamYaple | any(([''])) == True | 02:18 |
| dmsimard | not in my experience | 02:18 |
| dmsimard | no | 02:18 |
| SamYaple | yes | 02:18 |
| dmsimard | because filter was taking the empty string out | 02:18 |
| dmsimard | :P | 02:18 |
| SamYaple | but you said 02:01:41 < dmsimard> turns out when you do a split on an empty string, it puts the empty string as is in the list .. so it would iterate over it and just put nothing.. like "yum -y install " | 02:19 |
| SamYaple | if filter was doing that, this wouldnt be an issue | 02:19 |
| SamYaple | if filter is doing its job like yo usay, you don't need the if statements i have in there | 02:19 |
| dmsimard | SamYaple: http://paste.openstack.org/show/482467/ | 02:21 |
| dmsimard | what about we do a mix of my way and your way, I end up recursing if rpm_setup is not set and use filter instead of the if's | 02:22 |
| SamYaple | dmsimard: im not suggesting not using filter, I was just going on your earlier comment | 02:22 |
| dmsimard | yeah I was mostly ranting about the different things I tried (and failed) | 02:23 |
| SamYaple | im sure youll come up with something you like | 02:24 |
| dmsimard | SamYaple: while I like the recurse approach, the one thing I can see is that it's not efficient (it'll re-run the other for loops) | 02:25 |
| dmsimard | at this kind of scale/iterations it doesn't matter much but bah | 02:25 |
| SamYaple | well its also recursion which is "bad" by default for alot of people. i think its valid, but it does make it a bit more complicated and easier to break | 02:26 |
| *** rahulshr has joined #kolla | 02:26 | |
| SamYaple | the true safe way to do this would be a while loop probable | 02:26 |
| *** gfidente|afk has quit IRC | 02:28 | |
| SamYaple | dmsimard: http://paste.openstack.org/show/482468/ | 02:28 |
| SamYaple | that will prevent the purist from complaining | 02:28 |
| *** dims has joined #kolla | 02:29 | |
| *** sdake_ has joined #kolla | 02:34 | |
| *** sdake has quit IRC | 02:34 | |
| openstackgerrit | David Moreau Simard proposed openstack/kolla: Make the yum repositories configurable in the base image https://review.openstack.org/258017 | 02:36 |
| dmsimard | ^ I didn't put the while cause I don't feel it's a good use case to iterate over the two other params that are not expected to change | 02:37 |
| dmsimard | ttyl | 02:37 |
| *** unicell1 has quit IRC | 02:37 | |
| *** suro-patz has joined #kolla | 02:42 | |
| *** dims has quit IRC | 02:45 | |
| *** dims has joined #kolla | 02:47 | |
| *** salv-orlando has joined #kolla | 02:55 | |
| openstackgerrit | Martin André proposed openstack/kolla: Add a new tool to help find the latest versions https://review.openstack.org/248481 | 02:55 |
| *** tzn has joined #kolla | 02:58 | |
| *** tzn has quit IRC | 03:03 | |
| *** suro-patz has quit IRC | 03:04 | |
| *** stvnoyes has quit IRC | 03:05 | |
| *** stvnoyes has joined #kolla | 03:05 | |
| *** dims has quit IRC | 03:08 | |
| *** rahulshr has quit IRC | 03:11 | |
| *** coolsvap|away is now known as coolsvap | 03:21 | |
| *** suro-patz has joined #kolla | 03:27 | |
| *** kjelly has quit IRC | 03:28 | |
| *** ssurana has joined #kolla | 03:32 | |
| *** salv-orlando has quit IRC | 03:35 | |
| *** klint has joined #kolla | 03:50 | |
| *** tzn has joined #kolla | 03:59 | |
| *** tzn has quit IRC | 04:04 | |
| *** suro-patz has quit IRC | 04:07 | |
| *** ssurana has quit IRC | 04:07 | |
| openstackgerrit | Sam Yaple proposed openstack/kolla: pep8 the ansible modules https://review.openstack.org/260101 | 04:30 |
| openstackgerrit | Sam Yaple proposed openstack/kolla: Simplify config creation https://review.openstack.org/259624 | 04:30 |
| openstackgerrit | Sam Yaple proposed openstack/kolla: Convert to pymysql https://review.openstack.org/260080 | 04:30 |
| *** unicell has joined #kolla | 04:41 | |
| *** tzn has joined #kolla | 05:00 | |
| *** tzn has quit IRC | 05:05 | |
| *** diga has joined #kolla | 05:05 | |
| *** sdake_ has quit IRC | 05:23 | |
| openstackgerrit | Jeffrey Zhang proposed openstack/kolla: Add the rhel in the base_distro condition in swift dockerfile https://review.openstack.org/260294 | 05:34 |
| *** salv-orlando has joined #kolla | 05:41 | |
| *** tzn has joined #kolla | 06:01 | |
| *** tzn has quit IRC | 06:05 | |
| *** rahulshr_ has joined #kolla | 06:06 | |
| *** kjelly has joined #kolla | 06:07 | |
| rahulshr_ | I am following kolla getting started guide for bare metal machine with all in one configuration | 06:07 |
| rahulshr_ | getting below docker error while deploying kolla using kolla-ansible deploy | 06:08 |
| rahulshr_ | TASK: [common | Starting Ansible container] *********************************** failed: [localhost] => {"changed": true, "failed": true} msg: Docker API Error: Cannot start container 1a1276d5b5c282e07c30dca88d4723fd035af90a1a361647f9dbd43c91a1c7d2: [8] System error: not a directory FATAL: all hosts have already failed -- aborting PLAY RECAP ******************************************************************** to retry | 06:08 |
| rahulshr_ | I understand this is a docker related issue but have no idea how to resolve it | 06:09 |
| nihilifer | rahulshr_: try to do | 06:09 |
| nihilifer | pip install docker-py==1.5.0 | 06:09 |
| nihilifer | on the hosts you're trying to deploy kolla | 06:10 |
| rahulshr_ | I have installed installed docker-py 1.5.0 only | 06:10 |
| rahulshr_ | previously I have docker 1.6.x due to which I was getting client and server mismatch, then I installed 1.5.0 | 06:11 |
| Jeffrey4l | rahulshr_, could u run a container manually? | 06:13 |
| rahulshr_ | how can I run manually? I am pretty new to kolla and docker | 06:13 |
| Jeffrey4l | rahulshr_, docker run -it --rm centos bash | 06:15 |
| Jeffrey4l | try this. | 06:15 |
| Jeffrey4l | to make sure you docker service is ok. | 06:16 |
| *** salv-orlando has quit IRC | 06:16 | |
| rahulshr_ | docker run -it --rm ubuntu bash root@242667adc5c0:/ | 06:17 |
| Jeffrey4l | rahulshr_, it is ok. | 06:17 |
| nihilifer | rahulshr_: please do | 06:17 |
| nihilifer | docker ps -a | 06:17 |
| nihilifer | and send the output on some pastebin (paste.openstack.org / paste.fedoraproject.org) | 06:18 |
| Jeffrey4l | I doubt that the folder `/var/lib/kolla/dev/log` doesn't exist on the host. | 06:19 |
| nihilifer | rahulshr_: btw, you're doing a deployment on multiple nodes or only one?\ | 06:19 |
| Jeffrey4l | But I have no idea how it create, now? | 06:19 |
| rahulshr_ | http://paste.openstack.org/show/482475/ | 06:19 |
| rahulshr_ | all in one | 06:19 |
| Jeffrey4l | rahulshr_, could u check the folder `/var/lib/kolla/dev/` exist? | 06:20 |
| rahulshr_ | yes | 06:21 |
| Jeffrey4l | rahulshr_, pls use `docker logs rsyslog`. You rsyslog container don't run successfully. it should be the root cause. | 06:22 |
| rahulshr_ | it is failing | 06:24 |
| rahulshr_ | IOError: [Errno 2] No such file or directory: '/opt/kolla/config_files/config.json' | 06:24 |
| Jeffrey4l | rahulshr_, u may build the image in a wrong. | 06:25 |
| rahulshr_ | I built it with this cmd | 06:25 |
| * Jeffrey4l be back in a minute. | 06:25 | |
| rahulshr_ | tools/build.py --base ubuntu --type source | 06:26 |
| rahulshr_ | tools/build.py --base ubuntu --type source | 06:26 |
| *** ashestakov has joined #kolla | 06:34 | |
| ashestakov | coolsvap: ping | 06:35 |
| coolsvap | ashestakov, pong | 06:35 |
| ashestakov | look this one pls https://bugs.launchpad.net/kolla/+bug/1528341 | 06:35 |
| openstack | Launchpad bug 1528341 in kolla "Dropped root for ironic tftp server" [Undecided,New] | 06:35 |
| ashestakov | can i revert root user for that container? | 06:36 |
| rahulshr_ | for me ironic-pxe failed as well | 06:37 |
| nihilifer | rahulshr_: try to do the following | 06:37 |
| nihilifer | docker run -ti kollaglue/ubuntu-source-rsyslog:liberty-rc1 /bin/bash | 06:38 |
| nihilifer | ls /var/lib/kolla/config_files | 06:38 |
| nihilifer | let me know whether this dir exists inside container | 06:38 |
| coolsvap | ashestakov, i think it makes sense to remove the user ironic in pxe container | 06:38 |
| coolsvap | since its not inherited from ironic-base | 06:39 |
| rahulshr_ | root@7b28853ad50c:/# ls /var/lib/kolla/config_files ls: cannot access /var/lib/kolla/config_files: No such file or directory | 06:40 |
| nihilifer | rahulshr_: my mistake | 06:40 |
| nihilifer | exit this container (ctrl+d) | 06:40 |
| nihilifer | and try | 06:40 |
| rahulshr_ | same | 06:40 |
| rahulshr_ | kolla# ls /var/lib/kolla/config_files ls: cannot access /var/lib/kolla/config_files: No such file or directory | 06:40 |
| openstackgerrit | Andrey Shestakov proposed openstack/kolla: Remove user 'ironic' from ironic-pxe https://review.openstack.org/260347 | 06:41 |
| nihilifer | docker run -ti kollaglue/ubuntu-source-rsyslog:liberty-rc1 -v /etc/kolla/rsyslog:/var/lib/kolla/config_files /bin/bash | 06:41 |
| nihilifer | ls /var/lib/kolla/config_files | 06:41 |
| ashestakov | review pls ^^ | 06:41 |
| nihilifer | try now ^^ | 06:41 |
| nihilifer | coolsvap: i have another concern about some drop root stuff | 06:42 |
| rahulshr_ | :~/kolla# docker run -ti kollaglue/ubuntu-source-rsyslog:liberty-rc1 -v /etc/kolla/rsyslog:/var/lib/kolla/config_files /bin/bash exec: "-v": executable file not found in $PATH Error response from daemon: Cannot start container 64e8b1f0f2697059d8cb7795908577182d621e24a118e8430ca54df64fb127a7: [8] System error: exec: "-v": executable file not found in $PATH | 06:42 |
| nihilifer | coolsvap: recently we removed nova user from nova-libvirt container | 06:43 |
| nihilifer | coolsvap: because it didn't work as well | 06:43 |
| nihilifer | but i'm thinking whether we should try to drop root for libvirt process at all | 06:43 |
| nihilifer | shouldn't be libvirt ran but, have some polkit config allowing nova user to do stuff, and then use nova user only in nova-compute container? | 06:44 |
| nihilifer | shoudn't be libvirt ran as root* | 06:44 |
| coolsvap | i think its wortha try | 06:46 |
| coolsvap | i will have a look | 06:46 |
| nihilifer | ok, cool | 06:46 |
| kjelly | nihilifer, coolsvap: I will fix the issue about libvirt with root. Maybe I send a patch latter. | 06:47 |
| nihilifer | kjelly: kk | 06:47 |
| nihilifer | as far as i saw yesterday | 06:47 |
| nihilifer | libvirt process is ran by root and works fine | 06:48 |
| nihilifer | but nova-compute container uses nova user | 06:48 |
| coolsvap | kjelly, alright, will wait for your patch then | 06:48 |
| nihilifer | and has problems with accessing libvirt socket | 06:48 |
| nihilifer | rahulshr_: are you sure that the "/bin/bash" part of this command was copied properly and was in the same line as the rest? | 06:49 |
| nihilifer | rahulshr_: maybe try this: | 06:51 |
| nihilifer | docker run -v /etc/kolla/rsyslog:/var/lib/kolla/config_files -ti kollaglue/ubuntu-source-rsyslog:liberty-rc1 /bin/bash | 06:51 |
| rahulshr_ | this one worked | 06:51 |
| nihilifer | ok, cool | 06:51 |
| nihilifer | are you able to ls that dit? | 06:51 |
| nihilifer | dir* | 06:51 |
| nihilifer | it means | 06:51 |
| nihilifer | ls /var/lib/kolla/config_files | 06:52 |
| rahulshr_ | yup | 06:52 |
| rahulshr_ | root@630e4942256f:/# ls /var/lib/kolla/config_files config.json rsyslog.conf | 06:53 |
| nihilifer | ok | 06:53 |
| ashestakov | guys, have you any ideas how to build container based on multiple sources? | 06:53 |
| ashestakov | e.g. nova-base + ironic-base | 06:53 |
| nihilifer | rahulshr_: the other question - you tried to run ansible only once? | 06:53 |
| rahulshr_ | you mean to say kolla-ansible? | 06:54 |
| nihilifer | rahulshr_: yes | 06:54 |
| nihilifer | ashestakov: no, it's impossible | 06:54 |
| rahulshr_ | I tried 3-4 times | 06:54 |
| nihilifer | rahulshr_: did you remove all containers before re-trying? | 06:54 |
| ashestakov | nihilifer: is it hard to implement? | 06:54 |
| nihilifer | ashestakov: it's just impossible. docker doesn't support having multiple parents | 06:55 |
| rahulshr_ | yes i found some one mentioned to remove containers and try again | 06:55 |
| rahulshr_ | so I idi | 06:55 |
| rahulshr_ | idi* | 06:55 |
| rahulshr_ | did* | 06:55 |
| openstackgerrit | Kuo-tung Kao proposed openstack/kolla: drop root for libvirt and change sock grp to nova https://review.openstack.org/260352 | 06:55 |
| nihilifer | ashestakov: what things do you need from nova-base and ironic-base? | 06:56 |
| ashestakov | ironic requires separate nova-compute process with ironic's compute_manager | 06:56 |
| ashestakov | it is why nova-compute container not works with ironic now | 06:57 |
| ashestakov | ironic code is missing inside | 06:57 |
| kjelly | nihilifer, coolsvap: I sent a patch. It seems that adding nova user to nova-libvirt group and setting socket group to nova is good idea. | 06:57 |
| *** sdake has joined #kolla | 06:58 | |
| nihilifer | kjelly: it's not a good idea! | 06:58 |
| *** coolsvap is now known as coolsvap|away | 06:58 | |
| kjelly | nihilifer: But if you don't add user to nova-libvirtd container, the uid 1000 will not exist. | 06:59 |
| nihilifer | running libvirt as root and letting to write to socket by normal user is high security risk | 06:59 |
| nihilifer | without using polkit, of course | 06:59 |
| Jeffrey4l | kjelly, at least, there is no nova user in the libvirt image. it is based on base image rather than nova-base image. | 06:59 |
| nihilifer | yep, Jeffrey4l is right | 07:00 |
| nihilifer | nova-libvirt doesn't inherit from nova-base | 07:00 |
| nihilifer | so 2 reasons against doing it like that | 07:00 |
| kjelly | nihilifer, Jeffrey4: I add nova user to it. | 07:00 |
| Jeffrey4l | i have another PS <https://review.openstack.org/#/c/259329/> for doing almost the same thing. I am trying to use polkit. But it is still WIP. | 07:01 |
| *** tzn has joined #kolla | 07:02 | |
| kjelly | To solve the issue https://review.openstack.org/#/c/259329/, Now I know two ways. One is change libvirt-socket to 777. Another is change socket group. | 07:02 |
| nihilifer | omg, you're serious with 777? | 07:03 |
| kjelly | nihilifer: No, so I don't send patch with the way. | 07:03 |
| Jeffrey4l | security is a big thing. | 07:04 |
| *** coolsvap|away is now known as coolsvap | 07:04 | |
| *** ashestakov has quit IRC | 07:06 | |
| nihilifer | the nice thing about polkit - as fara as i know, you can give an access only to some things in lobvirt | 07:06 |
| kjelly | nihilifer, Jeffrey4l: If you have good idea, send a new patch. Now I have no idea to solve it. | 07:06 |
| nihilifer | so give a specific permissions | 07:06 |
| nihilifer | https://libvirt.org/aclpolkit.html | 07:06 |
| *** tzn has quit IRC | 07:06 | |
| nihilifer | giving acces to the whole socket is too huge risk imo | 07:06 |
| Jeffrey4l | nihilifer, agree. | 07:07 |
| kjelly | nihilifer: Ok. I try the way to solve it. | 07:08 |
| nihilifer | hmmm.. we seem to have 3 people willing to work on this :P | 07:08 |
| kjelly | nihilifer: Who are working on it? I assign bug to him | 07:09 |
| nihilifer | https://blueprints.launchpad.net/kolla/+spec/drop-root | 07:10 |
| nihilifer | the work about dropping root seems to be tracked here | 07:10 |
| nihilifer | i see coolsvap in nova | 07:10 |
| nihilifer | so | 07:10 |
| nihilifer | coolsvap: you want to work on this | 07:11 |
| kjelly | https://bugs.launchpad.net/kolla/+bug/1525583 and https://bugs.launchpad.net/kolla/+bug/1528328 is the same issue. | 07:11 |
| openstack | Launchpad bug 1525583 in kolla "nova_compute can not talk to nova_libvirt" [Undecided,In progress] - Assigned to Kuo-tung Kao (jelly) (coding1314) | 07:11 |
| openstack | Launchpad bug 1528328 in kolla "Nova_compute libvirt-sock permission denied" [Undecided,New] - Assigned to Kuo-tung Kao (jelly) (coding1314) | 07:11 |
| nihilifer | or should we make a work item "libvirt"? | 07:11 |
| coolsvap | i am currently occupied in something else | 07:12 |
| coolsvap | i am okay with creating a work item for libvirt | 07:12 |
| coolsvap | rather than tracking only this change in a bug | 07:12 |
| nihilifer | coolsvap: ok, thank you | 07:12 |
| nihilifer | kjelly: ok. i'll sign the second bug as a duplicate | 07:12 |
| nihilifer | i'd suggest to use both this bug and "Partially-Implements: blueprint drop-root" for the commit solving libvirt issues | 07:13 |
| kjelly | nihilifer: Ok. | 07:14 |
| kjelly | By the way, I found fuel and Suse Cloud 5 set libvirt-socket permission to 777. it's interesting. | 07:15 |
| *** sdake has quit IRC | 07:15 | |
| nihilifer | kjelly: that's strange | 07:16 |
| nihilifer | but it also means that kolla will be the best when we will succeed with polkit :D | 07:17 |
| *** ssurana has joined #kolla | 07:17 | |
| nihilifer | kjelly: but regarding fuel | 07:18 |
| nihilifer | kjelly: you see it in openstack/puppet-* repo or explicitly somewhere in openstack/fuel-library? | 07:19 |
| nihilifer | i cannot grep it both in puppet-nova and fuel-library | 07:22 |
| kjelly | nihilifer: I have fuel env. And I just see what is the permission in fuel env. | 07:24 |
| kjelly | nihilifer: I am not good at puppet. So I don't know how to find it from code. | 07:24 |
| rahulshr_ | nihilifer should I build the images again? | 07:25 |
| *** shardy has joined #kolla | 07:25 | |
| openstackgerrit | Jeffrey Zhang proposed openstack/kolla: Rename the kolla-build section to DEFAULT https://review.openstack.org/260363 | 07:30 |
| *** rahulshr_ has quit IRC | 07:35 | |
| *** rahulshr_ has joined #kolla | 07:35 | |
| *** rahulshr_ has quit IRC | 07:57 | |
| *** tzn has joined #kolla | 08:03 | |
| *** tzn has quit IRC | 08:08 | |
| *** tfukushima has joined #kolla | 08:10 | |
| *** coolsvap is now known as coolsvap|away | 08:19 | |
| *** coolsvap|away is now known as coolsvap | 08:21 | |
| *** unicell1 has joined #kolla | 08:31 | |
| *** unicell has quit IRC | 08:31 | |
| *** gfidente has joined #kolla | 08:40 | |
| *** gfidente has joined #kolla | 08:40 | |
| *** salv-orlando has joined #kolla | 08:40 | |
| *** akwasnie has joined #kolla | 08:40 | |
| *** kproskurin has joined #kolla | 08:41 | |
| *** egonzalez has joined #kolla | 08:44 | |
| *** diga has quit IRC | 08:48 | |
| *** salv-orl_ has joined #kolla | 08:51 | |
| *** salv-orl_ has quit IRC | 08:51 | |
| *** salv-orlando has quit IRC | 08:51 | |
| *** salv-orlando has joined #kolla | 08:52 | |
| *** neilus has quit IRC | 08:54 | |
| *** neilus has joined #kolla | 08:54 | |
| openstackgerrit | Jeffrey Zhang proposed openstack/kolla: use oslo.config instead of raw argparse.ArgumentParser https://review.openstack.org/260389 | 08:54 |
| *** salv-orl_ has joined #kolla | 08:55 | |
| *** salv-orlando has quit IRC | 08:58 | |
| *** Johanna_ has joined #kolla | 09:02 | |
| *** tzn has joined #kolla | 09:04 | |
| *** tzn has quit IRC | 09:09 | |
| *** mbound_ has joined #kolla | 09:09 | |
| *** tzn has joined #kolla | 09:22 | |
| *** ssurana has quit IRC | 09:22 | |
| *** ashestakov has joined #kolla | 09:34 | |
| *** rahulshr has joined #kolla | 09:39 | |
| *** diga has joined #kolla | 09:49 | |
| *** tzn has quit IRC | 09:49 | |
| coolsvap | ashestakov, checking the gates once before i approve the pxe patch | 09:50 |
| *** macsz has joined #kolla | 09:51 | |
| ashestakov | coolsvap: thanks | 09:52 |
| coolsvap | ashestakov, np | 09:52 |
| openstackgerrit | Michal Rostecki proposed openstack/kolla-mesos: Improvements in documentation https://review.openstack.org/260423 | 09:57 |
| ashestakov | so, no ideas how to build container with multiple bases? | 10:03 |
| nihilifer | hmm... maybe i have something. we use jinja2 to generate dockerfiles from templates, so maybe we can use somwhow {% include "[...]" %} | 10:05 |
| nihilifer | it's probably trivial to use | 10:05 |
| nihilifer | but it may be not so easy to eventually re-design the other containers | 10:06 |
| nihilifer | if we would like to introduce includes not only for ironic | 10:06 |
| nihilifer | i'd suggest to keep the re-usable fragments in directory like docker/commons/[...] | 10:08 |
| nihilifer | and try to do jinja2 include | 10:08 |
| *** rahulshr has quit IRC | 10:10 | |
| *** tfukushima has quit IRC | 10:16 | |
| *** diga has quit IRC | 10:24 | |
| *** vincent_1dk is now known as vincent_vdk | 10:47 | |
| *** Johanna_ has quit IRC | 10:50 | |
| *** Johanna_ has joined #kolla | 10:51 | |
| openstackgerrit | Merged openstack/kolla: Remove user 'ironic' from ironic-pxe https://review.openstack.org/260347 | 10:53 |
| coolsvap | Jeffrey4l, have a look at the pep8 failures in https://review.openstack.org/260389 and test if the images build with the changes | 10:54 |
| coolsvap | not sure if it failed with your changes or not | 10:54 |
| *** Johanna_ has quit IRC | 10:55 | |
| Jeffrey4l | coolsvap, ./tools/build.py:626:1: E302 expected 2 blank lines, found 1 | 10:55 |
| Jeffrey4l | it should be related to my changes. | 10:55 |
| Jeffrey4l | i will push another one to fix it. | 10:55 |
| coolsvap | not the pep8 issue, the image build also failed in the gate results | 10:56 |
| Jeffrey4l | coolsvap, I am checking it. | 10:57 |
| coolsvap | just wanted to let you know | 10:57 |
| Jeffrey4l | coolsvap, ok, thanks a lot. | 10:57 |
| *** rahulshr has joined #kolla | 11:06 | |
| rahulshr | nihilifer I am still stuck | 11:06 |
| rahulshr | v | 11:07 |
| rahulshr | TASK: [common | Starting Ansible container] *********************************** failed: [localhost] => {"changed": true, "failed": true} msg: Docker API Error: Cannot start container ad235b682b8391b54d661d2f49cf8768d055931ff9f421562738b40316126c11: [8] System error: not a directory FATAL: all hosts have already failed -- aborting PLAY RECAP ******************************************************************** to retry, | 11:07 |
| *** akwasnie has quit IRC | 11:08 | |
| *** shardy has quit IRC | 11:14 | |
| *** akwasnie has joined #kolla | 11:19 | |
| *** egonzalez has quit IRC | 11:23 | |
| openstackgerrit | Jeffrey Zhang proposed openstack/kolla: use oslo.config instead of raw argparse.ArgumentParser https://review.openstack.org/260389 | 11:27 |
| *** tzn has joined #kolla | 11:30 | |
| *** Jeffrey4l has quit IRC | 11:31 | |
| *** macsz has quit IRC | 11:34 | |
| *** egonzalez has joined #kolla | 11:36 | |
| *** Jeffrey4l has joined #kolla | 11:43 | |
| *** salv-orl_ has quit IRC | 11:44 | |
| openstackgerrit | Jeffrey Zhang proposed openstack/kolla: use oslo.config instead of raw argparse.ArgumentParser https://review.openstack.org/260389 | 11:46 |
| coolsvap | Jeffrey4l, i am not able to build images after using changes in patch set 260389 ^^ | 12:06 |
| *** egonzalez has quit IRC | 12:06 | |
| Jeffrey4l | coolsvap, what's the error? | 12:08 |
| coolsvap | unrecognized argument | 12:09 |
| coolsvap | for individual image build | 12:09 |
| coolsvap | like if i try to build keystone | 12:09 |
| coolsvap | build.py: error: unrecognized arguments: keystone | 12:09 |
| Jeffrey4l | coolsvap, could u show me the run command line? | 12:09 |
| coolsvap | wait creating paste | 12:10 |
| coolsvap | Jeffrey4l, http://paste.openstack.org/show/482504/ | 12:10 |
| * Jeffrey4l is reading | 12:11 | |
| Jeffrey4l | ok. My mistake. I never know the build script can work like that. | 12:12 |
| Jeffrey4l | I will try to fix it. | 12:12 |
| Jeffrey4l | coolsvap, ^ | 12:12 |
| coolsvap | Jeffrey4l, np | 12:12 |
| coolsvap | just move it to WIP for the time being, its good to use the oslo in here | 12:14 |
| Jeffrey4l | coolsvap, I found root cause. I can fix it in several minutes. | 12:16 |
| coolsvap | Jeffrey4l, np then | 12:17 |
| coolsvap | :) | 12:17 |
| Jeffrey4l | coolsvap, (just what make sure: you mean that you want me to make the PS WIP?)(my english is not very good..);P | 12:18 |
| coolsvap | yes but since it does not require much time you can dont need to do it | 12:19 |
| Jeffrey4l | thanks | 12:19 |
| *** egonzalez has joined #kolla | 12:21 | |
| *** egonzalez has quit IRC | 12:29 | |
| *** rhallisey has joined #kolla | 12:41 | |
| *** rhallisey has quit IRC | 12:42 | |
| *** salv-orlando has joined #kolla | 12:45 | |
| *** sdake has joined #kolla | 12:54 | |
| openstackgerrit | Jeffrey Zhang proposed openstack/kolla: use oslo.config instead of raw argparse.ArgumentParser https://review.openstack.org/260389 | 12:58 |
| *** dims has joined #kolla | 13:01 | |
| rahulshr | I am still trying to deploy kolla using kolla-ansible but it failing for one of the containers | 13:12 |
| rahulshr | TASK: [common | Starting Ansible container] *********************************** failed: [localhost] => {"changed": true, "failed": true} msg: Docker API Error: Cannot start container ad235b682b8391b54d661d2f49cf8768d055931ff9f421562738b40316126c11: [8] System error: not a directory FATAL: all hosts have already failed -- aborting | 13:14 |
| rahulshr | I am pretty new to kolla and docker | 13:14 |
| *** klint has quit IRC | 13:15 | |
| *** akwasnie has quit IRC | 13:17 | |
| *** salv-orlando has quit IRC | 13:18 | |
| *** akwasnie has joined #kolla | 13:18 | |
| *** salv-orlando has joined #kolla | 13:18 | |
| *** sdake has quit IRC | 13:19 | |
| Jeffrey4l | rahulshr, are u solve the rsyslog issue? is it working well? | 13:20 |
| *** akwasnie has quit IRC | 13:22 | |
| Jeffrey4l | mandre, pls check the comment here https://review.openstack.org/#/c/260363/1 | 13:22 |
| Jeffrey4l | it is neccessary to check the kolla-build section to DEFAULT | 13:22 |
| *** akwasnie has joined #kolla | 13:22 | |
| Jeffrey4l | s/check/change/ | 13:23 |
| rahulshr | Jeffrey4l, not able to solve it | 13:25 |
| rahulshr | what should I check | 13:25 |
| Jeffrey4l | rahulshr, still `IOError: [Errno 2] No such file or directory: '/opt/kolla/config_files/config.json'` this issue in rsyslog? | 13:26 |
| rahulshr | yeah that is still the same | 13:27 |
| *** akwasnie has quit IRC | 13:27 | |
| * Jeffrey4l is thinking hold on. | 13:28 | |
| *** mbound_ has quit IRC | 13:30 | |
| *** tzn has quit IRC | 13:30 | |
| *** dwalsh has joined #kolla | 13:31 | |
| Jeffrey4l | rahulshr, could u show the full logs for rsyslog? | 13:31 |
| Jeffrey4l | u can create a paste here paste.openstack.org | 13:31 |
| Jeffrey4l | http://paste.openstack.org/ | 13:31 |
| *** shardy has joined #kolla | 13:34 | |
| *** rahulshr has quit IRC | 13:36 | |
| *** athomas has joined #kolla | 13:36 | |
| dmsimard | mandre: replied to the yum setup review | 13:38 |
| dmsimard | I -1'd it to prevent merge since there was a +2 already | 13:39 |
| *** mbound_ has joined #kolla | 13:46 | |
| *** rahulshr has joined #kolla | 13:51 | |
| rahulshr | Jeffrey4l, sry got dced | 13:51 |
| rahulshr | can you please tell the cmd to get full logs | 13:51 |
| *** mbound_ has quit IRC | 13:54 | |
| Jeffrey4l | rahulshr, docker logs rsyslog should work | 13:55 |
| Jeffrey4l | rahulshr, do u changed any configuration? the `/opt` folder is wired. | 13:56 |
| rahulshr | did not make any config changes | 13:57 |
| rahulshr | opt folder is empty | 13:57 |
| rahulshr | http://paste.openstack.org/show/482514/ | 13:58 |
| Jeffrey4l | rahulshr, what kolla version are u using? | 13:59 |
| rahulshr | I have clone liberty branch | 13:59 |
| Jeffrey4l | ok. | 14:00 |
| Jeffrey4l | rahulshr, u can try master branch. | 14:00 |
| Jeffrey4l | I haven't try the liberty branch.. | 14:00 |
| rahulshr | ok then do I need to build again? | 14:01 |
| Jeffrey4l | rahulshr, yep. | 14:01 |
| rahulshr | ok...genrally what ppl do they use vagrant or they deploy on bare metal? | 14:03 |
| *** sdake has joined #kolla | 14:08 | |
| *** sdake has quit IRC | 14:08 | |
| *** sdake has joined #kolla | 14:09 | |
| Jeffrey4l | i am using vagrant. But bare metal should be the same. | 14:18 |
| Jeffrey4l | rahulshr, ^ | 14:18 |
| rahulshr | ok | 14:19 |
| rahulshr | I have a 8 gb system and vagrant seems to hang on it | 14:20 |
| *** athomas has quit IRC | 14:20 | |
| *** athomas has joined #kolla | 14:20 | |
| Jeffrey4l | rahulshr, it depends. My vagrant is allocated 5G and it works well. | 14:23 |
| *** inc0 has joined #kolla | 14:31 | |
| inc0 | good morning | 14:32 |
| *** dave-mccowan has joined #kolla | 14:33 | |
| inc0 | happy Sol Invictus birthday!;) | 14:33 |
| *** jtriley has joined #kolla | 14:43 | |
| *** rahulshr has quit IRC | 14:48 | |
| dmsimard | inc0: good morning :D | 14:56 |
| *** sdake has quit IRC | 15:10 | |
| britthouser | Its the shortest day the Lord has made. =) | 15:14 |
| *** zigo_ is now known as zigo | 15:19 | |
| inc0 | and today Space-X successfully landed stage 1 of rocket | 15:21 |
| inc0 | which is cool:) | 15:21 |
| *** thumpba has joined #kolla | 15:21 | |
| kproskurin | Hello guys, just noticed what http://docs.openstack.org/developer/kolla/dev-quickstart.html is still working | 15:29 |
| kproskurin | Which is bad afaik :-) | 15:29 |
| coolsvap | kproskurin, didnt get you :-/ | 15:30 |
| kproskurin | It’s old doc | 15:31 |
| kproskurin | New one is http://docs.openstack.org/developer/kolla/quickstart.html | 15:31 |
| coolsvap | kproskurin, ohk | 15:32 |
| kproskurin | and some google search leads to the old one | 15:32 |
| inc0 | kproskurin, mind popping an patchset? | 15:34 |
| kproskurin | I not sure what to edit, ‘cos I did edit the index.rst to point to new doc | 15:34 |
| kproskurin | but old one is still there and I just not sure how to fix this | 15:34 |
| kproskurin | perfect solution will be redirect imo | 15:35 |
| openstackgerrit | Michal Jastrzebski (inc0) proposed openstack/kolla: Playbook for nova upgrade https://review.openstack.org/254395 | 15:41 |
| openstackgerrit | David Moreau Simard proposed openstack/kolla: Make the yum repositories configurable in the base image https://review.openstack.org/258017 | 15:41 |
| dmsimard | SamYaple: ^ different implementation attempt based on mandre's comment, I kind of like it | 15:41 |
| *** Jeffrey4l has quit IRC | 15:44 | |
| openstackgerrit | Merged openstack/kolla: Convert to pymysql https://review.openstack.org/260080 | 15:47 |
| coolsvap | SamYaple, i think the patch had some dependent patches yesterday | 15:47 |
| coolsvap | ^^ | 15:47 |
| ashestakov | hi guys | 15:57 |
| ashestakov | any ideas how to better run fastcgi application with apache under container? | 15:57 |
| *** cloudnautique has joined #kolla | 16:01 | |
| *** cloudnautique has quit IRC | 16:01 | |
| *** cloudnautique has joined #kolla | 16:02 | |
| *** gfidente is now known as gfidente|pto | 16:03 | |
| ashestakov | or run two main processes in container | 16:06 |
| inc0 | ashestakov, we do that with supervisord | 16:06 |
| inc0 | it's hack, but sometimes required | 16:06 |
| inc0 | look at container neutron_agents | 16:07 |
| *** dwalsh has quit IRC | 16:07 | |
| inc0 | (hopefully we can get rid of this once docker 1.10 lands but for now it is what it is | 16:08 |
| *** dwalsh has joined #kolla | 16:08 | |
| ashestakov | supervisord... ok, thanks | 16:08 |
| *** dwalsh has quit IRC | 16:08 | |
| *** dwalsh has joined #kolla | 16:11 | |
| SamYaple | coolsvap: what do you mean? | 16:14 |
| SamYaple | ashestakov: im not sure what you are planning to do, but we aren't accepting anymore supervisord implementations under Kolla unless absolutely needed. What was that for? | 16:15 |
| ashestakov | i trying to implement radosgw container | 16:17 |
| inc0 | that's actually a very noble effort, can't do without fastcgi? | 16:18 |
| ashestakov | ironic needs swift api, but swift in kolla is not convenient | 16:18 |
| ashestakov | radosgw can runs as standalone fastcgi, and apache can use mod_proxy-fcgi to proxy requests | 16:18 |
| ashestakov | or apache can run fastcgi server | 16:19 |
| ashestakov | i prefer standalone process and proxy-fcgi, like in ceph docs | 16:19 |
| dmsimard | ashestakov: careful, fastcgi support is deprecated with radosgw | 16:19 |
| dmsimard | ashestakov: they want people to use civetweb now | 16:19 |
| dmsimard | they stopped mirroring the patched mod_fastcgi they provided for apache | 16:20 |
| ashestakov | dmsimard: can you give me link? | 16:20 |
| dmsimard | we have a review for adding civetweb support in puppet-ceph: https://review.openstack.org/#/c/254548/ | 16:20 |
| dmsimard | ashestakov: will try to find it, sec | 16:21 |
| ashestakov | and where is corresponding apache config for civetweb? | 16:23 |
| *** shardy has quit IRC | 16:24 | |
| *** rahulshr has joined #kolla | 16:25 | |
| SamYaple | ashestakov: oh I have the ceph radosgw and even CephFS stuff, i just didn't drop it in Kolla since we didn't have a need for it | 16:25 |
| *** coolsvap is now known as coolsvap|away | 16:26 | |
| ashestakov | SamYaple: can you add it? | 16:26 |
| SamYaple | ashestakov: I have to Kolla-fy it, but I can, yes | 16:26 |
| SamYaple | ashestakov: do you already have a patchset you would like me to work with you on? | 16:27 |
| *** athomas has quit IRC | 16:27 | |
| ashestakov | no, i just trying to get radosgw build and deploy | 16:27 |
| SamYaple | well we don't build ceph from source in Kolla at all | 16:27 |
| SamYaple | we just use the packaged ceph | 16:28 |
| ashestakov | also, ironic requires swift api to get images by tempurl, glance should use rgw as backend to get it works | 16:28 |
| ashestakov | is possible to implement this in kolls? | 16:29 |
| SamYaple | it is possible to back glance by rgw, yes | 16:29 |
| SamYaple | however, I don't understand the need for that | 16:30 |
| SamYaple | glance will stream the image from the backing store (rbd) to ironic via api | 16:30 |
| ashestakov | images uploaded by glance, should be available via swift api (via rgw) | 16:30 |
| dmsimard | ashestakov: The link I have for the thread on ceph-users looks dead, asked someone about it | 16:30 |
| ashestakov | SamYaple: have you implemented rgw with fcgi? | 16:31 |
| SamYaple | i guess im confused ashestakov, are you saying the glance images can only be acessed by speaking directly to swift (bypassing glance) in ironic? | 16:31 |
| dmsimard | SamYaple: ironic generates a swift tempurl for the images and downloads it over http | 16:32 |
| dmsimard | it "computes" the path to the images with a swift key | 16:32 |
| dmsimard | it doesn't download the image through glance | 16:32 |
| ashestakov | ironic should provide swift tempurl http://docs.openstack.org/kilo/config-reference/content/object-storage-tempurl.html | 16:32 |
| ashestakov | so these tempurls available only in swift api | 16:32 |
| ashestakov | when glance uses rdb as backend, images are not visible vi rgw | 16:33 |
| SamYaple | ah i follow. yea that makes sense. i wasnt aware thats how that worked | 16:33 |
| SamYaple | luckily we can have multiple backends with glance | 16:33 |
| SamYaple | rbd and rgw | 16:33 |
| ashestakov | glance backend calls rgw or swift? | 16:34 |
| SamYaple | ashestakov: "Starting with Mitaka the Bare Metal service also supports Ceph Object Gateway" | 16:34 |
| SamYaple | direct rgw | 16:34 |
| dmsimard | oh that's cool | 16:34 |
| SamYaple | ashestakov: yes i must be using the fastcgi since thats te compiled version of apache2 that ceph provides | 16:35 |
| SamYaple | ill double check its been a while since i dug into that | 16:35 |
| ashestakov | "Starting with Mitaka the Bare Metal service also supports Ceph Object Gateway (RADOS Gateway)" | 16:35 |
| SamYaple | if we must use supervisord there, thats not teh end of the world | 16:35 |
| ashestakov | it is just name of backend, anyway it needs swift api | 16:36 |
| SamYaple | hey inc0 did you see https://review.openstack.org/#/c/259624/ ? Look at the latest pastes | 16:36 |
| inc0 | SamYaple, yeah I am reviewing it slowly. lot's of files there;) | 16:36 |
| ashestakov | maybe run separate containers for apache and rgw? | 16:37 |
| SamYaple | ashestakov: not quite, with the way it _can_ work now, keystone is not in the picture | 16:37 |
| SamYaple | rgw with keystone auth is... bad | 16:37 |
| ashestakov | rgw not requires keystone, it can use ceph auth like in doc http://docs.ceph.com/docs/master/radosgw/config/ | 16:38 |
| *** dims has quit IRC | 16:42 | |
| *** ashestakov_ has joined #kolla | 16:42 | |
| *** ashestakov has quit IRC | 16:45 | |
| *** thumpba_ has joined #kolla | 16:46 | |
| ashestakov_ | meanwhile, can i use custom repositories to build containers? | 16:47 |
| inc0 | ashestakov_, sure, /etc/kolla/kolla-build.conf | 16:48 |
| inc0 | I mean, to some extend | 16:48 |
| inc0 | it works for source builds | 16:48 |
| ashestakov_ | i mean deb repos for ubuntu | 16:48 |
| inc0 | ah then I don't think so | 16:49 |
| *** thumpba has quit IRC | 16:49 | |
| ashestakov_ | is any plan to support custom repos in future? | 16:50 |
| inc0 | ashestakov_, if someone writes it, I'll surely +2 it;) | 16:53 |
| *** diga has joined #kolla | 16:53 | |
| inc0 | its feature we would like, that's for sure, just we don't have manpower to make it happen | 16:53 |
| ashestakov_ | ok, and i asked before how to build container with multiple bases (nova-base + ironic-base), any fresh ideas?) | 16:57 |
| inc0 | uhh...I don't think it's possible with docker | 16:58 |
| inc0 | you'd need to create new base with features you need | 16:58 |
| inc0 | if you need packages from both consider to move them to openstack-base | 16:59 |
| dmsimard | ashestakov_: I actually have a review up for that | 16:59 |
| dmsimard | ashestakov_: https://review.openstack.org/#/c/258017/ | 17:00 |
| dmsimard | This is for RPM repositories, though - if you build for Ubuntu this doesn't add support for it | 17:00 |
| dmsimard | inc0: ^ :) | 17:00 |
| *** kproskurin has quit IRC | 17:01 | |
| openstackgerrit | David Moreau Simard proposed openstack/kolla: Make the yum repositories configurable in the base image https://review.openstack.org/258017 | 17:02 |
| *** blahRus has joined #kolla | 17:02 | |
| inc0 | yeah I thought I saw it somewhere | 17:02 |
| dmsimard | oh no, I had put arguments in the wrong alphabetical order :( | 17:02 |
| ashestakov_ | dmsimard: cool! will you add same for ubuntu? | 17:04 |
| ashestakov_ | also will nice to support priorities for repos | 17:05 |
| dmsimard | Ubuntu is not at parity in general (binary support is still in review at https://review.openstack.org/#/c/260069/) | 17:06 |
| dmsimard | Someone can contribute repository configuration support with a similar logic to the RPM one once the binary containers land | 17:06 |
| dmsimard | that someone likely won't be me as I have my hands full :( | 17:07 |
| openstackgerrit | Michal Jastrzebski (inc0) proposed openstack/kolla: Playbook for keystone upgrade https://review.openstack.org/257568 | 17:10 |
| *** dims has joined #kolla | 17:12 | |
| *** unicell1 has quit IRC | 17:19 | |
| *** neilus has quit IRC | 17:20 | |
| *** neilus has joined #kolla | 17:24 | |
| inc0 | SamYaple, ad variable include | 17:29 |
| *** neilus has quit IRC | 17:29 | |
| inc0 | I don't think this is possible with Ansible | 17:29 |
| *** diga has quit IRC | 17:33 | |
| *** neilus has joined #kolla | 17:41 | |
| *** neilus has quit IRC | 17:46 | |
| SamYaple | inc0: i wasnt asking if it is possible. I was saying it is, and you need to do it | 17:46 |
| SamYaple | ashestakov_: yea we are doing custom repos for RPM right now, it will be easy to do it for ubuntu | 17:48 |
| inc0 | SamYaple, "Note that you cannot do variable substitution when including one playbook inside another." from http://docs.ansible.com/ansible/playbooks_roles.html | 17:49 |
| inc0 | and no, it does not work as you described in comment, tested it first | 17:50 |
| SamYaple | you absolutely can | 17:50 |
| SamYaple | you cannot in ansible 2.0 | 17:50 |
| inc0 | I don't run 2.0 | 17:51 |
| SamYaple | but you have different ways to achive that in 2.0 | 17:51 |
| SamYaple | then you can, you should try it | 17:51 |
| inc0 | I tried oit | 17:51 |
| SamYaple | you can do "include: {{ variable }}.yml" | 17:51 |
| SamYaple | that works | 17:51 |
| *** dwalsh has quit IRC | 17:52 | |
| inc0 | http://paste.openstack.org/show/482535/ >> ERROR: file could not read: /home/inc0/kolla/ansible/roles/keystone/tasks/{{ playbook }}.yml | 17:52 |
| inc0 | ansible --version >> 1.9.4 | 17:53 |
| SamYaple | Ill gve you a paste on how to do it in a minute | 17:54 |
| inc0 | ok thanks | 17:55 |
| *** openstackgerrit has quit IRC | 18:02 | |
| *** openstackgerrit has joined #kolla | 18:03 | |
| *** ssurana has joined #kolla | 18:05 | |
| SamYaple | http://paste.openstack.org/show/482536/ inc0 | 18:15 |
| SamYaple | inc0: http://paste.openstack.org/show/482539/ | 18:15 |
| *** dwalsh has joined #kolla | 18:17 | |
| inc0 | so it's for hosts right? | 18:17 |
| SamYaple | inc0: no its you didnt use set_fact right | 18:19 |
| inc0 | debug shows otherwise | 18:20 |
| inc0 | and error msg | 18:20 |
| inc0 | ansible --version plz | 18:20 |
| SamYaple | 1.9.4 | 18:21 |
| inc0 | with -e it works, with set_fact it does not | 18:23 |
| inc0 | even tho value of {{ playbook }} is ok | 18:23 |
| SamYaple | because you cant set it with set_fact, because that sets it per host | 18:23 |
| SamYaple | put it in globals.yml and it will work | 18:24 |
| SamYaple | there are many ways to make it work | 18:24 |
| *** unicell has joined #kolla | 18:25 | |
| inc0 | yeah, if I put it in globals then upgrade.yml doesn't override its value, so it's useless | 18:30 |
| inc0 | will play around with it, but so far it's trading lots of skips for not working at all | 18:30 |
| SamYaple | im telling you inc0, it absolutely works, you need to do it right though | 18:31 |
| SamYaple | where are you still hung up? | 18:31 |
| inc0 | so if I put it into globals.yml | 18:31 |
| inc0 | it works...only I have no way to override it when using upgrade.yml play instead of site.yml | 18:32 |
| SamYaple | nien | 18:32 |
| inc0 | and we need to do it because that's the idea behind this variable in the first place | 18:32 |
| SamYaple | use vars on the include | 18:32 |
| SamYaple | or i guess when you call the role in this case | 18:32 |
| SamYaple | there are like thirty ways to solve this | 18:33 |
| SamYaple | if you would like i can push a patchset with teh change im talking about | 18:33 |
| inc0 | I'll push what I have and you can make it work | 18:33 |
| inc0 | ok? | 18:33 |
| SamYaple | fair enough | 18:33 |
| SamYaple | team work! | 18:33 |
| openstackgerrit | Michal Jastrzebski (inc0) proposed openstack/kolla: Playbook for keystone upgrade https://review.openstack.org/257568 | 18:35 |
| inc0 | there you go, it's keystone, but whatever we get there I'll port to nova | 18:35 |
| inc0 | we can mitigate whole thing by not specifying this variable and on "kolla-ansible upgrade" add -e playbook=upgrade | 18:37 |
| inc0 | but well... | 18:37 |
| inc0 | I don't like that | 18:37 |
| SamYaple | no no the issue is you have the variable specified in the globals.yml | 18:37 |
| SamYaple | ill show you | 18:37 |
| inc0 | I know that | 18:38 |
| inc0 | otherwise it's set up on host | 18:38 |
| inc0 | and we need it earlier | 18:38 |
| inc0 | either way, it has to be set up in -e | 18:38 |
| *** chandankumar has joined #kolla | 18:43 | |
| *** jtriley has quit IRC | 18:46 | |
| *** salv-orlando has quit IRC | 18:48 | |
| *** rahulshr has quit IRC | 18:57 | |
| SamYaple | inc0: you are wrong | 18:59 |
| SamYaple | the only reason you think that is because of ansibles variable precedence | 19:00 |
| inc0 | well you know what we want to achieve, if you think of a way let's implement it | 19:02 |
| SamYaple | inc0: you arent stopping the keystone containers at all in this patchset? | 19:04 |
| inc0 | no | 19:04 |
| SamYaple | you need to stop them to upgrade the db, no? | 19:05 |
| inc0 | unless they do destructive migration, no | 19:05 |
| SamYaple | do they? how do we know? | 19:05 |
| inc0 | they didn't so far | 19:05 |
| inc0 | we gate on it I guess | 19:05 |
| inc0 | later on | 19:05 |
| SamYaple | "Progress has been made on supporting live sql migrations, however it is still recommended to bring down the heat service for the duration of the upgrade." | 19:05 |
| SamYaple | doesnt taht advice hold true for all services? | 19:06 |
| inc0 | SamYaple, I spoke with Dolph about that | 19:06 |
| inc0 | no, it doesnt | 19:06 |
| SamYaple | oh ok if you talked to dolph im good to go | 19:06 |
| inc0 | http://lists.openstack.org/pipermail/openstack-dev/2015-December/081467.html | 19:07 |
| inc0 | that's for nova | 19:07 |
| inc0 | we'll test it in and out | 19:07 |
| inc0 | but for now it seems ok | 19:07 |
| SamYaple | so the deal at the summit was we do a hard stop of the service until we know for sure. are yo usaying you talked to dolph about keystone specifically? | 19:08 |
| inc0 | yes, each service should be treated differently | 19:09 |
| SamYaple | so.. did you talk to dolph specifically about doing a live keystone upgrade? | 19:10 |
| SamYaple | or are you just guessing here? | 19:10 |
| inc0 | yes, I talked with Dolph specifically about live keystone upgrade, I sit 2m from him | 19:11 |
| SamYaple | that was the answer to my question | 19:11 |
| inc0 | if something bad merges, we want to catch that | 19:11 |
| inc0 | hence gate | 19:12 |
| SamYaple | thats not encoraging, are you saying we are gating a way no one else is? | 19:13 |
| inc0 | I'm saying non-destructive migration merging is policy-enforced | 19:13 |
| inc0 | and nova does gating on it | 19:13 |
| SamYaple | im asking about keystone | 19:13 |
| SamYaple | im looking at that keystone one rihgt now | 19:14 |
| inc0 | but because it is "just" a policy it potentially might break | 19:14 |
| inc0 | it wont right now as there were NO destructive migrations | 19:14 |
| SamYaple | as long as it breaks everyone, i am ok, if it just breaks _US_ thats not ok | 19:14 |
| inc0 | and I'm unwilling to put forced downtime on EVERYTHING unless we need to | 19:14 |
| *** dmsimard is now known as caffeine | 19:14 | |
| *** caffeine is now known as dmsimard | 19:14 | |
| SamYaple | you cant make unilateral decisions for Kolla when we agreed to different things at the summit inc0 | 19:14 |
| SamYaple | no matter your personal opinion | 19:15 |
| inc0 | it will break everyone doing upgrade, yes | 19:15 |
| inc0 | SamYaple, let's fix one problem at the time ok? right now we can do live upgrades | 19:15 |
| SamYaple | im just pointing out thats not the first time you have said you are going against what we all agreed to | 19:16 |
| SamYaple | thats not cool | 19:16 |
| inc0 | we changed our approach on ML | 19:16 |
| inc0 | after me proposing per-case approach | 19:17 |
| inc0 | this kinda invalidates what we agreed because we agreed on universal upgrade play | 19:17 |
| inc0 | with per-case approach we can have this kind of freedom | 19:17 |
| SamYaple | thats not what we agreed to, but im fine with this approach, im not fine with the "ill do it my way because i want to" attitude | 19:17 |
| inc0 | SamYaple, please care about your own attitude | 19:18 |
| inc0 | this was discussed on ML none of cores had problems with it | 19:18 |
| inc0 | if you have problem with this approach, let's discuss whats wrong with it and how to solve these problems | 19:18 |
| *** jtriley has joined #kolla | 19:31 | |
| *** dims has quit IRC | 19:45 | |
| *** salv-orlando has joined #kolla | 19:47 | |
| SamYaple | so inc0 I have done tweaks that work for me to that keystoen patchset | 19:47 |
| SamYaple | ive made it DRY where it was not before | 19:47 |
| SamYaple | pushing it momentarily, running it once more | 19:47 |
| inc0 | thanks | 19:47 |
| openstackgerrit | Sam Yaple proposed openstack/kolla: Playbook for keystone upgrade https://review.openstack.org/257568 | 19:48 |
| inc0 | SamYaple, add yourself as co-author plz | 19:49 |
| SamYaple | ok ill do that in a later patchset after yo ureview this one | 19:49 |
| SamYaple | if you notice, there is no upgrades dir as we reuse existing tasks to acomplish the same thing | 19:50 |
| inc0 | yeah, noticed that | 19:50 |
| inc0 | hmm....when I run upgrade it still tries to do bootstrapping of users | 19:53 |
| inc0 | - debug: msg="{{ playbook }}" >> install regardless of play I run | 19:55 |
| SamYaple | inc0: you should rebuild your containers, no? https://github.com/openstack/kolla/blob/master/docker/keystone/extend_start.sh | 19:55 |
| inc0 | I did that | 19:55 |
| SamYaple | im not sure how it can then | 19:56 |
| inc0 | I mean it still runs ansible tasks with it | 19:56 |
| SamYaple | oh. hold on it probably needs a rebase | 19:56 |
| SamYaple | let me check | 19:56 |
| inc0 | it doesn't rebase was done | 19:56 |
| inc0 | this morning | 19:56 |
| inc0 | try to add debug at the end of keystone's main.yml | 19:57 |
| SamYaple | can you do a pastebin? I do not see those results | 19:58 |
| openstackgerrit | Sam Yaple proposed openstack/kolla: Remove unneeded variables passed into keystone https://review.openstack.org/260693 | 20:01 |
| inc0 | SamYaple, ok, it was my mistake, globals.yml had leftovers | 20:02 |
| SamYaple | ah yes | 20:02 |
| SamYaple | so the reason that _wasnt_ working for you is because the globals.yml is technically a cli variable which has highest precedence, you can't override it withs 'vars:' | 20:03 |
| SamYaple | so we pass that variable with vars on the role only | 20:03 |
| *** dims has joined #kolla | 20:04 | |
| inc0 | yeah I figured it out | 20:04 |
| inc0 | I'm thinking to name it in different way tho | 20:04 |
| SamYaple | i cant figure out how to update commit message online in new gerrit :( | 20:04 |
| inc0 | "playbook" is misleading | 20:05 |
| inc0 | I don't think you can | 20:05 |
| SamYaple | i agree, but i didnt immediately think of something better | 20:05 |
| SamYaple | instead of "playbook" what about "task"? | 20:05 |
| inc0 | just update it during commit --ammend | 20:05 |
| SamYaple | yea but i hate that they removed tahtfeature... | 20:05 |
| inc0 | task is better | 20:06 |
| inc0 | still feels not-perfect, but good enough imho | 20:06 |
| SamYaple | i can change to task and add co-author | 20:06 |
| inc0 | do that plz | 20:06 |
| SamYaple | we can do "mission".... | 20:07 |
| inc0 | ... | 20:07 |
| SamYaple | "ansible" is from a book about war anyway | 20:07 |
| inc0 | quest | 20:07 |
| SamYaple | just saying | 20:07 |
| SamYaple | venture | 20:07 |
| SamYaple | job ? | 20:07 |
| inc0 | campaign? | 20:07 |
| SamYaple | task or job are my thoughts | 20:07 |
| inc0 | make it job | 20:07 |
| SamYaple | k | 20:08 |
| inc0 | task is ansible term, we don't want to confuse folks | 20:08 |
| inc0 | or action | 20:08 |
| inc0 | SamYaple, ^ | 20:08 |
| inc0 | how about "action"? | 20:08 |
| SamYaple | is action protected? | 20:08 |
| SamYaple | nope were good | 20:09 |
| SamYaple | action is fine | 20:09 |
| inc0 | cool | 20:09 |
| inc0 | do these and I'll +1 | 20:09 |
| inc0 | also try it out plz, I don't have access to proper multi-node | 20:09 |
| inc0 | for now at leasy | 20:10 |
| SamYaple | what do you think of deploy vs install? | 20:10 |
| SamYaple | i think it should be deploy since we call that action deploy elsewhere | 20:10 |
| inc0 | I guess | 20:10 |
| inc0 | yeah | 20:10 |
| inc0 | becayse then with kolla-ansible we'll have either "deploy" or "upgrade" | 20:11 |
| SamYaple | yea | 20:11 |
| *** dims_ has joined #kolla | 20:11 | |
| openstackgerrit | Sam Yaple proposed openstack/kolla: Playbook for keystone upgrade https://review.openstack.org/257568 | 20:12 |
| inc0 | hey dims_ btw this fix doesn't seem to help:/ | 20:12 |
| inc0 | I'll dig more | 20:12 |
| dims_ | inc0 : the syslog one? | 20:12 |
| inc0 | yeah | 20:12 |
| inc0 | still can't see traces | 20:12 |
| inc0 | I'll debug some more | 20:13 |
| SamYaple | yea dims_ traces go to stderr | 20:13 |
| SamYaple | so we need to close stdin/out/err properly to redirect it? | 20:13 |
| SamYaple | do* | 20:13 |
| inc0 | SamYaple, disclaimer - I tested it out outside container and problem seems to be present there | 20:13 |
| inc0 | as well | 20:13 |
| SamYaple | well the one thing we arent doing that regular daemons do is closing stdin/out/err | 20:14 |
| *** dims has quit IRC | 20:14 | |
| inc0 | dhellman thought it's about way oslo.log gets output from traceback | 20:14 |
| inc0 | hmm...I did that with running service on my terminal, so strout/err was open as well | 20:15 |
| SamYaple | yup | 20:15 |
| SamYaple | just a thought, no evidence to back that up | 20:15 |
| inc0 | worth to try | 20:15 |
| SamYaple | inc0: did you make it through my config patch? | 20:16 |
| *** tzn has joined #kolla | 20:16 | |
| inc0 | omw back to it | 20:16 |
| *** neilus has joined #kolla | 20:18 | |
| *** neilus has quit IRC | 20:23 | |
| inc0 | SamYaple, so way it works - instead of making templates on dest host and running template there, we create it on deployment node and ship result? | 20:30 |
| *** tzn has quit IRC | 20:32 | |
| SamYaple | yup | 20:33 |
| *** smartshader has quit IRC | 20:37 | |
| dims_ | inc0 : just tested this code - https://gist.github.com/dims/226218dd289857f6c933 | 20:41 |
| dims_ | seems to work on my os x | 20:41 |
| inc0 | with rsyslog? | 20:41 |
| SamYaple | dims_: that works for me too, ubuntu 14.04 | 20:42 |
| inc0 | hmm | 20:42 |
| inc0 | Ala mentioned something about code calling log.error instead of exception | 20:43 |
| inc0 | let me check few things | 20:44 |
| dims_ | inc0 : doesn't log.error need an explicit exc_info? | 20:45 |
| inc0 | dims_, maybe, and if code uses it incorrectly, that might be reason for it | 20:46 |
| inc0 | or as SamYaple mentioned, maybe we need to close stderr | 20:46 |
| dims_ | ok, you know where to find me :) | 20:46 |
| inc0 | yup, thanks a lot! | 20:48 |
| *** salv-orlando has quit IRC | 20:49 | |
| SamYaple | thanks dims_ | 20:50 |
| *** dwalsh has quit IRC | 20:51 | |
| *** neilus has joined #kolla | 20:52 | |
| *** gfidente|pto has quit IRC | 20:57 | |
| *** cloudnautique has quit IRC | 20:57 | |
| *** cloudnautique has joined #kolla | 21:02 | |
| *** tzn has joined #kolla | 21:32 | |
| openstackgerrit | Sam Yaple proposed openstack/kolla: Remove unneeded variables passed into keystone https://review.openstack.org/260693 | 21:32 |
| *** tzn has quit IRC | 21:38 | |
| *** mbound_ has joined #kolla | 22:05 | |
| *** mbound_ has quit IRC | 22:07 | |
| openstackgerrit | Sam Yaple proposed openstack/kolla: Remove unneeded variables passed into keystone https://review.openstack.org/260693 | 22:19 |
| openstackgerrit | Sam Yaple proposed openstack/kolla: Pin shade and os-client-config https://review.openstack.org/260732 | 22:19 |
| SamYaple | inc0: i split up the patch | 22:19 |
| inc0 | damn Death Magnetic is one cool album | 22:21 |
| *** ashestakov has joined #kolla | 22:22 | |
| *** tzn has joined #kolla | 22:24 | |
| SamYaple | its ok | 22:25 |
| SamYaple | its better than some of there other albums | 22:25 |
| *** tzn has quit IRC | 22:25 | |
| *** cloudnautique has quit IRC | 22:35 | |
| *** jtriley has quit IRC | 22:44 | |
| *** bradjones has quit IRC | 23:10 | |
| *** inc0 has quit IRC | 23:11 | |
| *** cloudnautique has joined #kolla | 23:11 | |
| *** bradjones has joined #kolla | 23:13 | |
| *** bradjones has quit IRC | 23:13 | |
| *** bradjones has joined #kolla | 23:13 | |
| *** dims has joined #kolla | 23:18 | |
| *** dims_ has quit IRC | 23:19 | |
| *** ashestakov has quit IRC | 23:33 | |
| *** cloudnautique has quit IRC | 23:40 | |
| *** salv-orlando has joined #kolla | 23:47 | |
| *** Jeffrey4l has joined #kolla | 23:55 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!