*** diogogmt has quit IRC | 00:00 | |
*** cloudnautique has quit IRC | 00:10 | |
*** achanda has joined #kolla | 00:21 | |
*** tpot has quit IRC | 00:44 | |
*** dims_ has joined #kolla | 00:50 | |
*** vinkman has quit IRC | 00:53 | |
*** dims_ has quit IRC | 00:55 | |
*** dims_ has joined #kolla | 00:58 | |
*** erkules_ has joined #kolla | 01:21 | |
*** erkules has quit IRC | 01:23 | |
*** SourabhP has quit IRC | 01:26 | |
*** dims_ has quit IRC | 01:30 | |
*** mfalatic has quit IRC | 01:39 | |
*** mfalatic has joined #kolla | 01:40 | |
*** loth has quit IRC | 01:41 | |
*** loth has joined #kolla | 01:41 | |
*** vinkman has joined #kolla | 01:51 | |
*** unicell has quit IRC | 02:18 | |
*** CBR09 has joined #kolla | 02:19 | |
*** tfukushima has joined #kolla | 02:24 | |
CBR09 | Sam, you around?. I have a question for ceph | 02:39 |
---|---|---|
*** tpot has joined #kolla | 02:43 | |
*** sdake has joined #kolla | 02:49 | |
*** achanda has quit IRC | 02:50 | |
*** klint has joined #kolla | 02:54 | |
sdake | evenint folks | 03:07 |
*** sdake has quit IRC | 03:09 | |
*** sdake has joined #kolla | 03:10 | |
openstackgerrit | Kuo-tung Kao proposed openstack/kolla: eval var for brace expansion in cleanup-conatiners https://review.openstack.org/231293 | 03:11 |
*** sdake has quit IRC | 03:13 | |
SamYaple | CBR09: I am around | 03:22 |
CBR09 | with osd | 03:28 |
CBR09 | if I use partition | 03:28 |
CBR09 | is it same as disk ? | 03:29 |
CBR09 | for disk, only type: parted $DISK -s -- mklabel gpt mkpart KOLLA_CEPH_OSD_BOOTSTRAP 1 -1 | 03:30 |
openstackgerrit | Sam Yaple proposed openstack/kolla: Bring Kolla inline with FHS https://review.openstack.org/229761 | 03:30 |
openstackgerrit | Sam Yaple proposed openstack/kolla: Common start.sh https://review.openstack.org/229371 | 03:30 |
CBR09 | and or partition, is it sme? | 03:30 |
CBR09 | same | 03:30 |
SamYaple | CBR09: if I detect that partition I will use the _entire_ disk | 03:32 |
SamYaple | you can't just use a single partition | 03:33 |
*** sdake has joined #kolla | 03:33 | |
*** CBR09 has quit IRC | 03:45 | |
kjelly | Can I work on the bug? https://bugs.launchpad.net/kolla/+bug/1459355 | 03:57 |
openstack | Launchpad bug 1459355 in kolla "Add a new tool to Kolla that sets up a local Docker Registry" [Wishlist,Confirmed] | 03:57 |
sdake | kjelly soundsgood :) | 03:58 |
sdake | assigned to "jelly" I assume thats you | 03:58 |
bmace | based on our setup.cfg it looks like kolla is still at 0.1.0. is the assumption that anyone that wants to use it will make up their own versioning scheme or is that version going to get bumped at some point? :) | 03:59 |
sdake | well i would go with version 1.0 if the abi was stable | 03:59 |
sdake | but im not convinced it is | 03:59 |
asalkeld | hi sdake | 04:00 |
sdake | we still need upgrade functionality | 04:00 |
sdake | hey asalkeld sup | 04:00 |
bmace | sure, so maybe 1.0 for M? | 04:00 |
asalkeld | i should chat to you sometime about kolla, i *might* get more involved this cycle | 04:00 |
sdake | yes - i may bump before then | 04:00 |
sdake | cool tell me more asalkeld | 04:00 |
asalkeld | i just need to learn a bit more | 04:00 |
asalkeld | you got some docs/info | 04:01 |
sdake | docs.openstack.org/kolla | 04:01 |
SamYaple | hey sdake | 04:01 |
sdake | hey SamYaple | 04:02 |
SamYaple | sdake: i was under the impression Liberty was our 1.0 release | 04:02 |
SamYaple | if the api isn't going to be stable then whats the point of bawking at backports? | 04:03 |
sdake | ya - i'mgame to change to 1.0 if nobody has objections | 04:03 |
SamYaple | ok cool | 04:03 |
sdake | i'll ask at wednesday's meeting | 04:03 |
SamYaple | sounds like a plan | 04:03 |
SamYaple | ive got the start.sh refactor done | 04:03 |
SamYaple | as well as a patch to actually pull stable liberty tarballs | 04:04 |
SamYaple | https://review.openstack.org/#/c/229874/ | 04:04 |
SamYaple | rather than master | 04:04 |
*** CBR09 has joined #kolla | 04:06 | |
CBR09 | so above parted command not work for partition, only for disk. | 04:06 |
SamYaple | CBR09: all of the ceph stuff expects a full disk | 04:07 |
bmace | feels good enough for a 1.0 to me :) | 04:07 |
SamYaple | hey whatever bot pops up wit hthe proposals should be updated so we see the backports | 04:08 |
CBR09 | yes, Sam | 04:08 |
*** walters has quit IRC | 04:09 | |
CBR09 | ah, I see nova-libvirt have pid=host flag | 04:10 |
CBR09 | pros for this? | 04:10 |
SamYaple | CBR09: it allows for a restart of the libvirt container or even an upgrade of docker without stopping instances | 04:11 |
SamYaple | in some cases without even losing networking ot the instance | 04:11 |
CBR09 | Oh, very nice | 04:14 |
CBR09 | and for osd, I alse see pid = host | 04:15 |
*** walters has joined #kolla | 04:16 | |
SamYaple | thats to work around what i believe is a bug in ceph | 04:16 |
SamYaple | cant convince them though | 04:16 |
CBR09 | I understand | 04:18 |
kjelly | Sam Yaple, sdake: If I work on the bug, Do I need to write blurpint first? https://bugs.launchpad.net/kolla/+bug/1459355 | 04:20 |
openstack | Launchpad bug 1459355 in kolla "Add a new tool to Kolla that sets up a local Docker Registry" [Wishlist,Confirmed] - Assigned to jelly (coding1314) | 04:20 |
sdake | kjelly nah not necessary | 04:20 |
SamYaple | bug should be fine | 04:20 |
SamYaple | depends on what you ahve planned though | 04:20 |
SamYaple | i mean a docker registry is setup with a single command already | 04:20 |
SamYaple | docker run -d registry | 04:21 |
*** cloudnautique has joined #kolla | 04:21 | |
kjelly | sdake, SamYaple: I see. | 04:22 |
SamYaple | when ryan initially proposed that bug I believe he was saying Kolla itself should have a tool to build and push to a registry | 04:23 |
SamYaple | which is a bit more intense | 04:23 |
SamYaple | at this point I was under the impression we just wanted docs around the subject | 04:23 |
SamYaple | not sure though | 04:23 |
openstackgerrit | Sam Yaple proposed openstack/kolla: Set sysctl values https://review.openstack.org/231297 | 04:24 |
sdake | ya docs shoudl solve it | 04:24 |
*** SourabhP has joined #kolla | 04:25 | |
kjelly | It seems that we should solve the bug using document instead of writing new script, is it? | 04:26 |
SamYaple | sdake: https://wiki.openstack.org/wiki/Documentation/HowTo#Git_commit_messages_and_backports | 04:28 |
SamYaple | we should probably start doing that as a thing | 04:28 |
SamYaple | backport: none | 04:28 |
SamYaple | backport: liberty | 04:28 |
*** banix has quit IRC | 04:31 | |
*** daneyon has joined #kolla | 04:35 | |
*** vinkman has quit IRC | 04:35 | |
*** daneyon_ has joined #kolla | 04:35 | |
*** daneyon has quit IRC | 04:39 | |
*** tfukushima has quit IRC | 04:41 | |
*** achanda has joined #kolla | 04:42 | |
*** unicell has joined #kolla | 04:42 | |
*** SourabhP_ has joined #kolla | 04:42 | |
*** SourabhP has quit IRC | 04:44 | |
*** SourabhP_ is now known as SourabhP | 04:44 | |
*** dims_ has joined #kolla | 04:46 | |
sdake | samyaple that may be a new thing - not familiar with taht | 04:47 |
sdake | i dlslike backport: none | 04:48 |
sdake | https://wiki.openstack.org/wiki/Documentation/DocImpact | 04:50 |
sdake | here is how to setup docimpact | 04:50 |
sdake | yay some docs on that :) | 04:50 |
SamYaple | sdake: DocImpact is already setup | 04:51 |
SamYaple | thats been handled | 04:51 |
SamYaple | backport: is not a new thing | 04:51 |
SamYaple | its been around since the beginning | 04:51 |
SamYaple | is just not one does 'backport: none' | 04:51 |
sdake | typically projects use launchpad to track backport potential | 04:52 |
SamYaple | well thats not entirely true | 04:52 |
SamYaple | but i prefer the backport in the commit message | 04:52 |
SamYaple | that said we are pretty invested in laucnhpad | 04:52 |
SamYaple | so either way works for me | 04:52 |
SamYaple | i guess | 04:52 |
SamYaple | without the backport in the commit message it is impossible to use git commands to track potential backports | 04:53 |
SamYaple | sdake: https://git.openstack.org/cgit/openstack-infra/project-config/tree/gerrit/projects.yaml#n1172 | 04:53 |
sdake | yup i see it thanks | 04:53 |
SamYaple | hey if you want to save me the trouble of constantly rebasing that start.sh patch you could merge it | 04:54 |
sdake | sure, i approved it this afternoon | 04:54 |
SamYaple | otherwise its going to rebase every commit | 04:54 |
SamYaple | oh i see that. | 04:55 |
SamYaple | yea i had to rebase it so i lost the votes | 04:55 |
SamYaple | sdake: should i block that start.sh patch? I tested that and the FHS patchset (next is series) at the same time | 04:58 |
SamYaple | i can't recall if one was explicitly tested without the other | 04:58 |
openstackgerrit | Merged openstack/kolla: Common start.sh https://review.openstack.org/229371 | 04:59 |
SamYaple | welp. too late | 05:00 |
*** jpeeler has quit IRC | 05:02 | |
*** SourabhP has quit IRC | 05:04 | |
*** jpeeler has joined #kolla | 05:07 | |
*** sdake_ has joined #kolla | 05:09 | |
*** dramakri has joined #kolla | 05:10 | |
*** dramakri has left #kolla | 05:11 | |
*** sdake has quit IRC | 05:13 | |
openstackgerrit | Merged openstack/kolla: Bring Kolla inline with FHS https://review.openstack.org/229761 | 05:20 |
openstackgerrit | Merged openstack/kolla: Set sysctl values https://review.openstack.org/231297 | 05:20 |
*** daneyon_ has quit IRC | 05:23 | |
*** daneyon has joined #kolla | 05:25 | |
*** daneyon_ has joined #kolla | 05:28 | |
*** daneyon has quit IRC | 05:30 | |
*** diga has joined #kolla | 05:31 | |
cloudnautique | question, is there something special that needs to be done to make ansible look at the local images instead of trying to pull from docker hub? | 05:36 |
cloudnautique | I built the images locally | 05:36 |
cloudnautique | and now ansible is failing to pull it... from docker.io | 05:36 |
*** sdake_ has quit IRC | 05:38 | |
cloudnautique | ah, do not set global.yml to pull always :) | 05:40 |
*** tfukushima has joined #kolla | 05:41 | |
*** tfukushima has quit IRC | 05:43 | |
*** tfukushima has joined #kolla | 05:43 | |
SamYaple | cloudnautique: you got it!, set pull_policy to missing | 05:48 |
cloudnautique | is it assumed that the playbooks are running as root? there are tasks for sysctl that don't have the sudo flag turned on | 05:50 |
SamYaple | cloudnautique: that is correct | 05:52 |
*** unicell has quit IRC | 05:52 | |
SamYaple | in the M cycle security (ie not running as root) is the top priority | 05:53 |
SamYaple | so containers and ansible will be able to drop privleges when not needed | 05:53 |
*** unicell has joined #kolla | 05:53 | |
cloudnautique | Seems reasonable. | 05:53 |
*** unicell1 has joined #kolla | 05:59 | |
*** unicell has quit IRC | 06:00 | |
openstackgerrit | Kuo-tung Kao proposed openstack/kolla: put conatiner name to array in cleanup-containers https://review.openstack.org/231293 | 06:01 |
*** alisonh has quit IRC | 06:04 | |
*** cloudnautique has quit IRC | 06:08 | |
kjelly | SamYaple: Could you tell me what is your bash version? I am interesting in the issue. | 06:11 |
SamYaple | kjelly: same version of bash, just run without the subshells | 06:21 |
SamYaple | kjelly: those are the oneliners i had i just dropped them into the script and wrapped them in the subshells | 06:21 |
SamYaple | the subshells were what is breaking this | 06:21 |
*** alisonh has joined #kolla | 06:22 | |
kjelly | SamYaple: I see. | 06:28 |
*** inc0 has joined #kolla | 06:32 | |
inc0 | good monring | 06:33 |
SamYaple | morning inc0 | 06:34 |
*** tpot has quit IRC | 06:46 | |
kjelly | SamYaple, sdake: Should I solve the issue by writting new document ? https://bugs.launchpad.net/kolla/+bug/1459355 | 06:46 |
openstack | Launchpad bug 1459355 in kolla "Add a new tool to Kolla that sets up a local Docker Registry" [Wishlist,Confirmed] - Assigned to jelly (coding1314) | 06:46 |
SamYaple | kjelly: that works for me | 06:47 |
inc0 | but we already have option to push from build | 06:48 |
inc0 | so it's one command to setup registry and running build.py with flag | 06:48 |
SamYaple | no acutally setting up the registry inc0 | 06:49 |
inc0 | docker run -d -p 5000:5000 --restart=always --name registry registry; build.py --namespace 127.0.0.1:5000 --push | 06:50 |
inc0 | what else? | 06:50 |
SamYaple | documentatoin about what it is how to use it what to look for etc | 06:51 |
SamYaple | and 127.0.0.1 is bad for multinode, thats kinda the point | 06:51 |
SamYaple | and you need docker flags to make that work if its insecure (which yours is) | 06:51 |
inc0 | ok, document is cool, sorry kjelly didn't read this is about docs (first coffie is still ahead of me) | 06:52 |
inc0 | I'd be against having a separate tool for that | 06:52 |
inc0 | also I'm +1 on any new docs if it's correct and at least remotely connected to topic;) | 06:53 |
kjelly | SamYaple, inc0: Ok. I will work on it. | 06:54 |
inc0 | thank you kjelly | 06:55 |
openstackgerrit | Merged openstack/kolla: put conatiner name to array in cleanup-containers https://review.openstack.org/231293 | 06:58 |
*** alisonh has quit IRC | 07:03 | |
*** alisonh has joined #kolla | 07:08 | |
*** achanda has quit IRC | 07:18 | |
*** achanda has joined #kolla | 07:22 | |
*** erkules_ is now known as erkules | 07:25 | |
*** erkules has quit IRC | 07:25 | |
*** erkules has joined #kolla | 07:25 | |
CBR09 | about https://blueprints.launchpad.net/kolla/+spec/ansible-task-generate-openrc | 07:29 |
CBR09 | I think it's good | 07:30 |
SamYaple | CBR09: so the issue here is where does the openrc end up? | 07:31 |
SamYaple | one host? all hosts? | 07:31 |
CBR09 | if it is agreed, I can do it | 07:31 |
SamYaple | i dont really like the idea of pushing admin credentials to all hosts | 07:31 |
CBR09 | I think it on controller | 07:31 |
CBR09 | and on deployment host | 07:31 |
SamYaple | right, but there is no "controller" | 07:31 |
SamYaple | and there are no tasks that run on the deployment host | 07:31 |
SamYaple | all of the hosts that are referenced are sepcific groups, but "controller" | 07:31 |
SamYaple | rather "keystone" or "glance-api" | 07:31 |
SamYaple | im all for generating an openrc file, we just haven't figured out _where_ we were going to drop it and because its got the username/password it should be secured somehow | 07:33 |
CBR09 | can put on keystone host and fetch it deployment hót | 07:33 |
CBR09 | host | 07:33 |
inc0 | CBR09, but that means you have deployment host in your inventory, and that might not be the case | 07:33 |
SamYaple | and i dont agree with putting it on the keystone host specifically at all | 07:33 |
SamYaple | really the best thing here in my opinion would be a script that could generate/print out teh openrc file | 07:34 |
inc0 | maybe a python script that will read configs and produce file? | 07:34 |
inc0 | not ansible | 07:34 |
SamYaple | ansible could do it | 07:34 |
inc0 | I know it could | 07:34 |
inc0 | I'm +1 on script as well | 07:34 |
SamYaple | i was thinking possibly a task in a playbook that has to be specifically run | 07:35 |
SamYaple | sitting beside site.yml like upgrade.yml or openrc.yml | 07:35 |
CBR09 | I did | 07:35 |
inc0 | I think that would be overkill | 07:35 |
inc0 | I mean...it's just glorified print | 07:35 |
SamYaple | well for one writing a python script will be overkill and many more lines of code | 07:36 |
CBR09 | I did it for me | 07:36 |
CBR09 | http://paste.openstack.org/show/475417/ | 07:36 |
SamYaple | for two, no i mean it would drop the file on the host that you specify | 07:36 |
CBR09 | generate openrc on keystone host and fetch it | 07:36 |
CBR09 | deployment host only is host run ansible | 07:36 |
inc0 | hey...how about just printing it to stdout? | 07:37 |
SamYaple | CBR09: yea it could me much simpler than that by having a template for the openrc file | 07:37 |
inc0 | that would solve problem of placement | 07:37 |
CBR09 | After deploy openstack, I can source /etc/kolla/openrc-admin, it's simple | 07:37 |
SamYaple | not the problem of automating it and dropping it on all hosts | 07:37 |
SamYaple | CBR09: I think a simple task in an openrc.yml will work fine. then you could just specify the host on the command line as the deployer | 07:39 |
SamYaple | to ensure it only goes where you want | 07:39 |
inc0 | again, how about printing it after finished deployment? | 07:39 |
inc0 | if someone want's to dump it into file - he can do it | 07:40 |
SamYaple | inc0: thats not automated, i answered you above | 07:40 |
inc0 | ok, well | 07:41 |
inc0 | I don't know, having to call another ansible playbook and specify inventory just for openrc seems overkill to me | 07:41 |
SamYaple | `kolla openrc` | 07:42 |
SamYaple | doesnt seem like much to me | 07:42 |
inc0 | ok, but what will that do? | 07:42 |
inc0 | find out group in ansible inventroy and run playbook with single task in it? | 07:42 |
inc0 | you still need to specify inventory | 07:42 |
SamYaple | by default deploy on all hosts, `kolla openrc controller|compute` could be for more specific | 07:43 |
SamYaple | and no you dont | 07:43 |
SamYaple | currently we have `kolla deplyo` | 07:43 |
inc0 | I guess my point is - while I'm ok with generating opnrc, it's cool, I don't see use case of having that much of code for that | 07:45 |
SamYaple | a single task? | 07:45 |
SamYaple | you suggested making a pythong script way more code there | 07:45 |
SamYaple | lol pythong | 07:45 |
inc0 | most people would like to kolla deploy -> source ./openrc -> nova list | 07:45 |
inc0 | and unless they're on aio or deploy from one of controllers, kolla openrc would require them to ssh to controller and do whatever | 07:46 |
SamYaple | youll need to justify "most people" | 07:46 |
SamYaple | from an operator point of view, i go to the control nodes | 07:46 |
SamYaple | where the logs are | 07:46 |
SamYaple | thats where the opernc should exist | 07:46 |
SamYaple | so its going to very person to person | 07:47 |
SamYaple | a simple task that you can specify the host is the best option | 07:47 |
inc0 | so why put that much automation to it? | 07:47 |
SamYaple | ..... kolla is automation | 07:47 |
SamYaple | we are talking like 3 lines here | 07:47 |
SamYaple | anyway a single task is fine | 07:47 |
inc0 | ok, well, just write it then? | 07:47 |
SamYaple | we need some way to do this and that is the simplest way | 07:48 |
SamYaple | it shouldnt be writen to all hosts | 07:48 |
SamYaple | we shouldnt be making the choice about where it writes out too | 07:48 |
SamYaple | thats the whole conversation here | 07:48 |
SamYaple | which is solved with a small openrc.yml wher ethe deploy specifies the destination hosts | 07:48 |
inc0 | hmm | 07:48 |
SamYaple | whats teh pushback again? | 07:49 |
CBR09 | so, write simple task in a openrc.yml beside site.yml? | 07:49 |
SamYaple | CBR09: yes | 07:49 |
inc0 | for start, yeah | 07:49 |
SamYaple | dont specify hosts in the task | 07:49 |
SamYaple | that will be dont from the commandline by the deployer | 07:49 |
CBR09 | and ask where put it? | 07:49 |
inc0 | operator will specify it himself during playbook run | 07:50 |
CBR09 | ok, I understand | 07:50 |
inc0 | I liked fact that we have just one playbook to rule them all (site.yml), but we can sort that out later | 07:51 |
CBR09 | when run openrc.yml, it'll ask operator about where it should be placed | 07:51 |
SamYaple | inc0: that was never the end plan | 07:51 |
SamYaple | inc0: that wont do upgrades or config things correctly | 07:51 |
inc0 | yeah I guess, we need to think how to preserve cleanliness and have these options | 07:52 |
SamYaple | i agree | 07:52 |
inc0 | good topic over kirin (japanese beer) | 07:52 |
SamYaple | the plan was always to have some liberty-mitaka.yml for upgrade, but if a better way to do it presents itself... | 07:52 |
CBR09 | so discussed above, can i work on it? | 07:55 |
SamYaple | CBR09: of course :) | 07:56 |
inc0 | yeah, just add another file | 07:56 |
inc0 | we'll sort this out later, don't let it block you | 07:56 |
*** achanda has quit IRC | 07:57 | |
CBR09 | ok, hope it get approved : ) | 07:59 |
*** achanda has joined #kolla | 08:05 | |
*** achanda has quit IRC | 08:16 | |
kjelly | Where is the best place to put the document how to set up docker registry? new file or image-building.rst? | 08:19 |
inc0 | I think additional section in image-building is ok | 08:27 |
kjelly | kjelly: ok | 08:28 |
*** achanda has joined #kolla | 08:39 | |
*** jmccarthy has quit IRC | 08:40 | |
*** jmccarthy has joined #kolla | 08:41 | |
openstackgerrit | Kuo-tung Kao proposed openstack/kolla: update doc about how to set up local registry https://review.openstack.org/231358 | 08:41 |
*** pbourke has joined #kolla | 08:47 | |
*** gfidente has joined #kolla | 08:49 | |
*** achanda has quit IRC | 08:50 | |
*** mbound has joined #kolla | 09:07 | |
openstackgerrit | James McCarthy proposed openstack/kolla: Updated haproxy config to include swift api https://review.openstack.org/231378 | 09:11 |
jmccarthy | I think there is a small bit missing there, stuck in what is hopefully a quick patch for swift | 09:12 |
SamYaple | jmccarthy: need a bug in there | 09:13 |
jmccarthy | Oh, ok I'll check there may not be one yet | 09:14 |
openstackgerrit | James McCarthy proposed openstack/kolla: Updated haproxy config to include swift api https://review.openstack.org/231378 | 09:22 |
jmccarthy | Stuck a quick bug in, is that enough info ? | 09:23 |
SamYaple | i left a comment | 09:25 |
SamYaple | were trying a new thing | 09:25 |
jmccarthy | Where does the tag go ? | 09:28 |
jmccarthy | Ah right ok yea in the commit, one sec | 09:29 |
openstackgerrit | James McCarthy proposed openstack/kolla: Updated haproxy config to include swift api https://review.openstack.org/231378 | 09:30 |
*** exploreshaifali has joined #kolla | 09:52 | |
*** CBR09 has quit IRC | 09:52 | |
SamYaple | jmccarthy: do you want to do the backport of that to the stable/liberty branch? | 10:02 |
openstackgerrit | Merged openstack/kolla: Updated haproxy config to include swift api https://review.openstack.org/231378 | 10:03 |
openstackgerrit | Merged openstack/kolla: add "--template-only" flag for tools/build.py https://review.openstack.org/230882 | 10:05 |
jmccarthy | I can give that a try, where to begin ? =) | 10:07 |
SamYaple | jmccarthy: just checkout the stable/liberty branch | 10:08 |
SamYaple | then do a `git cherry-pick -x <git commit id of master patch>` | 10:08 |
jmccarthy | SamYaple: Checking - this id ? Change-Id: I5c233298312dabeb43efa4821855c77fd81aae67 | 10:28 |
SamYaple | jmccarthy: no https://github.com/openstack/kolla/commit/c879b1f2711db6aefe7dbf88b0b08d3a32e1e02a | 10:28 |
SamYaple | `git cherry-pick -x c879b1f2711db6aefe7dbf88b0b08d3a32e1e02a` | 10:29 |
jmccarthy | Ok I think that went ahead ok =) | 10:35 |
*** kjelly_ has joined #kolla | 10:49 | |
*** sdake has joined #kolla | 10:52 | |
*** daneyon_ has quit IRC | 11:03 | |
*** athomas has quit IRC | 11:06 | |
*** athomas has joined #kolla | 11:09 | |
*** tfukushima has quit IRC | 11:15 | |
sdake | morning | 11:44 |
SamYaple | morning sdake | 11:46 |
sdake | hey sam | 11:46 |
jmccarthy | o/ | 11:46 |
sdake | hey jmccarthy | 11:46 |
sdake | jmccarthy is it accurate you wont be at summit? | 11:46 |
inc0 | hey sdake | 11:46 |
sdake | yo inc0 | 11:46 |
sdake | inc0 did you get an invitation to th core reviewer party at summit? | 11:47 |
jmccarthy | Sadly yea I won't be =( | 11:47 |
sdake | jmccarthy bummer I was going to get you an invitation to the cr party | 11:48 |
sdake | just ordered a 1tb macbook 15" pro | 11:49 |
* sdake excited | 11:49 | |
sdake | now i can play diablo 3 on my laptop :) | 11:49 |
sdake | 2.8ghz processor and high performance gpu | 11:49 |
jmccarthy | Dangit, sounds good sdake but not this time around | 11:49 |
inc0 | sdake, I did | 11:49 |
jmccarthy | pbourke says he will party twice as hard for me ;) | 11:49 |
inc0 | I'm wondering if I should go tho, I dislike idea of separate parties for cores and non-cores | 11:50 |
inc0 | community is community | 11:50 |
sdake | inc0 its a sweet party | 11:50 |
sdake | the problem with havinga party for eveyrone is expense and venue space | 11:50 |
SamYaple | inc0: you stand on your morals! im going | 11:50 |
sdake | there are several noncores going as well | 11:50 |
SamYaple | just the special ones | 11:51 |
sdake | ptls can ask for invitations for non core folks | 11:51 |
sdake | what is sweetabout the venue is it is where the 47 ronin are buried | 11:51 |
jmccarthy | SamYaple: inc0: you stand on your morals! im going - LOL | 11:52 |
pbourke | I'll protest at the bar | 11:52 |
pbourke | :p | 11:52 |
SamYaple | ill compromise and join you pbourke | 11:52 |
inc0 | "let's drink for non-cores!" | 11:52 |
*** dwalsh has joined #kolla | 11:52 | |
SamYaple | here here | 11:52 |
inc0 | there should be separate party for ATC tho | 11:54 |
SamYaple | ive been ATC for the past few summits, this is the first one im going too | 11:55 |
sdake | there has been an atc party in the past | 11:56 |
sdake | but i think its too expensive now | 11:57 |
sdake | there are like 1500 atcs iirc | 11:57 |
SamYaple | ok just the ATC that are also core on a project | 11:57 |
SamYaple | wait... | 11:57 |
sdake | atc doen'thave to be core | 11:59 |
sdake | an atc is defined as somoene who has a commit in the last 2 cycles I think to any bi tent project | 11:59 |
sdake | they may have changed it to just 1 cycle | 11:59 |
sdake | i think there are about 300-400 core reviewers | 12:00 |
inc0 | I don't think there is a core who is not an atc | 12:00 |
inc0 | there shouldn't be in any case | 12:00 |
sdake | yes to be core you have to be an atc | 12:00 |
pbourke | the hp/scality party is big enough for everyone | 12:01 |
pbourke | make sure you guys rsvp for that if around | 12:01 |
sdake | i guess i'll do that now | 12:02 |
inc0 | Paris one was epic, but Vancouver...not so | 12:02 |
sdake | i didn't rsvp in time for vc one | 12:02 |
inc0 | I just went | 12:03 |
inc0 | no invintation or anyting | 12:03 |
inc0 | but a bit later | 12:03 |
sdake | i had heard they weren't letting people in without rsvp | 12:04 |
sdake | pbourke where od you rsvp for the parties? | 12:04 |
sdake | I dont see iton the summit track schedule | 12:04 |
SamYaple | its a secret schedule | 12:04 |
SamYaple | for the cool people | 12:04 |
sdake | well share teh secret and make me cool :) | 12:04 |
sdake | son up at 5am | 12:04 |
* sdake groans | 12:05 | |
SamYaple | THATS NOT HOW COOLNESS WORKS | 12:05 |
inc0 | how does it work? | 12:05 |
SamYaple | idk ask pbourke i didnt get an invite either | 12:05 |
sdake | hey folks, just a reminder please vote for the tc election, even if you decide not to vote for me :) | 12:08 |
inc0 | is voting open already? | 12:08 |
sdake | i found it , search for RSVP | 12:09 |
sdake | inc0 yes you should have got an email last week i think | 12:09 |
inc0 | aaa yeah, I missed it | 12:09 |
*** sdake has quit IRC | 12:09 | |
*** sdake has joined #kolla | 12:10 | |
sdake | samyaple any chance you can fix heat so the domain stuff works | 12:13 |
sdake | SamYaple that is | 12:13 |
SamYaple | sdake: all the domain stuff works, whatcha mean? | 12:13 |
sdake | oh didthat get fixed? | 12:13 |
sdake | i wasn't sure if wait conditions were working or not | 12:13 |
sdake | and if they aren't we need them to be :) | 12:14 |
SamYaple | idk anything about heat | 12:14 |
SamYaple | i can spin up things and delete them | 12:14 |
SamYaple | and thats asking alot | 12:14 |
sdake | haha :) | 12:14 |
sdake | if I make you a template can you get it working? | 12:14 |
sdake | (one with waitconditions) | 12:14 |
SamYaple | idk | 12:15 |
SamYaple | i used your steak template thing | 12:15 |
sdake | well it is something we can develop as a backport | 12:15 |
sdake | since its a bug fix and not a feature | 12:15 |
SamYaple | honestly i think we can backport just about everything | 12:16 |
sdake | ya it is probalby something we should dicuss at summit | 12:17 |
*** exploreshaifali has quit IRC | 12:17 | |
sdake | but I'm open to a more liberal backport policy | 12:17 |
*** dwalsh has quit IRC | 12:17 | |
*** athomas has quit IRC | 12:17 | |
SamYaple | the way i see it, no structure changes (folders moving and what not) | 12:17 |
SamYaple | new containers _might_ be ok to backport, if they dont affect existing | 12:17 |
sdake | i htink for the most part our structure is set | 12:17 |
SamYaple | and existing containers get bugfixes only | 12:17 |
sdake | i can't see any need to reorg any containers | 12:18 |
sdake | we need to add upgrade migrations to the containers | 12:18 |
sdake | but that should be distinct and not break backwards compatibility i think | 12:18 |
SamYaple | sdake: https://blueprints.launchpad.net/kolla/+spec/openstack-common-container | 12:18 |
SamYaple | you have a TODO on that | 12:18 |
sdake | roger | 12:18 |
sdake | i'm going to actualloy work on that today | 12:18 |
sdake | and finish it up | 12:19 |
*** athomas has joined #kolla | 12:22 | |
sdake | morning athomas | 12:23 |
* sdake so excited for new laptop | 12:24 | |
* sdake loves vestmas :) | 12:25 | |
sdake | christmas comes two months early in the steak household :) | 12:25 |
*** rhallisey has quit IRC | 12:27 | |
*** dwalsh has joined #kolla | 12:30 | |
*** shardy has joined #kolla | 12:35 | |
*** rhallisey has joined #kolla | 12:36 | |
*** loth has quit IRC | 12:53 | |
*** jruano has joined #kolla | 12:53 | |
*** loth has joined #kolla | 12:55 | |
*** klint has quit IRC | 12:58 | |
*** loth has quit IRC | 12:59 | |
*** loth has joined #kolla | 13:00 | |
sdake | samyaple can ou look at master and submit cherypick backports for each change that shold go into liberty? | 13:02 |
sdake | SamYaple that is^ | 13:02 |
openstackgerrit | Kuo-tung Kao proposed openstack/kolla: update doc about how to set up local registry https://review.openstack.org/231358 | 13:03 |
SamYaple | sdake: i have been | 13:03 |
SamYaple | sdake: tehre is a few more but ive been up with most or all of them | 13:04 |
sdake | sounds good | 13:04 |
*** sdake has quit IRC | 13:09 | |
*** thumpba has joined #kolla | 13:10 | |
*** sdake has joined #kolla | 13:10 | |
*** sdake_ has joined #kolla | 13:12 | |
sdake_ | SamYaple rabbitmq 3.5.5 is out | 13:14 |
*** sdake has quit IRC | 13:14 | |
*** kjelly_ has quit IRC | 13:17 | |
SamYaple | sdake_: yea but ubuntu is pinned to 3.5.4 | 13:21 |
SamYaple | i have no contorl of that | 13:21 |
SamYaple | you dont have to, just saying | 13:21 |
SamYaple | anything >=3.5.4 is good | 13:21 |
sdake_ | any objection updating to 3.5.5 for ubuntu? | 13:22 |
sdake_ | i tried a 3.5.4 build and plugin enable locke up | 13:22 |
* sdake_ groans | 13:22 | |
sdake_ | rather 3.5.5 | 13:22 |
sdake_ | trying with --ffline now | 13:22 |
sdake_ | SamYaple any objection to doing rabbitmq plugin enablement in start.sh? | 13:23 |
sdake_ | INFO:__main__:rabbitmq:Step 3 : RUN /usr/lib/rabbitmq/bin/rabbitmq-plugins enable rabbitmq_management --offline | 13:23 |
sdake_ | this line just locks up | 13:23 |
pbourke | sdake_: https://www.eventbrite.com/e/ichi-go-ichi-e-openstack-summit-tokyo-2015-official-evening-event-tickets-18837896656 | 13:27 |
sdake_ | pbourke i foudn it earlier thanks | 13:27 |
pbourke | cool | 13:27 |
sdake_ | pbourke are you bringing a significant other | 13:27 |
sdake_ | you will get a chance to meet my wife :) | 13:27 |
pbourke | sdake_: no she would have loved to go though! | 13:28 |
sdake_ | ya my wife is going on work | 13:28 |
sdake_ | she works at ansible | 13:28 |
sdake_ | 50% of her job is community management for openstack iirc | 13:28 |
pbourke | nice | 13:29 |
sdake_ | or maybe 25% | 13:29 |
sdake_ | not really sure what she does :) | 13:29 |
vincent_vdk | sdake_: would that be Robyn | 13:31 |
sdake_ | roger | 13:31 |
vincent_vdk | :) | 13:32 |
sdake_ | do kolla builds enable the tty? | 13:35 |
sdake_ | and if not, how can i do so? | 13:35 |
SamYaple | sdake_: rabbitmq would still need to do the endablement in --offline mode in start.sh | 13:43 |
SamYaple | sdake_: i dont think they do | 13:43 |
SamYaple | but you can do fake TTY | 13:43 |
sdake_ | SamYaple so the issue I am running into is rabbitmq with the plugins locks up during startup | 13:43 |
SamYaple | and i *think* you can enable one in the dockerfile | 13:43 |
sdake_ | not rabbitmq, but the plugin thing | 13:43 |
SamYaple | yea i hear ya | 13:44 |
sdake_ | i am not sure if its the tty or somethign else | 13:44 |
SamYaple | do it by hand then | 13:44 |
sdake_ | did you run into thi porblem in ubnut? | 13:44 |
SamYaple | nope | 13:44 |
sdake_ | do which by hand? | 13:44 |
sdake_ | you mean install and enable pllugin? | 13:44 |
*** banix has joined #kolla | 13:44 | |
SamYaple | run the command on the image right before it installs the plugin | 13:44 |
SamYaple | and see the issue | 13:44 |
sdake_ | ya i'll try taht with a tty | 13:44 |
SamYaple | ill need more info than "locks up" | 13:44 |
SamYaple | btw i would say the openstack-common stuff is a bigger priority sdake | 13:45 |
SamYaple | the rabbit is a simple backport | 13:45 |
sdake_ | http://ur1.ca/nxgiw -> http://paste.fedoraproject.org/275319/39239144 | 13:47 |
sdake_ | this is as much of the typescript as I can paste | 13:47 |
sdake_ | but basicallly it enters a loop where it trie to close 1->65535 fds in a loop | 13:47 |
SamYaple | esh idk. seems wierd | 13:48 |
sdake_ | running with a tty works | 13:48 |
sdake_ | so we may need to modify build.py or something to get a tty | 13:48 |
sdake_ | any ideas on that? | 13:48 |
SamYaple | try running with `TERM=vt102 /usr/lib/rabbitmq/bin/rabbitmq-plugins enable --offline rabbitmq_management` | 13:49 |
sdake_ | if I do the rabbitmq enablement with a tty, it works - so thatis the olution | 13:49 |
sdake_ | i already ran that | 13:49 |
sdake_ | you mean in the build? | 13:49 |
SamYaple | look at the first part | 13:49 |
SamYaple | the TERM variable | 13:49 |
sdake_ | ok try8ing | 13:49 |
sdake_ | same looping result | 13:50 |
SamYaple | yea so fake term isnt going to work there | 13:51 |
SamYaple | we can probably do this let me check the api | 13:51 |
SamYaple | nope cant build with a tty | 13:52 |
sdake_ | https://github.com/docker/docker/pull/4882 | 13:52 |
SamYaple | api is what matters | 13:53 |
*** diogogmt has joined #kolla | 13:53 | |
sdake_ | right | 13:53 |
SamYaple | http://docker-py.readthedocs.org/en/latest/api/#build | 13:53 |
sdake_ | so, start.sh then? | 13:53 |
SamYaple | no thats not a start.sh thing | 13:53 |
SamYaple | this needs ot be figured out | 13:53 |
sdake_ | no i mean run the enable in start.sh | 13:53 |
SamYaple | i know | 13:53 |
SamYaple | and no thats not where it goes | 13:54 |
SamYaple | youll still have the same issue | 13:54 |
sdake_ | the containers have ttys | 13:54 |
sdake_ | do they not? | 13:54 |
SamYaple | nope | 13:54 |
sdake_ | oh | 13:54 |
SamYaple | we _can_ launch them with it but that has security problems i believe | 13:54 |
SamYaple | anyway that changes our entire bootstrap beahviour | 13:55 |
SamYaple | there are plenty of ways to fake a tyy | 13:55 |
sdake_ | open to other ideas :) | 13:55 |
SamYaple | you hsould look into those | 13:55 |
SamYaple | i have to head home | 13:55 |
sdake_ | ok i'll give ita go | 13:55 |
SamYaple | i dont view that as liberty blocker, just pin it to a working version then, 3.5.4 would be prefereed but eh i guess | 13:55 |
SamYaple | oh wait this is for master | 13:55 |
SamYaple | shoot | 13:55 |
SamYaple | idk | 13:55 |
SamYaple | ill think about it | 13:56 |
sdake_ | hmm I can't find any simulations for /dev/tty | 14:06 |
*** diogogmt has quit IRC | 14:13 | |
sdake_ | SamYaple I found a solution, not sure why it works | 14:15 |
sdake_ | butseems to ge tthe job done | 14:15 |
sdake_ | http://ur1.ca/nxgw5 -> http://paste.fedoraproject.org/275335/41409891 | 14:16 |
sdake_ | http://ur1.ca/nxgxj -> http://paste.fedoraproject.org/275337/14441411 | 14:19 |
sdake_ | better version ^ | 14:19 |
*** achanda has joined #kolla | 14:29 | |
*** achanda has quit IRC | 14:37 | |
*** diga has quit IRC | 14:47 | |
*** achanda has joined #kolla | 14:57 | |
*** diogogmt has joined #kolla | 14:58 | |
*** achanda has quit IRC | 15:01 | |
*** diogogmt has quit IRC | 15:07 | |
*** diogogmt has joined #kolla | 15:09 | |
*** vinkman has joined #kolla | 15:10 | |
*** daneyon has joined #kolla | 15:16 | |
*** daneyon has quit IRC | 15:17 | |
*** shardy has quit IRC | 15:18 | |
openstackgerrit | Paul Bourke proposed openstack/kolla: Fix incorrect dd param in cinder-guide.rst https://review.openstack.org/231544 | 15:21 |
sdake_ | pbourke are youw orking to get cinder in an operational state? | 15:21 |
pbourke | sdake_: not exactly, just going through the basics to test a fix for https://bugs.launchpad.net/kolla/+bug/1487780 | 15:22 |
openstack | Launchpad bug 1487780 in kolla "/dev needs to be bindmounted into cinder" [Critical,Confirmed] - Assigned to Paul Bourke (pauldbourke) | 15:22 |
sdake_ | i dont necessairly want to jut merge thatwithout a holistic solution to the problem | 15:22 |
sdake_ | (the /dev bindmount) | 15:22 |
*** sdake_ is now known as sdake | 15:23 | |
sdake | all those cinder bugs are various states of my understanding of the problem | 15:23 |
sdake | i'm not sure if they are all consistent | 15:23 |
sdake | for exaple, /dev may notactually need to be bindmounted | 15:23 |
sdake | it is unclear without a holistic solution if it is or not | 15:23 |
pbourke | unfortunately you may have to remind me of "the problem" | 15:24 |
*** blahRus has joined #kolla | 15:24 | |
sdake | cinder won't attach volumes | 15:25 |
sdake | atm the issue is that tgtd is not running (at all) | 15:25 |
sdake | apparently it was reoved in someone's patch to cinder at some point in the past | 15:25 |
sdake | that is the issue that needs solving first | 15:25 |
sdake | the idea is to get cinder + lvm + iscsi working together | 15:25 |
*** shardy has joined #kolla | 15:25 | |
sdake | that is why I thought we may need /dev to be bindmounted | 15:26 |
sdake | but i am not certain if thatis accurate or not | 15:26 |
*** shardy has quit IRC | 15:26 | |
pbourke | hmm ok | 15:26 |
sdake | probably whathsould happen with all those icnder bugs is they should allbe closed | 15:26 |
sdake | and a new one opened to solve the remaining problem oflvm + iscsi + cinder | 15:26 |
pbourke | yeah sounds like you should file a higher level bug with all current info | 15:26 |
sdake | I think ceph + cinder works | 15:26 |
sdake | i dont have any modern info | 15:26 |
sdake | but agree that owuld be a better approach | 15:26 |
*** dims_ has quit IRC | 15:27 | |
pbourke | I just grabbed above because I saw it on the board as critical and you had requested everyone grab a bug for the next release ;) | 15:27 |
*** daneyon has joined #kolla | 15:27 | |
sdake | master doesn't seem to work | 15:27 |
pbourke | did seem a little strange though that if it was really that easy, it would be done already | 15:27 |
sdake | i get stuck at TASK: [glance | Creating the Glance service and endpoint] ********************* | 15:28 |
sdake | in ansible | 15:28 |
* sdake groans | 15:29 | |
sdake | does master work for you pbourke? | 15:29 |
sdake | I just pulled i tt this morning | 15:29 |
sdake | SamYaple around? | 15:29 |
inc0 | sdake, docker logs shows anything? | 15:30 |
inc0 | and confirm that keystone is up and running | 15:30 |
sdake | i dunno i'm trying again with --vv on the playbook launch | 15:31 |
sdake | ya keystone appears to be running | 15:31 |
sdake | keystone appears wedged | 15:32 |
*** dims_ has joined #kolla | 15:32 | |
inc0 | wedged? | 15:33 |
sdake | ya as in locked up | 15:33 |
inc0 | new word everyday;) look at ks logs maybe | 15:33 |
inc0 | ? | 15:33 |
sdake | ya i'll debug, i was just curious IF ANYONE ELSE HAS WORKING MASSTER :) | 15:34 |
inc0 | docker exec -it log_data /var/log/keystone/keystone-error.log | 15:34 |
sdake | or if its me only | 15:34 |
inc0 | I've deployed master today | 15:34 |
sdake | there is no /var/log/keystone in log_data | 15:35 |
pbourke | sdake: at least glance+keystone seem to come up | 15:35 |
pbourke | but I do seem to get random slowness with ansible | 15:35 |
sdake | my database appears locked up | 15:36 |
pbourke | I have a /var/log/keystone | 15:36 |
sdake | <minime-01> REMOTE_MODULE command docker exec -t kolla_ansible /usr/bin/ansible localhost -m kolla_keystone_service -a "service_name=glance service_type=image description='Openstack Image' endpoint_region=RegionOne admin_url='http://192.168.1.148:9292' internal_url='http://192.168.1.148:9292' public_url='http://broked.selfip.net:9292' region_name=RegionOne auth={{ openstack_glance_auth }}" -e "{'openstack_glance_auth':{'username': 'admin', 'project_name': | 15:36 |
sdake | 'admin', 'password': 'password', 'auth_url': 'http://192.168.1.148:35357/v2.0'}}" | 15:36 |
sdake | in which container pbourke? | 15:36 |
pbourke | rsyslog | 15:36 |
sdake | rsyslogd: cannot create '/dev/log': Address already in use | 15:37 |
sdake | is that normal? | 15:37 |
sdake | ya no keystone log at all | 15:37 |
sdake | no keystone dir i mean | 15:37 |
sdake | in log_data or rsyslog containers | 15:37 |
sdake | perhaps I need a new /etc/kolla directory | 15:38 |
*** daneyon_ has joined #kolla | 15:43 | |
sdake | jpeeler can you give m ean update on the dictionary thing you were working on | 15:45 |
jpeeler | i stopped it | 15:45 |
*** daneyon has quit IRC | 15:46 | |
sdake | as in abandoned that work? | 15:46 |
jpeeler | yeah - https://review.openstack.org/#/c/229535/ | 15:46 |
sdake | why abaonded? | 15:47 |
sdake | not feasible? | 15:47 |
jpeeler | read the comment on patchset 4 from me | 15:47 |
sdake | because of the merge? | 15:47 |
sdake | right i read it | 15:47 |
sdake | because of the replacement of key/value pairs? | 15:48 |
jpeeler | yeah, it was either have an inconsistent style | 15:48 |
jpeeler | or not do it at all | 15:49 |
jpeeler | and after spending hours doing a bunch of modifications, i gave up | 15:49 |
sdake | well inconsistent style isn't ideal | 15:49 |
sdake | thanks for the investigaiton, it looked promising | 15:49 |
jpeeler | would have never started it if i had known dicts didn't merge | 15:49 |
sdake | too bad it didn't work out in the end,bu tthat is common in software engineering :) | 15:49 |
jpeeler | i changed so much text i saw git messages i had never seen before, with rewrite percentages per file | 15:50 |
sdake | haproxy si busted | 15:52 |
sdake | is busted | 15:52 |
*** inc0 has quit IRC | 15:52 | |
sdake | or keepalived | 15:52 |
sdake | damn inc0 left | 15:52 |
sdake | pbourke is haproxy running for you ? | 15:57 |
*** unicell has joined #kolla | 15:58 | |
*** unicell1 has quit IRC | 15:59 | |
sdake | it seems like keepalived and haproxy are running with "run_once: True | 16:02 |
sdake | but that isn't specified in the ansible playbook | 16:02 |
sdake | I wonder i fthe playbook is somehow globally fubars | 16:02 |
*** cloudnautique has joined #kolla | 16:02 | |
sdake | cloudnautique did you have some questions from yesterday that didn't get answered? | 16:03 |
cloudnautique | I'm making progress :) | 16:04 |
cloudnautique | thanks for asking. As I keep digging into it though I'm sure I'll be jumping in here. | 16:05 |
sdake | cool | 16:05 |
pbourke | sdake: I haven't tried haproxy master recently | 16:15 |
sdake | i figured out why haproxxy wans't running, my network section in the invetory file didn't have the 2 non-network nodes | 16:15 |
sdake | which Ig uess makes sense | 16:15 |
sdake | I'm not sure if haproxy has anything ot do with why glance is not workign | 16:15 |
sdake | it seems like keystone is blocked up in some way | 16:16 |
*** mbound has quit IRC | 16:16 | |
sdake | it is responding but doesn't seem to be giving out keys or something | 16:16 |
sdake | as in i can telenet to the port 5000 and it works | 16:16 |
* sdake wtb functional testing!! | 16:17 | |
*** daneyon_ has left #kolla | 16:19 | |
sdake | so starting with a limited set of network nodes is what cusaes the problem | 16:21 |
sdake | it cuases everything to go shits up | 16:21 |
*** banix has quit IRC | 16:30 | |
openstackgerrit | Steven Dake proposed openstack/kolla: Pin rabbitmq to 3.5.5 https://review.openstack.org/231583 | 16:35 |
*** unicell has quit IRC | 16:35 | |
openstackgerrit | Steven Dake proposed openstack/kolla: Pin rabbitmq to 3.5.5 https://review.openstack.org/231583 | 16:36 |
*** dims_ has quit IRC | 16:37 | |
*** Hui has joined #kolla | 16:40 | |
*** dims_ has joined #kolla | 16:41 | |
*** dims_ has quit IRC | 16:41 | |
*** dims_ has joined #kolla | 16:42 | |
*** jtriley has joined #kolla | 16:43 | |
*** dims__ has joined #kolla | 16:44 | |
*** dims_ has quit IRC | 16:45 | |
*** inc0 has joined #kolla | 16:52 | |
*** sdake has quit IRC | 17:03 | |
*** sdake has joined #kolla | 17:05 | |
*** jtriley_ has joined #kolla | 17:07 | |
*** jtriley has quit IRC | 17:09 | |
*** thumpba has quit IRC | 17:10 | |
*** unicell has joined #kolla | 17:12 | |
*** inc0 has quit IRC | 17:12 | |
*** banix has joined #kolla | 17:12 | |
*** jainman has joined #kolla | 17:16 | |
jainman | Bug 1502633] Re: Not able to Launch Instances | 17:17 |
openstack | bug 1502633 in kolla "Not able to Launch Instances" [Critical,Triaged] https://launchpad.net/bugs/1502633 | 17:17 |
sdake | hey | 17:17 |
sdake | need some debughelp with that? | 17:17 |
jainman | Yes | 17:18 |
sdake | what distro ar eyou running | 17:18 |
jainman | liberty-rc3 | 17:18 |
sdake | no os distro like fedora ubuntu cento | 17:18 |
jainman | centos | 17:18 |
sdake | how many nodes | 17:19 |
jainman | 7.x | 17:19 |
jainman | allinonenode | 17:19 |
sdake | did you setup a private registry on your centos node? | 17:19 |
jainman | yes | 17:19 |
jainman | sudo systemctl restart docker.service sudo docker run -d -p 5010:5010 --restart=always --name registry registry:2 sudo service libvirtd disable sUDO service libvirtd stop | 17:20 |
sdake | did you disable selinux and iptables? | 17:20 |
sdake | or firewalld? | 17:20 |
jainman | I don't think so | 17:20 |
jainman | No I didn't | 17:20 |
sdake | ok lets do that | 17:21 |
sdake | setenforce 0 | 17:21 |
sdake | will disable selinux | 17:21 |
sdake | sudo vi /etc/selinux/config | 17:21 |
sdake | change whatever it is to disabled I think | 17:21 |
sdake | SELINUX=disabled | 17:21 |
sdake | is what I hve in my config file | 17:21 |
sdake | sudo service firewalld stop | 17:22 |
sdake | sudo service firewalld disable | 17:22 |
sdake | sudo iptables stop | 17:22 |
sdake | sudo iptables disable | 17:22 |
jainman | Ok, I will check it. My VPN is not working | 17:22 |
sdake | sudo ip6tables stop | 17:22 |
sdake | sudo ip6tables disable | 17:22 |
jainman | Do I have to restart container after that | 17:23 |
sdake | we need to debug live, can't give you one instruciton and send you off, it will take foreer to do it thatway :) | 17:23 |
sdake | first step - get your bare metal system into shape for deployment | 17:23 |
sdake | second step - make sure you built images properly | 17:23 |
sdake | third step - deploy with a proper inventory file | 17:23 |
sdake | the third step involves checking if your config file is correct | 17:24 |
sdake | and if its not correct, fixing the dificencies | 17:24 |
jainman | I used default inventry file | 17:24 |
sdake | after you get your bare metal system into operational state, use cleanup-containers -fv | 17:24 |
*** achanda has joined #kolla | 17:24 | |
sdake | take care to look at tools/cleanup-containers and make sure its the versio nthat doesn't erase the existing containers (such as your registry) | 17:25 |
sdake | there is a new version which doesn't wipe out the registry on cleanup | 17:25 |
sdake | you will want to run that | 17:25 |
sdake | i am not sure which version rc1 has | 17:25 |
sdake | can yo uget your vpn operational so we can take it step by step please? | 17:26 |
jainman | I am trying, please give me some time | 17:26 |
sdake | sure i'm here all day :) | 17:26 |
*** jmccarthy has quit IRC | 17:27 | |
sdake | did you do a minimal os install of centos? | 17:27 |
*** dims_ has joined #kolla | 17:29 | |
*** jainman_ has joined #kolla | 17:30 | |
*** jainman has quit IRC | 17:30 | |
*** dims__ has quit IRC | 17:32 | |
*** cloudnautique has quit IRC | 17:35 | |
openstackgerrit | Steven Dake proposed openstack/kolla: Remove debug output in base image yum command https://review.openstack.org/231627 | 17:38 |
sdake | jainman_ did you do a minima los install of centos? | 17:38 |
*** exploreshaifali has joined #kolla | 17:50 | |
openstackgerrit | Hui Kang proposed openstack/kolla: Add consul containers for dynamic configuration, discovery, and monitoring. https://review.openstack.org/228719 | 17:50 |
*** dims_ has quit IRC | 17:53 | |
*** jainman_ has quit IRC | 17:55 | |
*** dims_ has joined #kolla | 18:01 | |
*** dims_ has quit IRC | 18:05 | |
*** dims_ has joined #kolla | 18:06 | |
openstackgerrit | Steven Dake proposed openstack/kolla: Remove which from neutron-server https://review.openstack.org/231645 | 18:06 |
*** athomas has quit IRC | 18:13 | |
*** dwalsh has quit IRC | 18:30 | |
*** cloudnautique has joined #kolla | 18:55 | |
*** dwalsh has joined #kolla | 19:03 | |
*** thumpba has joined #kolla | 19:06 | |
*** cloudnautique has quit IRC | 19:16 | |
*** cloudnautique has joined #kolla | 19:25 | |
*** cloudnautique has quit IRC | 19:28 | |
*** exploreshaifali has quit IRC | 19:40 | |
sdake | mr SamYaple I am in need of your expertise | 19:46 |
sdake | SamYaple do you have any ideas on https://bugs.launchpad.net/kolla/+bug/1503412 | 19:56 |
openstack | Launchpad bug 1503412 in kolla "centos from source sudo failing" [Critical,Triaged] - Assigned to Steven Dake (sdake) | 19:56 |
*** thumpba has quit IRC | 20:01 | |
*** sdake has quit IRC | 20:27 | |
*** sdake has joined #kolla | 20:28 | |
Hui | TASK: [nova | Disabling netfilter for bridges] ******************************** | 20:34 |
Hui | skipping: [control02] => (item=iptables) | 20:34 |
Hui | skipping: [control02] => (item=ip6tables) | 20:34 |
Hui | skipping: [control01] => (item=iptables) | 20:34 |
Hui | skipping: [control01] => (item=ip6tables) | 20:34 |
Hui | failed: [compute01] => (item=iptables) => {"failed": true, "item": "iptables"} | 20:34 |
Hui | msg: setting net.ipv4.bridge-nf-call-iptables failed: sysctl: cannot stat /proc/sys/net/ipv4/bridge-nf-call-iptables: No such file or directory | 20:34 |
Hui | failed: [compute01] => (item=ip6tables) => {"failed": true, "item": "ip6tables"} | 20:34 |
Hui | msg: setting net.ipv4.bridge-nf-call-ip6tables failed: sysctl: cannot stat /proc/sys/net/ipv4/bridge-nf-call-ip6tables: No such file or directory | 20:34 |
Hui | Hi, my recent deployment of compute node failed due to the above error | 20:35 |
Hui | is there anything change in rdo? Thanks. | 20:35 |
*** rhallisey has quit IRC | 20:43 | |
*** juggler has quit IRC | 20:43 | |
*** cloudnull has quit IRC | 20:43 | |
*** rhallisey has joined #kolla | 20:46 | |
*** juggler has joined #kolla | 20:46 | |
*** cloudnull has joined #kolla | 20:46 | |
*** sdake has quit IRC | 20:53 | |
*** sdake has joined #kolla | 20:58 | |
sdake | hui i also had this porblem | 20:59 |
sdake | you can revert the commit that says somethign about sysctl in the git log | 20:59 |
sdake | and it will resole th prolbe | 20:59 |
sdake | m | 20:59 |
sdake | it is a change that apparent ydoesnt' seem to work in all systems | 20:59 |
sdake | Hui ^^ | 20:59 |
sdake | hui https://bugs.launchpad.net/kolla/+bug/1503416 | 21:08 |
openstack | Launchpad bug 1503416 in kolla "iptables sysctl broken on centos 7" [Critical,Confirmed] - Assigned to Sam Yaple (s8m) | 21:08 |
openstackgerrit | Steven Dake proposed openstack/kolla: Enable source for openstack-base https://review.openstack.org/231716 | 21:12 |
Hui | sdake, thanks for the info | 21:12 |
sdake | i suspect sam will fix it tonight | 21:13 |
Hui | nice | 21:14 |
Hui | sdake, i am putting more advanced check in consul to demonstrate its value | 21:14 |
Hui | do you think a video demo will helpful? | 21:15 |
sdake | emmensly helpful | 21:15 |
sdake | otherwise i have to figure out how to get it to work | 21:15 |
Hui | sure, I will upload to youtube | 21:15 |
sdake | i'd figure it out if we were keen to merge it but the core team thinking so far is not worth the pain | 21:16 |
Hui | understand | 21:16 |
sdake | raining cold and windy | 21:18 |
sdake | bad weather = nice for once :) | 21:18 |
sdake | not 110 f out today | 21:18 |
Hui | the weather is so nice in east | 21:18 |
sdake | where arey ou located | 21:19 |
Hui | yorktown heights, NY | 21:19 |
sdake | scottsdale az here | 21:19 |
Hui | who? | 21:20 |
sdake | that is where I live | 21:20 |
sdake | arizona | 21:20 |
Hui | I see | 21:20 |
Hui | I thought you were in 110, CA | 21:20 |
*** dwalsh has quit IRC | 21:20 | |
sdake | 110F in arizona typically this time of year :( | 21:21 |
sdake | or 105 | 21:21 |
sdake | not particularly pleasant | 21:21 |
sdake | today its about 50f | 21:21 |
Hui | wow | 21:21 |
sdake | hui what do you think of the general design of kolla | 21:22 |
sdake | see areas for improvement? | 21:23 |
Hui | I like the design of kolla | 21:23 |
Hui | but for production depoyment, we are working on the HA side | 21:24 |
Hui | that's why I propose the consul idea | 21:24 |
sdake | we being, who? | 21:24 |
Hui | internally IBM | 21:24 |
sdake | which part of ha is deficient | 21:24 |
Hui | internally we have some debates on this micro-service style HA enabled by consul and docker versus pacemaker on VM | 21:25 |
sdake | we just use haproxy + keepalived | 21:25 |
sdake | no pacemake | 21:25 |
Hui | now the haproxy only checks if the port is open or not | 21:25 |
Hui | that is not enough | 21:25 |
sdake | ya need deper heatlhchecking i'll grant you that:) | 21:26 |
Hui | yes | 21:26 |
*** jtriley_ has quit IRC | 21:27 | |
Hui | sdake, do you think the ha in kolla is enough? | 21:29 |
sdake | i am not concerned with split brain | 21:29 |
sdake | I have alwyas thought hardware fencing is overkill for ha | 21:29 |
sdake | proper ha systems are designed to tolerate partitions | 21:29 |
sdake | i think kolla's ha system could use deper healthchecking | 21:30 |
sdake | but i'd like to keep things as simple as possible | 21:30 |
sdake | what i'd like to se is a 1/2/3 level where level 3 takes 5 seconds to run, 2 takes 1 econd to run, 1 takes .1 seconds to run | 21:30 |
Hui | with cinder and glance be integrated into the picture, split brain may become an issue | 21:30 |
sdake | with intervals of 1, 10, 60 seconds | 21:30 |
sdake | not when usingceph as a bckend | 21:31 |
sdake | which is what we promote | 21:31 |
Hui | ok | 21:31 |
harmw | we could use Consul for doing more thourough healthchecking | 21:33 |
harmw | I'm not aware of the current Consul status in Kolla though :) | 21:33 |
sdake | hui has ap atch up for that | 21:35 |
sdake | but I tend to think tis a bit overkill | 21:36 |
sdake | i'd prefer not introduce new big thigns into kolla | 21:36 |
sdake | couldn't compelx heatlhcehcking just be done through a helathcheck script implemented in haproxy? | 21:37 |
sdake | this seems way more simple to me | 21:37 |
harmw | yep | 21:37 |
sdake | hui any other areas for improvement in kolla you suggest besides ha? | 21:38 |
sdake | hui will you be at openstack summit | 21:38 |
harmw | I've seen the consul patch or something, I haven't looked at it though | 21:38 |
Hui | sdake, I have no plan for the summit yet; waiting for my manager's approval | 21:41 |
sdake | cuttingclose on hotels and air travel | 21:41 |
sdake | if you happen to make it, would loe to have you in the kolla design sessions :) | 21:41 |
sdake | hui any other weak spotsyou see in kolla besides ha? | 21:42 |
Hui | no | 21:42 |
sdake | any other room for improvement? | 21:42 |
Hui | I am fine with the current design of kolla | 21:42 |
sdake | just looking for ideas - its ok to be critical :) | 21:42 |
Hui | yes, I will speak out in this channel when I can think of anything | 21:43 |
sdake | cool thanks | 21:43 |
sdake | we have done quite a bit of gold plating | 21:43 |
Hui | is there a special design session for kolla in the summit? | 21:43 |
sdake | we have 8 sessions iirc | 21:43 |
sdake | 3 fishbowl 5 regular | 21:44 |
*** banix has quit IRC | 21:44 | |
sdake | and a halfday contributor meetup | 21:44 |
Hui | sounds interesting | 21:44 |
Hui | hi, harmw, please add your comment in the consul patch when you have a minute to review them. thanks. | 21:46 |
Hui | sdake, are the 8 session all about kolla? | 21:46 |
sdake | yes, plsu we have one main conference presentation | 21:46 |
harmw | Hui: hi! I will, but first let me get my hands on that minute ;) | 21:46 |
sdake | we have basically a full day of conference space | 21:47 |
sdake | an entire room :) | 21:47 |
Hui | yes, I realized main presentation about kolla plus ansible | 21:47 |
Hui | nice | 21:47 |
*** thumpba has joined #kolla | 21:47 | |
*** jruano has quit IRC | 22:04 | |
*** jtriley has joined #kolla | 22:06 | |
*** thumpba has quit IRC | 22:11 | |
*** thumpba has joined #kolla | 22:13 | |
*** thumpba has quit IRC | 22:28 | |
*** thumpba has joined #kolla | 22:29 | |
*** thumpba has quit IRC | 23:00 | |
*** thumpba has joined #kolla | 23:13 | |
*** dwalsh has joined #kolla | 23:15 | |
openstackgerrit | Steven Dake proposed openstack/kolla: Install agent dependencies from source for CentOS https://review.openstack.org/231754 | 23:22 |
*** thumpba has quit IRC | 23:28 | |
*** blahRus has quit IRC | 23:34 | |
*** thumpba has joined #kolla | 23:36 | |
*** achanda_ has joined #kolla | 23:36 | |
*** achanda has quit IRC | 23:40 | |
*** cloudnautique has joined #kolla | 23:41 | |
*** achanda_ has quit IRC | 23:41 | |
*** sdake_ has joined #kolla | 23:41 | |
*** cloudnautique has quit IRC | 23:42 | |
*** jruano has joined #kolla | 23:42 | |
*** sdake has quit IRC | 23:43 | |
openstackgerrit | Merged openstack/kolla: Fix incorrect dd param in cinder-guide.rst https://review.openstack.org/231544 | 23:44 |
*** thumpba_ has joined #kolla | 23:44 | |
*** thumpba has quit IRC | 23:44 | |
*** thumpba has joined #kolla | 23:45 | |
*** thumpba has quit IRC | 23:46 | |
*** thumpba_ has quit IRC | 23:49 | |
*** sdake_ has quit IRC | 23:52 | |
*** dwalsh has quit IRC | 23:53 |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!