| asalkeld | other option is to break bps into bugs | 00:00 |
|---|---|---|
| asalkeld | (just saying) | 00:00 |
| sdake | asalkeld too lazy for that :) | 00:00 |
| sdake | we are talking like 30-40 containers | 00:00 |
| *** jasonsb has quit IRC | 00:01 | |
| sdake | this is a perfect test of work items for distributed work - if your willing to distribute we can give it a go - i'll get blueprint setup | 00:02 |
| *** jasonsb_ has quit IRC | 00:02 | |
| sdake | SamYaple ^^ | 00:04 |
| SamYaple | sure, thats fine | 00:06 |
| sdake | cool | 00:06 |
| sdake | got a sample of what yu have been prototypingg for the structure of work items? | 00:06 |
| SamYaple | launchpad sets the strucuture | 00:09 |
| SamYaple | task: STATUS | 00:09 |
| sdake | we can't put a name next to the task? | 00:10 |
| asalkeld | "(asalkeld) do this thing": DONE | 00:10 |
| sdake | are we going to drop permissions inside contianer or outside container | 00:11 |
| sdake | samyaple ^^ | 00:11 |
| SamYaple | we can | 00:11 |
| *** jasonsb has joined #kolla | 00:12 | |
| sdake | either i or SamYaple is lagged ;-) | 00:13 |
| *** tpot has joined #kolla | 00:15 | |
| sdake | well i guess we will experiment with outside the container privilege drop | 00:16 |
| asalkeld | that means I have to solve this seperately? | 00:16 |
| asalkeld | (for mesos case) | 00:17 |
| sdake | i think thats what it means | 00:19 |
| asalkeld | :-( | 00:19 |
| sdake | its just a docker flag | 00:19 |
| asalkeld | ok | 00:19 |
| sdake | the big challenge for you will be adding the docker flag to mesos | 00:19 |
| SamYaple | what is "outside the container priviliege drop"? | 00:19 |
| SamYaple | just running ansible as non-root? | 00:19 |
| sdake | docker run -u, --user= Username or UID (format: <name|uid>[:<group|gid>]) | 00:20 |
| SamYaple | no we would bake that into the Dockerfile | 00:20 |
| sdake | you mean inside the contianer priv drop? | 00:20 |
| SamYaple | no i mean set USER in hte dockerfile | 00:21 |
| sdake | cool didnt know that was a flag | 00:21 |
| *** vilobhmm11 has quit IRC | 00:21 | |
| SamYaple | check out yaodu | 00:21 |
| SamYaple | this work is all done | 00:21 |
| sdake | so asalkeld this solves your case as well | 00:21 |
| sdake | nice | 00:21 |
| asalkeld | yeay | 00:21 |
| SamYaple | sdake: was talking in #infra and im moving forward on the true multinode gating stuff | 00:22 |
| SamYaple | start with two nodes then move on | 00:23 |
| sdake | dnd didn't work? | 00:23 |
| SamYaple | it does | 00:23 |
| SamYaple | but the OOM is a killer | 00:23 |
| sdake | right | 00:23 |
| sdake | what did infra say that changed your mind | 00:23 |
| SamYaple | nothing i changed it yesterday when i figured out how to bypass the registry | 00:23 |
| sdake | cool | 00:24 |
| SamYaple | with stripping down services i still only have 1G usable ram | 00:24 |
| SamYaple | so its just not viable i dont think | 00:24 |
| sdake | right | 00:24 |
| SamYaple | i mean we could really limit the nodes,but then is that a real test? | 00:24 |
| sdake | well booting 1 vm would be hugely beneficial to what we have now | 00:24 |
| sdake | it would test out compute kit in a dead chicken sort of way | 00:24 |
| sdake | but gree we want real tests not dead chicken tests | 00:25 |
| SamYaple | the fact is we don't even validate the containers are running yet, we need to do that | 00:27 |
| SamYaple | pbourke: any thoughts on the tempest stuff | 00:28 |
| *** diogogmt has joined #kolla | 00:29 | |
| asalkeld | sdake: that demo/admin thing is more about the internal/external address than the user/project right? cos' that's confusing | 00:29 |
| diogogmt | sdake: do you guys have a continuos integration pipeline where you can automate the build and push of the images to docker hub? | 00:30 |
| SamYaple | diogogmt: no | 00:30 |
| SamYaple | and we likely won't until dockerhub is better | 00:31 |
| diogogmt | SamYaple: better in which sense? | 00:31 |
| SamYaple | right now we end up pushing the same layer n-number of times and the push takes 12 hours | 00:31 |
| SamYaple | there are several critical bugs with docker 2.0 and pushing | 00:31 |
| sdake | docker registry 2.0 | 00:34 |
| SamYaple | right that | 00:34 |
| diogogmt | SamYaple: but even if it takes a while to push the images, by having the pipeline in place whenever docker hub improves their layer caching machanism the pipeline would already be in place to take the benefits from the speed improvement | 00:34 |
| SamYaple | i agree diogogmt. we are time contrained | 00:34 |
| sdake | diogogmt infra wont let us hve a job run longe then 90 minutes | 00:34 |
| diogogmt | what i’m trying to say is that by having official images for the docker services would reduce the barrier of people trying kolla | 00:34 |
| sdake | i may be able to negotiate 120 mins | 00:34 |
| sdake | but thats maxed oout | 00:34 |
| sdake | diogogmt nothing can be done on our end to solve the problem but i agree with your statement | 00:35 |
| diogogmt | sdake: oh i see, so would it be 120mins to build and push all images? | 00:35 |
| sdake | that is our limit | 00:35 |
| sdake | actually our limit is 90 minutes | 00:35 |
| sdake | it takes me 5 hours | 00:35 |
| diogogmt | can’t you parallelize the jobs for each image individually? how many nodes can you scale horizontally? | 00:35 |
| asalkeld | sdake: what do you mean by "CR"? | 00:36 |
| sdake | cr = core reviewer i think | 00:36 |
| sdake | asalkeld depends on context | 00:37 |
| sdake | typically whe n I say cr, I mean core reiewer | 00:37 |
| asalkeld | yeah, never heard that | 00:37 |
| asalkeld | brb rebooting | 00:37 |
| *** asalkeld has quit IRC | 00:37 | |
| sdake | diogogmt we could push sevearl images in parlale with docker in docker | 00:38 |
| sdake | but each new push would push all the same old data | 00:38 |
| sdake | because docker is broke | 00:38 |
| *** asalkeld has joined #kolla | 00:39 | |
| sdake | https://github.com/docker/docker/issues/14018 | 00:39 |
| *** mbound has quit IRC | 00:40 | |
| diogogmt | fair enough, so the plan is to wait for docker to fix the issue with the layer checksum and later implement a ci pipeline? | 00:40 |
| sdake | adding to the ci pipeline a push operation is a 1-2 liner patch | 00:41 |
| sdake | so yes blocked on docker | 00:41 |
| sdake | then blocked possibly on security | 00:41 |
| sdake | becaue our credentials would have to go in the gate | 00:42 |
| sdake | thsi is a more complex problem to solve | 00:42 |
| sdake | the infra guys gave me hope this can be solved | 00:42 |
| sdake | or infra individuals rather | 00:42 |
| openstackgerrit | Sam Yaple proposed openstack/kolla: Add new gate for multinode https://review.openstack.org/242282 | 00:43 |
| diogogmt | i’m not very familiar with the internal processes to build and release openstack services but i see what you are saying | 00:43 |
| diogogmt | just by curiosity, who runs the infrastructure for openstack? | 00:43 |
| sdake | docker requires credentials to push to hub | 00:43 |
| sdake | diogogmt openstack | 00:43 |
| sdake | we need those credentials in the gate job | 00:43 |
| sdake | we dont just want to check those straight into our repo | 00:44 |
| sdake | they need to come from a secure server that only infra can access | 00:44 |
| sdake | anyway as I said this problem is unsolved | 00:44 |
| sdake | but solveable | 00:44 |
| sdake | blocked on docker atm ;) | 00:44 |
| SamYaple | we can push several images in parallel right now | 00:44 |
| SamYaple | dockerhub has no problem with that | 00:44 |
| diogogmt | would be nice if docker hub had access keys in place to segregate access in a rbac model | 00:45 |
| sdake | SamYaple when I do that the push happens completely for all layers | 00:45 |
| sdake | so it doesn't actually help | 00:46 |
| sdake | but ya, it does work | 00:46 |
| openstackgerrit | Sam Yaple proposed openstack/kolla: Add new gate for multinode https://review.openstack.org/242282 | 00:46 |
| SamYaple | sdake: it does work you can do it in a way that works | 00:47 |
| SamYaple | i typically do | 00:47 |
| sdake | hmm would thatwork for us in 90 mins then? | 00:47 |
| SamYaple | no we cant push from the gate | 00:48 |
| sdake | because? | 00:48 |
| sdake | SamYaple on a orthogonal topic why not adopt this push method in --push? | 00:49 |
| SamYaple | sdake: it only works when pushing to the dockerhub | 00:50 |
| sdake | roger | 00:50 |
| SamYaple | sdake: it still takes longer than 90 minutes and the disks are slow in te gate | 00:50 |
| *** dims has joined #kolla | 00:52 | |
| SamYaple | ok infra helped me out | 01:02 |
| *** tfukushima has joined #kolla | 01:03 | |
| SamYaple | we should have an experiemental 2-node gate :) | 01:04 |
| SamYaple | sdake: https://review.openstack.org/#/c/242282/ if you want multinode gating, lets get this reviewed faster than infra | 01:08 |
| *** kjelly has joined #kolla | 01:12 | |
| *** ssurana has quit IRC | 01:16 | |
| *** ssurana has joined #kolla | 01:16 | |
| sdake | there ya go sam, experiment launched :) | 01:16 |
| *** dims has quit IRC | 01:17 | |
| sdake | SamYaple i gotta run out for 1 hour to pick up my kids glasses, but I'll be back to look at that patch you requested above | 01:17 |
| sdake | bbiaf | 01:17 |
| *** ssurana has quit IRC | 01:21 | |
| *** sdake has quit IRC | 01:22 | |
| *** sdake has joined #kolla | 01:24 | |
| sdake | groan sounded like an idiot in my email | 01:25 |
| sdake | oh well not the first or last time ;) | 01:25 |
| *** jasonsb has quit IRC | 01:26 | |
| britthouser | Read the email, but didn't see where you sounded like an idiot sdake? | 01:44 |
| *** tobe has joined #kolla | 01:55 | |
| sdake | britthouser i repeated the deadline twice | 02:07 |
| SamYaple | sdake: youre right you sounded like an idiot. is that what yo uwant to hear? are you happy now? | 02:09 |
| sdake | huh? | 02:09 |
| britthouser | Ah...I just took that as being emphatic. =) | 02:09 |
| SamYaple | sdake: your emails are fine | 02:10 |
| sdake | thanks | 02:10 |
| *** tzn has quit IRC | 02:10 | |
| sdake | i like the etherpads because they have underline and crossout and colors and stuff | 02:10 |
| *** bmace__ has quit IRC | 02:10 | |
| sdake | but we can always combine if needed | 02:11 |
| sdake | for something simple like this work items should do the trick | 02:11 |
| *** bmace__ has joined #kolla | 02:11 | |
| SamYaple | im not quite sure we had the same vision here | 02:11 |
| sdake | feel free to change asy ou like | 02:12 |
| sdake | which part did you not like? | 02:12 |
| SamYaple | i was expecting a larger blueprint with large workitems (drop root user: TODO, drop CAPs: TODO, etc) rather than broken out per service like this | 02:12 |
| sdake | i se | 02:13 |
| *** stvnoyes has quit IRC | 02:13 | |
| SamYaple | i dont neccessarily think this is the right way t ouse work items | 02:13 |
| sdake | part of this blueprint is to recruit new dev | 02:13 |
| sdake | i hear what your saying | 02:13 |
| SamYaple | sure im just letting you know | 02:13 |
| sdake | we can adopt that model going forward | 02:13 |
| sdake | and kick off dev | 02:13 |
| * britthouser took the bait | 02:13 | |
| sdake | most of the core team sleeping atm :) | 02:13 |
| sdake | need to wake the dogs :) | 02:13 |
| SamYaple | mwhahaa | 02:13 |
| *** stvnoyes has joined #kolla | 02:14 | |
| sdake | SamYaple once every 11 years cisco pays 27 times per year rather then 26 | 02:15 |
| sdake | and i'm so lucky it happend my first year | 02:15 |
| * sdake groans | 02:15 | |
| sdake | -ETOOMUCHTAXES | 02:15 |
| kjelly | Now I awake :) | 02:16 |
| SamYaple | sdake: thats.... wierd | 02:16 |
| *** unicell has quit IRC | 02:17 | |
| britthouser | Its just that 14 doesn't divide 365 evenly. | 02:17 |
| britthouser | I think cisco was late to catch it this time. Last time they gave us the heads up MUCH sooner in the year. | 02:18 |
| sdake | finance guy A to finance guy B: "this is going to be a bad quarter." | 02:18 |
| sdake | "Why?" | 02:18 |
| sdake | "We ehave to make payroll one more time this year" | 02:18 |
| sdake | "groaaan." | 02:18 |
| SamYaple | my companies pays twice a month | 02:19 |
| sdake | ya so did red hat | 02:19 |
| sdake | where my wife works ansible paid twice a month too | 02:19 |
| sdake | i like tiwce a month an every 2nd friday | 02:19 |
| sdake | because my paycheck rotates aroudn and robyns stays the sme | 02:19 |
| sdake | its pretty sweet | 02:19 |
| SamYaple | i just budget so i use last months pay for this months bills | 02:20 |
| SamYaple | i always know how much i have t osepnd | 02:20 |
| sdake | ya my dad is on a yearly budget using that technique | 02:20 |
| sdake | he pays hismelf once a year | 02:21 |
| sdake | on the 31st | 02:21 |
| sdake | of dec | 02:21 |
| britthouser | no tax pre-payments that way? | 02:21 |
| sdake | roger | 02:21 |
| sdake | no estimated taxes ftw | 02:21 |
| britthouser | Yup! | 02:22 |
| sdake | i am suppsoed to make estimated tax payments but too lazy to sort that crap out | 02:22 |
| sdake | doing my taxes one a year is trying | 02:23 |
| sdake | ocne a yer that is | 02:23 |
| sdake | can't imgine having to do it 4 times a year | 02:23 |
| britthouser | Yeah no thanks! once a year is just frequent enough that I've forgotten how I did it last year. | 02:23 |
| sdake | i am joe turbotax ftw | 02:24 |
| britthouser | me too, but I forget how to use turbotax. =P | 02:24 |
| sdake | i can't take any deductions or i have to pay amt | 02:24 |
| sdake | so wife (we are not actually married) takes all the deductions | 02:24 |
| britthouser | Ew | 02:24 |
| britthouser | Ew to the AMT, not the wife. =) | 02:25 |
| sdake | ya | 02:25 |
| sdake | i actualy can take 1 deduction | 02:25 |
| sdake | before amt kicks in | 02:25 |
| sdake | amt = lets make it harder | 02:25 |
| sdake | Page 35, "How many physical compute nodes do OpenStack clouds have?" | 02:30 |
| sdake | 10-99: 42% | 02:30 |
| sdake | 1-9: 36% | 02:30 |
| sdake | 100-999: 15% | 02:30 |
| sdake | 1000-9999: 7% | 02:30 |
| sdake | we hit 80% of the openstack deployments yay ;-) | 02:31 |
| britthouser | I'm sure within that 100-999 range, 90% are 100-110 =P | 02:32 |
| SamYaple | and 80% of those "plan on having 100 nodes, but we just have a devstack lab right now" | 02:33 |
| bmace__ | scaled deploy should be an interesting problem to work out | 02:34 |
| sdake | bmace__ define scaled deploy | 02:37 |
| *** tfukushima has quit IRC | 02:42 | |
| *** tfukushima has joined #kolla | 02:43 | |
| sdake | sam did you watch your video of ur talk | 02:45 |
| sdake | SamYaple ^^ | 02:45 |
| SamYaple | a bit of it | 02:45 |
| SamYaple | im quite attractive | 02:46 |
| sdake | lol | 02:46 |
| sdake | we did a solid job | 02:46 |
| sdake | surprised that demo worked :) | 02:46 |
| SamYaple | i think so | 02:47 |
| SamYaple | would prefer config prep ahead of time | 02:47 |
| SamYaple | and more questinos | 02:47 |
| sdake | ya needed more time for q&a | 02:47 |
| sdake | you took longer then i expected to do the config expination | 02:47 |
| sdake | but it is what it is ;) | 02:47 |
| SamYaple | what can i say, i hadnt practiced it one | 02:48 |
| SamYaple | once* | 02:48 |
| SamYaple | its fine | 02:48 |
| sdake | ya like i said solid job | 02:48 |
| SamYaple | i want a lab | 02:48 |
| sdake | ya a lab would rock | 02:49 |
| SamYaple | the novnc stuff borked the nova.conf | 02:56 |
| SamYaple | to many rebases | 02:56 |
| SamYaple | patch incoming | 02:56 |
| sdake | the one I did? | 02:56 |
| SamYaple | no me | 02:57 |
| SamYaple | when i added spice | 02:57 |
| sdake | this is where a boot gat would help ;) | 02:58 |
| SamYaple | it wouldnt have caught this | 02:58 |
| SamYaple | it would have had to try to hit the console | 02:59 |
| sdake | oh | 02:59 |
| sdake | YOU HAVE FAILED ME FOR tHE LAST tiME | 02:59 |
| sdake | THATS IT | 03:00 |
| SamYaple | no multinode gate for you! | 03:00 |
| sdake | THE REBELS ARE THERE | 03:00 |
| sdake | i have a darth vadar destiy decider | 03:00 |
| SamYaple | http://darthjarjar.com/original-theory-by-lumpawarroo/ | 03:01 |
| SamYaple | i want you to read that | 03:01 |
| SamYaple | ironically or non-ironically you need to read that | 03:01 |
| sdake | http://shop.hallmark.com/gifts/darth-vader-destiny-decider-1SHP3003.html | 03:01 |
| SamYaple | you will be a believe in the end | 03:01 |
| SamYaple | first you will laugh. then you will believe | 03:01 |
| sdake | sounds viable | 03:07 |
| sdake | i could se e george chicke nout ;) | 03:08 |
| sdake | because boy, jar jar wa annoying | 03:08 |
| sdake | my favorite character in all tar wars was "dac" | 03:08 |
| sdake | "I feel like I could take on the entire empire" | 03:09 |
| sdake | 5 minutes later -> dead | 03:09 |
| sdake | so epic | 03:09 |
| sdake | and his name | 03:09 |
| sdake | digital analog converter | 03:09 |
| SamYaple | yea | 03:11 |
| SamYaple | man though | 03:12 |
| SamYaple | if that really was jar jars original role, that could have been epic | 03:12 |
| SamYaple | would explain the horribleness of some of the original triology | 03:12 |
| SamYaple | prequel* | 03:12 |
| SamYaple | sdake: i think I am going to convert an existing ceph cluster into the docker version i have in kolla | 03:22 |
| *** cloudnautique has joined #kolla | 03:22 | |
| SamYaple | that ceph role is almost standalone | 03:22 |
| SamYaple | i may push it to the galaxy of ansible | 03:22 |
| *** tzn has joined #kolla | 03:25 | |
| *** tzn has quit IRC | 03:30 | |
| *** tzn has joined #kolla | 03:31 | |
| openstackgerrit | Merged openstack/kolla: Allow disabling of sysctl values https://review.openstack.org/241856 | 03:31 |
| openstackgerrit | Merged openstack/kolla: bootstrap takes place on the wrong node https://review.openstack.org/241989 | 03:32 |
| *** sdake has quit IRC | 03:36 | |
| SamYaple | a huge chunk of my spice patch disappered in a rebase it looks like :( | 03:39 |
| *** Jeffrey4l has joined #kolla | 03:45 | |
| openstackgerrit | Sam Yaple proposed openstack/kolla: Finish implementation of spice https://review.openstack.org/242320 | 03:46 |
| openstackgerrit | Sam Yaple proposed openstack/kolla: Finish implementation of spice https://review.openstack.org/242320 | 03:48 |
| bmace__ | i hope it is an important service, because he who controls the spice, controls the galaxy | 03:59 |
| bmace__ | actually, i think it was universe.. | 04:01 |
| openstackgerrit | Sam Yaple proposed openstack/kolla: Make the database json variable more readable https://review.openstack.org/242024 | 04:01 |
| SamYaple | bmace__: actually that was never in the books! | 04:02 |
| SamYaple | bmace__: for shame! | 04:02 |
| bmace__ | hey, i read all the books! not the drivel written by his son though | 04:02 |
| bmace__ | but i still enjoyed the movie | 04:02 |
| bmace__ | even the scifi mini-series | 04:02 |
| SamYaple | god emperor was pretty drivel | 04:02 |
| SamYaple | and he wrote that | 04:02 |
| SamYaple | sooooo | 04:02 |
| bmace__ | well, the first couple were the best, and it went downhill, but i enjoyed them all to varying degrees | 04:03 |
| SamYaple | first one i read several times a year | 04:03 |
| SamYaple | the second one ive read a few times | 04:03 |
| SamYaple | then the third.... i think ive only read it through once | 04:03 |
| bmace__ | i think you have an extra endif in your spice change, though i'm no jinja expert. | 04:04 |
| openstackgerrit | Sam Yaple proposed openstack/kolla: Make the database json variable more readable https://review.openstack.org/242024 | 04:06 |
| bmace__ | i hate to -1 it since i am not 100% sure.. but if it isn't then jinja syntax is odd :) | 04:06 |
| SamYaple | bmace__: i ran it through | 04:06 |
| SamYaple | it works | 04:06 |
| SamYaple | my guess is you arent tracking the first if clause (enable_nova) | 04:06 |
| bmace__ | sure.. so maybe it can work with a spurious endif? | 04:06 |
| bmace__ | ah.. maybe that is hidden.. so it is way up there | 04:07 |
| SamYaple | in this patchset i and and if/elif/endif block | 04:07 |
| SamYaple | thats all i add, so anything that was this is already there | 04:07 |
| bmace__ | having all this logic shifted to the left makes for terrible nesting context | 04:07 |
| SamYaple | yea thatss why we dont do it in the templates | 04:08 |
| bmace__ | yup, found it, looks good | 04:08 |
| SamYaple | for the docker file | 04:08 |
| SamYaple | haproxy has just sorta grown | 04:08 |
| SamYaple | its also format dependantant to a certain degree | 04:08 |
| SamYaple | change the spacing changes the output | 04:08 |
| *** tfukushima has quit IRC | 04:28 | |
| *** sdake has joined #kolla | 04:35 | |
| sdake | SamYaple could yoou run ceph plan by me again | 04:35 |
| sdake | you want to break it out separately? | 04:35 |
| sdake | or you want to take someone elsess and put it in? | 04:36 |
| SamYaple | sdake: im not quite sure | 04:39 |
| SamYaple | it is pretty much standalone right now | 04:39 |
| SamYaple | it requires nothing from the other services or openstack | 04:39 |
| sdake | i am not keen on taking someone elses code and putting it in | 04:39 |
| SamYaple | nope | 04:39 |
| sdake | if yu want to break it out separately that wfm | 04:39 |
| sdake | the license perits anyone to do that | 04:40 |
| SamYaple | well see. i think it might be better suited as a galaxy role that we consume | 04:40 |
| SamYaple | right | 04:40 |
| sdake | i dont know endough about galaxy consumption to know if that would be good for us or not | 04:40 |
| SamYaple | yea me either | 04:40 |
| SamYaple | just a thought | 04:40 |
| SamYaple | ive found myself using just the ceph role for things outside of kolla | 04:40 |
| sdake | but making it a sparate galaxy thing couldn't hurt the codebase | 04:40 |
| sdake | but depending on it is another q :) | 04:41 |
| SamYaple | yea im not quite sure how i want to go about this to be honest | 04:41 |
| SamYaple | it just seems a waste to appear so tied to kolla when its really not | 04:41 |
| sdake | i am not keen on introducing deps | 04:42 |
| sdake | but having a duplicate is not harmful to kolla | 04:42 |
| SamYaple | i know | 04:43 |
| SamYaple | but im not keep on duplicates :P | 04:43 |
| SamYaple | keen* | 04:43 |
| SamYaple | sdake: you notice the centos-source gate is busted? | 04:44 |
| sdake | i saw this afternoon before i went to pic up kids glaases | 04:45 |
| sdake | but then just woke up from power nap | 04:45 |
| sdake | so i guess i'll ook at it | 04:45 |
| SamYaple | itss been a few days | 04:45 |
| SamYaple | i think it was a key issue | 04:45 |
| openstackgerrit | Merged openstack/kolla: Pin the liberty branch to the liberty RPM repositories https://review.openstack.org/242229 | 04:46 |
| sdake | what key? | 04:46 |
| SamYaple | like a pgp key | 04:47 |
| SamYaple | idk i only looks for a few seconds | 04:47 |
| sdake | i think source and binary centos ar ethee ssame | 04:48 |
| sdake | ryn changed something recently with centos | 04:49 |
| SamYaple | you import the percona key in openstack-base | 04:50 |
| SamYaple | according to the logs | 04:51 |
| SamYaple | idk anything about centos | 04:51 |
| *** sdake has quit IRC | 04:53 | |
| mandre | it looks like it tries to install percona-server-devel instead of mariadb-devel | 04:54 |
| mandre | http://logs.openstack.org/29/242229/1/check/gate-kolla-build-images-centos-source/aa0eef6/console.html#_2015-11-06_04_21_51_453 | 04:54 |
| mandre | and later fails because it conflicts with installed files: http://logs.openstack.org/29/242229/1/check/gate-kolla-build-images-centos-source/aa0eef6/console.html#_2015-11-06_04_22_02_910 | 04:54 |
| *** diga has joined #kolla | 04:54 | |
| SamYaple | i am here to approve patches | 04:55 |
| *** sdake has joined #kolla | 04:58 | |
| sdake | damn build.py is usted | 04:59 |
| *** vilobhmm11 has joined #kolla | 04:59 | |
| *** vilobhmm11 has quit IRC | 05:02 | |
| SamYaple | how so? | 05:06 |
| SamYaple | i just used it | 05:06 |
| SamYaple | on master | 05:06 |
| sdake | build.py -b rhel -t rdo fails | 05:06 |
| sdake | ryan definately broke it | 05:07 |
| SamYaple | the gate built what ryan had | 05:07 |
| SamYaple | i made sure | 05:07 |
| SamYaple | but how is _build.py_ broken | 05:07 |
| sdake | run the command above | 05:07 |
| sdake | he made -t be rhel instead of -b be rhel | 05:07 |
| SamYaple | based on the patch he added, there is no way it was working before | 05:08 |
| SamYaple | so if you thought ti was working, it wasn't | 05:08 |
| SamYaple | https://github.com/openstack/kolla/commit/0340bc4043bd65fbc975c11a86ba89e20293be0a | 05:08 |
| sdake | he just didn't know how to run it apparently | 05:08 |
| SamYaple | while i agree rhel is wrong there, there is no way this worked before | 05:09 |
| sdake | it was working when i commited it | 05:09 |
| SamYaple | i dont see how | 05:09 |
| SamYaple | look there is no rhos tag | 05:09 |
| sdake | the rhos tag is in the bae image | 05:10 |
| sdake | it uses mixed iirc | 05:10 |
| SamYaple | right but look there is not rhos code in build.py | 05:10 |
| SamYaple | this couldnt have worked | 05:10 |
| sdake | profiles patch was prior | 05:10 |
| SamYaple | you broke this with the install type and metatype patch sdake | 05:11 |
| SamYaple | https://github.com/openstack/kolla/commit/0e99b69de48c1deb7f6f1f707e78f445caa5c3d0 | 05:11 |
| sdake | that is where i made it work actually | 05:11 |
| SamYaple | not possible | 05:11 |
| SamYaple | there is no rhos info | 05:11 |
| SamYaple | feel free to look at the code | 05:11 |
| SamYaple | there would be no metatype | 05:12 |
| sdake | ya i am looing | 05:12 |
| sdake | perhaps i left out a commit | 05:12 |
| SamYaple | looks like you need to do -t rhel-rdo | 05:13 |
| SamYaple | you split on - | 05:13 |
| sdake | right | 05:13 |
| sdake | that needs to go in docs | 05:13 |
| SamYaple | anyway, i dont htink ryan broke rhel | 05:13 |
| SamYaple | but i could be wrong, im not rhel head | 05:14 |
| sdake | good news nobody broke anything :) | 05:14 |
| sdake | [sdake@bigiron kolla]$ tools/build.py -t rhel-rdo | 05:16 |
| sdake | also fails | 05:16 |
| sdake | -t rhel fails badly | 05:17 |
| sdake | -b rhel -t rhel is what is needed | 05:17 |
| mandre | so replacing mysql-devel with Percona-Server-devel-55 make centos source openstack-base image build again | 05:19 |
| sdake | mandre cn you geta commit up plz | 05:20 |
| *** vilobhmm11 has joined #kolla | 05:20 | |
| mandre | sure, i haven't yet build all images but the gate will probably build faster than I can | 05:21 |
| SamYaple | you guys really should be using mysql-devel or percona-devel | 05:23 |
| SamYaple | you should be using mariadb-devel | 05:23 |
| sdake | huh | 05:23 |
| sdake | you mised an n't somewhere in those last two sentences | 05:23 |
| mandre | mariadb-devel is already there, i have no idea why mysql-devel was installed | 05:24 |
| sdake | use backport tag please | 05:24 |
| SamYaple | mandre: its old cruft | 05:24 |
| SamYaple | remove it | 05:24 |
| *** tfukushima has joined #kolla | 05:24 | |
| mandre | i will | 05:24 |
| SamYaple | sdake: first sentence shouldnt | 05:24 |
| SamYaple | yuo should NOT be using mysql-devel or percona-server-devel | 05:25 |
| *** tfukushima has quit IRC | 05:27 | |
| *** tfukushima has joined #kolla | 05:28 | |
| openstackgerrit | Martin André proposed openstack/kolla: Remove mysql-devel cruft https://review.openstack.org/242336 | 05:40 |
| mandre | ^ this should fix the centos source gate | 05:42 |
| *** dayou has joined #kolla | 05:47 | |
| openstackgerrit | Lin Hua Cheng proposed openstack/kolla: fix typo in README.rst https://review.openstack.org/242338 | 05:47 |
| *** tpot has quit IRC | 05:48 | |
| openstackgerrit | Jeffrey Zhang proposed openstack/kolla: Use the absoluate path https://review.openstack.org/242339 | 05:56 |
| *** sdake has quit IRC | 06:00 | |
| *** sdake has joined #kolla | 06:00 | |
| openstackgerrit | Martin André proposed openstack/kolla: Remove mysql-devel cruft https://review.openstack.org/242336 | 06:08 |
| mandre | SamYaple, sdake ^^ go crazy on the +2 | 06:09 |
| openstackgerrit | Steven Dake proposed openstack/kolla: Make RHEL build properly https://review.openstack.org/242343 | 06:13 |
| openstackgerrit | Merged openstack/kolla: fix typo in README.rst https://review.openstack.org/242338 | 06:13 |
| *** vilobhmm111 has joined #kolla | 06:18 | |
| *** vilobhmm11 has quit IRC | 06:19 | |
| *** Jeffrey4l has quit IRC | 06:24 | |
| *** unicell has joined #kolla | 06:24 | |
| *** tfukushima has quit IRC | 06:31 | |
| *** openstackgerrit has quit IRC | 06:31 | |
| *** openstackgerrit has joined #kolla | 06:31 | |
| *** tfukushima has joined #kolla | 06:31 | |
| *** tfukushima has quit IRC | 06:34 | |
| *** tfukushima has joined #kolla | 06:35 | |
| *** Jeffrey4l has joined #kolla | 06:36 | |
| *** vilobhmm111 has quit IRC | 06:54 | |
| *** unicell1 has joined #kolla | 07:03 | |
| *** unicell has quit IRC | 07:04 | |
| openstackgerrit | Steven Dake proposed openstack/kolla: Make RHEL build properly https://review.openstack.org/242343 | 07:09 |
| *** Jeffrey4l has quit IRC | 07:09 | |
| openstackgerrit | Jeffrey Zhang proposed openstack/kolla: Use the absoluate path https://review.openstack.org/242339 | 07:11 |
| *** sdake has quit IRC | 07:14 | |
| *** sdake has joined #kolla | 07:14 | |
| *** Jeffrey4l has joined #kolla | 07:21 | |
| SamYaple | mandre: never! | 07:22 |
| openstackgerrit | Merged openstack/kolla: Remove mysql-devel cruft https://review.openstack.org/242336 | 07:24 |
| *** Jeffrey4l has quit IRC | 07:26 | |
| *** sdake has quit IRC | 07:26 | |
| mandre | thanks SamYaple, since you're a good guy I approved your ntp doc change | 07:27 |
| *** sdake has joined #kolla | 07:27 | |
| openstackgerrit | Merged openstack/kolla: Add additional ntp information around ceph https://review.openstack.org/236758 | 07:28 |
| SamYaple | what the what | 07:28 |
| SamYaple | no ways | 07:28 |
| SamYaple | mandre: on the spice on yes huge portions of the implementation were missing | 07:28 |
| SamYaple | the only thing that survived was some nova.conf and group_vars/all.yml changes | 07:29 |
| SamYaple | _and_ it broke deploys | 07:29 |
| SamYaple | so all around, yay! | 07:29 |
| SamYaple | i could live in tokyo | 07:30 |
| SamYaple | i could do that | 07:30 |
| mandre | what do you mean it broke deploys? | 07:32 |
| openstackgerrit | Steven Dake proposed openstack/kolla: Make RHEL build properly https://review.openstack.org/242343 | 07:33 |
| openstackgerrit | Steven Dake proposed openstack/kolla: Make RHEL build properly https://review.openstack.org/242343 | 07:33 |
| sdake | what broke, is stable busted? | 07:37 |
| *** sdake_ has joined #kolla | 07:40 | |
| *** Jeffrey4l has joined #kolla | 07:40 | |
| kjelly | Hi, I want to work on the bug. https://bugs.launchpad.net/kolla/+bug/1513018 murano ui is needed to be installed extra. where is the palce to put install script? dockerfile or ansible ? | 07:40 |
| openstack | Launchpad bug 1513018 in kolla "murano gui does not work" [Critical,New] - Assigned to Kuo-tung Kao (jelly) (coding1314) | 07:40 |
| sdake_ | kjelly no idea we haven't sorted out plugins yet | 07:41 |
| kjelly | it seems that ansible is good way | 07:41 |
| sdake_ | feel free to tackle it ;) | 07:42 |
| kjelly | sdake_: Ok. | 07:42 |
| *** sdake has quit IRC | 07:42 | |
| openstackgerrit | Merged openstack/kolla: Use the absoluate path https://review.openstack.org/242339 | 07:43 |
| sdake_ | mandre can you ack https://review.openstack.org/#/c/242282/ | 07:48 |
| *** Jeffrey4l has quit IRC | 07:49 | |
| *** shardy has joined #kolla | 07:49 | |
| openstackgerrit | Merged openstack/kolla: Add new gate for multinode https://review.openstack.org/242282 | 07:52 |
| SamYaple | kjelly: sdake_ the murano thing is part of the bigger "horizon external plugins" stuff | 08:00 |
| openstackgerrit | Sam Yaple proposed openstack/kolla: Remove unused setup_fedora.sh https://review.openstack.org/242364 | 08:05 |
| openstackgerrit | Sam Yaple proposed openstack/kolla: Adjust gate btrfs setup https://review.openstack.org/242365 | 08:05 |
| openstackgerrit | Sam Yaple proposed openstack/kolla: Restrucutre gate scripts https://review.openstack.org/242366 | 08:05 |
| openstackgerrit | Sam Yaple proposed openstack/kolla: bootstrap takes place on the wrong node https://review.openstack.org/242367 | 08:05 |
| openstackgerrit | Sam Yaple proposed openstack/kolla: Add new gate for multinode https://review.openstack.org/242368 | 08:05 |
| openstackgerrit | Sam Yaple proposed openstack/kolla: Allow specifying erasure as the pool type https://review.openstack.org/242371 | 08:09 |
| openstackgerrit | Sam Yaple proposed openstack/kolla: Allow ceph pool names to be configurable https://review.openstack.org/242372 | 08:09 |
| openstackgerrit | Sam Yaple proposed openstack/kolla: Automate setting up a cache tier https://review.openstack.org/242373 | 08:09 |
| openstackgerrit | Sam Yaple proposed openstack/kolla: Add new gate for multinode https://review.openstack.org/242368 | 08:13 |
| openstackgerrit | Sam Yaple proposed openstack/kolla: Restrucutre gate scripts https://review.openstack.org/242366 | 08:13 |
| openstackgerrit | Sam Yaple proposed openstack/kolla: bootstrap takes place on the wrong node https://review.openstack.org/242367 | 08:13 |
| openstackgerrit | Sam Yaple proposed openstack/kolla: Remove unused setup_fedora.sh https://review.openstack.org/242364 | 08:14 |
| openstackgerrit | Sam Yaple proposed openstack/kolla: Adjust gate btrfs setup https://review.openstack.org/242365 | 08:14 |
| openstackgerrit | Sam Yaple proposed openstack/kolla: HAProxy does not have murano-api configuration https://review.openstack.org/242379 | 08:20 |
| openstackgerrit | Sam Yaple proposed openstack/kolla: Add missing murano-api to site.yml https://review.openstack.org/242380 | 08:20 |
| openstackgerrit | Sam Yaple proposed openstack/kolla: Download, install, and enable rabbitmq_cluster https://review.openstack.org/242381 | 08:20 |
| openstackgerrit | Sam Yaple proposed openstack/kolla: Use rabbitmq-clusterer when deploying https://review.openstack.org/242382 | 08:20 |
| openstackgerrit | Sam Yaple proposed openstack/kolla: Small refactor for site.yml https://review.openstack.org/242383 | 08:20 |
| openstackgerrit | Sam Yaple proposed openstack/kolla: Remove vip for rabbitmq https://review.openstack.org/242384 | 08:20 |
| openstackgerrit | Sam Yaple proposed openstack/kolla: heat containers depends on 'common' (rsyslog, log_data, kolla-ansible) https://review.openstack.org/242385 | 08:26 |
| *** sdake_ has quit IRC | 08:33 | |
| *** zigo has quit IRC | 08:37 | |
| *** zigo has joined #kolla | 08:42 | |
| *** openstackstatus has quit IRC | 08:42 | |
| openstackgerrit | Martin André proposed openstack/kolla: Remove mysql-devel cruft https://review.openstack.org/242394 | 08:42 |
| *** akwasnie has joined #kolla | 08:47 | |
| *** kproskurin has joined #kolla | 08:50 | |
| *** mbound has joined #kolla | 08:59 | |
| *** mbound has quit IRC | 09:02 | |
| *** mbound has joined #kolla | 09:02 | |
| kproskurin | Hi all | 09:03 |
| kproskurin | mandre: Mind to disscuss your comment on https://review.openstack.org/#/c/240827 ? | 09:03 |
| mandre | hey kproskurin | 09:04 |
| SamYaple | any cores around ecept mandre? | 09:05 |
| SamYaple | pbourke: coolsvap jpeeler harmw ? | 09:05 |
| kproskurin | Hi! I agree on var and dir vs file, but debian thing… well, it’s a patch to support *Ubuntu* and, hopefully, not break other things. May be it’s same for debian… I not sure | 09:05 |
| SamYaple | guy we are slower than infra! | 09:06 |
| mandre | yeah, I understand you're trying to fix ubuntu, but if we can generalize and fix all debian-based distros I'd much prefer that | 09:07 |
| kproskurin | “all debian-based distros” is more than Debian + Ubuntu :-) Oh, I could try to test Debian, but just my “patch scope” | 09:09 |
| *** tzn has quit IRC | 09:10 | |
| mandre | kproskurin, if you want you can do it in a follow up patch | 09:10 |
| mandre | i don't want to block your patch | 09:10 |
| kproskurin | I just not sure if Debian is supported by Kolla at all. :-) | 09:11 |
| *** cemmason has joined #kolla | 09:11 | |
| kproskurin | And openstack in general | 09:11 |
| mandre | I suspect ubuntu just install pip packages in the same location as debian does, it would be good to confirm | 09:12 |
| nihilifer | yep, agree, probably all "debian family" distros have the similar policy regarding pip. but I also agree that Debian should be covered in the separate patch | 09:12 |
| nihilifer | because probably no one here has environment with Debian | 09:13 |
| *** diga has quit IRC | 09:16 | |
| kproskurin | Yep | 09:16 |
| SamYaple | i use debian.... | 09:18 |
| cemmason | you would! | 09:18 |
| SamYaple | go back to your hole cemmason! | 09:18 |
| cemmason | freenode is my hole! | 09:18 |
| cemmason | =P | 09:18 |
| SamYaple | i thought you decided to put yourself into a coma until fallout 4 came out | 09:19 |
| SamYaple | waiting was to hard | 09:19 |
| cemmason | *cries* | 09:19 |
| *** achanda has quit IRC | 09:21 | |
| *** jmccarthy has quit IRC | 09:22 | |
| mandre | SamYaple, where are your pip packages installed? | 09:22 |
| kproskurin | asking the right question! | 09:23 |
| SamYaple | debian based systems use /usr/local/ | 09:23 |
| kproskurin | git clone kolla | 09:23 |
| kproskurin | as expected | 09:23 |
| *** jmccarthy has joined #kolla | 09:24 | |
| SamYaple | if you look at that crazy sed in docker/base/Dockerfile.j2 for ubuntu it actually changes that for our images | 09:24 |
| SamYaple | unrelated just thought you should know | 09:24 |
| kproskurin | So guys, sould I add Debian to a list of checked OS in my patch for /usr/local/? Or make it a separate patch? | 09:24 |
| mandre | isn't there a generic way to find the distro family? | 09:25 |
| mandre | or you can check for existence of /usr/share/kolla or /usr/local/share/kolla, so that it will always find the right place | 09:28 |
| *** gfidente has joined #kolla | 09:32 | |
| *** gfidente has joined #kolla | 09:32 | |
| kproskurin | Oh well, then I start to write this patch I was suprissed what there is no easy and one way on every linux distro to actualy tell what is this distro name. Personaly, I think checking /usr/local/../kolla vs /usr/…/kolla is bad way to do things. It’s not really relaible. | 09:38 |
| *** tzn has joined #kolla | 09:48 | |
| *** tzn has quit IRC | 09:54 | |
| *** tobe has quit IRC | 09:55 | |
| SamYaple | kproskurin: /win 18 | 09:57 |
| kproskurin | whhhaaat?.. | 09:58 |
| SamYaple | clearly i was asking you to join me in /win 18 | 10:00 |
| SamYaple | :) | 10:00 |
| SamYaple | kproskurin: so anyway i dont agree with checking /usr/local either | 10:00 |
| SamYaple | but we can throw in a small debian check to make everyone happy | 10:00 |
| kproskurin | Let’s make it separate patch then. I really want to merge this and commit fixes to new doc after. And I still don’t get “/win 18” joke(?) %-) Sry! | 10:02 |
| britthouser | So I"m going through the Kolla Dev Quickstart guide. It says to use docker 1.8.2 b/c 1.8.3 is busted. But I know 1.9.0 just came out. Anybody tried it with kolla yet? | 10:03 |
| SamYaple | britthouser: anything past 1.8.2 will not work | 10:04 |
| SamYaple | not until ansible 2.0 | 10:04 |
| SamYaple | kproskurin: /win 18 was a typo. i use irssi and i was trying to change windows | 10:05 |
| kproskurin | oh %-) | 10:05 |
| britthouser | thanks SamYaple. is the details of the incompatibility documented somewhere? Just want to satisfy my curiousity | 10:07 |
| SamYaple | britthouser: docker-devs in thier infinity wisdom changed the docker api between version without bumping the api version | 10:08 |
| SamYaple | the ansible module blew up | 10:08 |
| SamYaple | ansible modules are fixed now, but ansible wont cut another 1.9.x tag | 10:08 |
| SamYaple | so we have to wait for 2.0 and then upgrade all the playbooks | 10:08 |
| openstackgerrit | Proskurin Kirill proposed openstack/kolla: Few improvements in Kolla tools. Added Ubuntu support. https://review.openstack.org/240827 | 10:09 |
| *** dayou has quit IRC | 10:09 | |
| SamYaple | britthouser: https://github.com/ansible/ansible-modules-core/pull/2258 | 10:09 |
| *** athomas has joined #kolla | 10:09 | |
| britthouser | Ok cool. We hit some issues with docker1.9 API changing as well. | 10:10 |
| SamYaple | britthouser: i may or may not be rolling an ansible docker module for kolla to control | 10:11 |
| SamYaple | ansible has proven unreliable here | 10:11 |
| SamYaple | since the docker module is critical to us.... | 10:11 |
| *** Slower has quit IRC | 10:11 | |
| britthouser | Yeah if everyone rolls forward and nobody backports, then its just throwing darts. =) | 10:12 |
| SamYaple | nope | 10:12 |
| SamYaple | its backported | 10:12 |
| SamYaple | made sure of that myself | 10:12 |
| SamYaple | ansible wont snap a new 1.9.x tag | 10:13 |
| SamYaple | hard no | 10:13 |
| britthouser | backported but never released? | 10:13 |
| *** mfalatic has joined #kolla | 10:13 | |
| SamYaple | bingo | 10:13 |
| britthouser | brilliant | 10:13 |
| *** Slower has joined #kolla | 10:13 | |
| SamYaple | its not the first docker module problem we have had either | 10:13 |
| SamYaple | so rolling our own, how ever non-opensource that is, is the safest option for the stability of kolla | 10:14 |
| *** mfalatic_ has quit IRC | 10:15 | |
| *** slotti has joined #kolla | 10:17 | |
| openstackgerrit | Britt Houser proposed openstack/kolla: Update Centos7 docker commands https://review.openstack.org/242425 | 10:25 |
| kproskurin | Oh, I actually got same fixes in the incoming doc update :-) | 10:32 |
| britthouser | Is it up for review kproskurin? | 10:33 |
| kproskurin | not yet, it’s waiting for other commit as a dependency | 10:33 |
| SamYaple | kproskurin: you can do dependant reviews | 10:34 |
| britthouser | Ok. I can abandon mine if you want. I was also noticing some other inconsistencies around when/how pip is installed. I was gonna fix that next, but can wait if you've got that covered. | 10:35 |
| *** tfukushima has quit IRC | 10:35 | |
| kproskurin | britthouser: “inconsistencies around when/how pip is installed” care to eleborate? I actually working on that too | 10:35 |
| kproskurin | SamYaple: what tag do I need use for that? | 10:36 |
| SamYaple | kproskurin: just make a second patch ontop of the other one | 10:36 |
| SamYaple | then git review | 10:36 |
| SamYaple | it will ask you if you want to push both, say yes | 10:36 |
| nihilifer | I have one silly question regarding keepalived container - was some magic done to succesfully "modprobe ip_vs" in start script? | 10:37 |
| britthouser | Well, e.g. the doc says to pip install requirements at the top, but then lower it says to install pip. Also it says to easy_install pip for Centos, but to install pip from packaging for ubuntu | 10:37 |
| kproskurin | oh, I thought it’s some kind of tag like “Dependend-on: #number” | 10:37 |
| britthouser | I was gonna add "yum install -y epel-release" followed by "yum install -y python-pip" for centos | 10:37 |
| britthouser | But if you have that covered, go for it. | 10:38 |
| SamYaple | kproskurin: depends-on is for cross-repo stuff (like when nova requires a patch to merge in neutron) | 10:38 |
| kproskurin | britthouser: “ pip install requirements” this will go away in new doc. | 10:38 |
| SamYaple | nihilifer: what do you mean, magic? | 10:38 |
| kproskurin | “Also it says to easy_install pip for Centos, but to install pip from packaging for ubuntu” this is actually ok, no? | 10:38 |
| SamYaple | nihilifer: we modprobe in all kinds of places | 10:38 |
| britthouser | Well just not consistent. Why not install pip from packaging on centos? | 10:38 |
| kproskurin | it requires EPEL | 10:39 |
| kproskurin | afaik | 10:39 |
| nihilifer | SamYaple: is making mount /lib/modules:/lib/modules:ro enough to do "modprobe" inside container without any errors? | 10:39 |
| britthouser | Yeah..."yum install epel-release" and then "yum install python-pip" | 10:39 |
| SamYaple | nihilifer: da | 10:39 |
| britthouser | or is epel bad news here? | 10:39 |
| kproskurin | i tjink it’s bad idea to install epel only for pip. And yes, it’s bad in general(personal opinion) | 10:39 |
| britthouser | Ok. Its lesser of two evils in my opinion, but its just that, my opinion. =) | 10:40 |
| britthouser | kproskurin: are you also covering the fact that gcc needs to be installed before doing "pip install -U python-openstackclient" ? | 10:41 |
| kproskurin | yes :) | 10:41 |
| britthouser | Ok...you've got all my bases covered then. =) | 10:42 |
| kproskurin | I could send you full diff if want | 10:42 |
| kproskurin | may be you’ll find something else | 10:42 |
| britthouser | no I'll just wait for review to post | 10:42 |
| SamYaple | kproskurin: just post the patch so we can all review it :) | 10:42 |
| kproskurin | ok, ok. :-) | 10:43 |
| kproskurin | I just wasent sure about dependencies. New doc need my tools patch to be merged and I didnt know how to show it | 10:43 |
| britthouser | so if you've made them as two sequential commits in your same topic branch, then you can just git review on the second commit, and both will be pushed with the correct dependacy | 10:45 |
| *** shardy has quit IRC | 10:46 | |
| britthouser | if you've made them different patches in unique topic branches, then you could rebase the tools patch into the doc branch (or vice versa) and then submit the review. | 10:46 |
| kproskurin | well, it’s a different branches in my git | 10:46 |
| britthouser | just make sure you get the ordering right. | 10:46 |
| kproskurin | git is scary | 10:47 |
| britthouser | You can rebase the patch from one branch to another. | 10:47 |
| britthouser | sdake just taught me this a few months ago. | 10:47 |
| britthouser | =) | 10:47 |
| britthouser | so in your docs branch, do a 'git log' | 10:48 |
| britthouser | copy down the sha of your docs commit | 10:48 |
| britthouser | then switch to bug/1512302 | 10:48 |
| britthouser | git rebase -i origin/master | 10:48 |
| kproskurin | http://imgs.xkcd.com/comics/git.png | 10:49 |
| britthouser | you'll get an editor that says something like "pick 12312 Few improvements...." | 10:49 |
| britthouser | under that line add "pick <sha>" | 10:49 |
| britthouser | then git review | 10:50 |
| britthouser | and it'll ask you about posting both. | 10:50 |
| britthouser | lol kproskurin | 10:50 |
| pbourke | SamYaple: are we backporting clusterer to liberty? | 10:51 |
| pbourke | (or have we) | 10:51 |
| britthouser | pbourke: it was on the list at the bottom of: https://etherpad.openstack.org/p/kolla-mitaka-upgrade | 10:51 |
| pbourke | britthouser: k thanks | 10:52 |
| kproskurin | mandre: oh, you right, I add func and never use it. What is wrong with me… | 10:59 |
| kproskurin | Do you thin it should be a func, tho’? Oneliner… | 10:59 |
| mandre | no worries kproskurin, it happens | 10:59 |
| mandre | up to you, I wouldn't bother writing a function for it personally | 11:00 |
| SamYaple | pbourke: for 1.1 we will be, yes | 11:01 |
| pbourke | im raging at gerrit right now | 11:01 |
| SamYaple | pbourke: https://review.openstack.org/#/c/242364/ <-- that patch series needs to merge. its holding up infra | 11:02 |
| SamYaple | it doesnt include ceph or rabbitmq | 11:02 |
| pbourke | i just moved to firefox and cant get my filters to work http://imgur.com/qted3mU | 11:02 |
| pbourke | does anyone have any useful gerrit filters other than status:open | 11:02 |
| pbourke | SamYaple: can I workflow that patch given the gates are red | 11:03 |
| openstackgerrit | Merged openstack/kolla: Update Centos7 docker commands https://review.openstack.org/242425 | 11:03 |
| SamYaple | pbourke: thats unrelated entirely | 11:03 |
| pbourke | i know but... | 11:03 |
| SamYaple | yes you can | 11:03 |
| openstackgerrit | Proskurin Kirill proposed openstack/kolla: Few improvements in Kolla tools. Added Ubuntu support. https://review.openstack.org/240827 | 11:03 |
| mandre | pbourke, the gate will be back to green after you +a https://review.openstack.org/#/c/242394/ | 11:04 |
| SamYaple | mandre: i think he already did | 11:04 |
| mandre | hint hint | 11:04 |
| mandre | he then retracted :) | 11:05 |
| pbourke | yeah i was concerned about +a with red gates | 11:05 |
| pbourke | got shouted at for that before | 11:05 |
| openstackgerrit | Merged openstack/kolla: Remove unused setup_fedora.sh https://review.openstack.org/242364 | 11:06 |
| SamYaple | pbourke: in this case we are literally the ones holding up infra | 11:06 |
| openstackgerrit | Merged openstack/kolla: Remove mysql-devel cruft https://review.openstack.org/242394 | 11:06 |
| SamYaple | pbourke: all the patches up to https://review.openstack.org/#/c/242368/ have to merge for the infra patch to merge | 11:07 |
| SamYaple | they are cherry-picks, but feel free to -1 if you have a problem | 11:07 |
| SamYaple | with the red gate i mean | 11:07 |
| SamYaple | i dont want to push you if you dont feel comfortable | 11:08 |
| openstackgerrit | Merged openstack/kolla: Adjust gate btrfs setup https://review.openstack.org/242365 | 11:09 |
| openstackgerrit | Merged openstack/kolla: Restrucutre gate scripts https://review.openstack.org/242366 | 11:10 |
| openstackgerrit | Merged openstack/kolla: bootstrap takes place on the wrong node https://review.openstack.org/242367 | 11:10 |
| openstackgerrit | Merged openstack/kolla: Add new gate for multinode https://review.openstack.org/242368 | 11:10 |
| openstackgerrit | Merged openstack/kolla: heat containers depends on 'common' (rsyslog, log_data, kolla-ansible) https://review.openstack.org/242385 | 11:11 |
| SamYaple | so with those patches in line, we will now have an experimental 2 node gate | 11:13 |
| SamYaple | i can do actual multinode gate testing | 11:13 |
| britthouser | w00t! | 11:14 |
| *** pbourke has quit IRC | 11:15 | |
| *** pbourke has joined #kolla | 11:15 | |
| *** achanda has joined #kolla | 11:22 | |
| *** achanda has quit IRC | 11:26 | |
| *** dims has joined #kolla | 11:27 | |
| *** gfidente is now known as gfidente|lunch | 11:40 | |
| *** dayou has joined #kolla | 11:42 | |
| *** akwasnie has quit IRC | 11:50 | |
| *** tzn has joined #kolla | 11:51 | |
| pbourke | mandre: did you ever have errors around ip6tables with vagrant-libvirt ? | 11:54 |
| *** tzn has quit IRC | 11:56 | |
| mandre | pbourke, nope | 12:10 |
| mandre | what kind of issue do you have/ | 12:10 |
| mandre | ? | 12:10 |
| pbourke | mandre: http://paste.openstack.org/show/478175/ | 12:13 |
| pbourke | I haven't figured out why its using ipv6 yet | 12:13 |
| pbourke | or why it thinks it needs to use ip6tables rather than iptables | 12:14 |
| pbourke | ipv6 should be disabled | 12:14 |
| mandre | do you have the right kernel modules loaded? | 12:16 |
| *** rhallisey has joined #kolla | 12:16 | |
| mandre | lsmod | grep ip6 | 12:18 |
| openstackgerrit | Merged openstack/kolla: Few improvements in Kolla tools. Added Ubuntu support. https://review.openstack.org/240827 | 12:22 |
| kproskurin | tx guys! | 12:23 |
| openstackgerrit | Proskurin Kirill proposed openstack/kolla: Few improvements in the quickstart doc https://review.openstack.org/242467 | 12:24 |
| *** achanda has joined #kolla | 12:24 | |
| mandre | kproskurin, you're now ATC, congrats! | 12:25 |
| kproskurin | britthouser: check this docs update, I hope I covered your things | 12:25 |
| kproskurin | mandre: ATC?... | 12:25 |
| mandre | active technical contributor | 12:25 |
| kproskurin | sounds awesome %-) | 12:26 |
| mandre | it means you have at least 1 commit in a project in the openstack namespace | 12:26 |
| kproskurin | hey, it’s my second merged commit! -) | 12:26 |
| kproskurin | not first :-) | 12:26 |
| mandre | what it really means is you will have a free ticket to the openstack summit :) | 12:26 |
| mandre | cool, your other patch took less time to land | 12:28 |
| kproskurin | “ATC codes provide a discount of $600 USD off all ticket prices” - not really free if you’re slowpoke for buying tickets actually. :-) | 12:29 |
| *** achanda has quit IRC | 12:30 | |
| mandre | it's free for early bird registration | 12:31 |
| kproskurin | Guys, just so you know, I started working on adding binary images for Ubuntu. If anyone doing the same, we could split our work. -) | 12:31 |
| mandre | I believe nobody else is working on it at this moment | 12:34 |
| mandre | or if they do they didn't say so | 12:34 |
| kproskurin | Well, you never know. I think we(me + my collegue) could do it during next week or so | 12:36 |
| *** zigo has quit IRC | 12:36 | |
| *** zigo has joined #kolla | 12:37 | |
| *** akwasnie has joined #kolla | 12:40 | |
| britthouser | kproskurin: I added a couple of comments | 12:43 |
| britthouser | gnocchi has no RPMs available...this is expected, right? | 12:51 |
| *** tzn has joined #kolla | 12:52 | |
| *** tzn has quit IRC | 12:56 | |
| kproskurin | britthouser: tx! Looking into it | 13:08 |
| kproskurin | britthouser: “Not sure this section is necessary since we had them install gcc prior to installing the openstackclients.” Well, install of openstackclients is optional, so it could be skiped. I’ll try to move this section a bit | 13:12 |
| britthouser | Yeah that's true. | 13:13 |
| SamYaple | kproskurin: binary packages for ubuntu are pretty much a no go | 13:30 |
| SamYaple | kproskurin: ubuntu doesnt provide a delorean repo at all | 13:30 |
| SamYaple | so they have liberty packages _after_ liberty is released | 13:31 |
| kproskurin | SamYaple: yeah, just discuss it like 1 min ago with collegue… | 13:31 |
| kproskurin | That is sad | 13:32 |
| SamYaple | its ok, te cloud-archive packages are always bad and broken | 13:32 |
| SamYaple | just like delorean | 13:32 |
| SamYaple | source is far superior for this type of thing | 13:32 |
| kproskurin | Building from source it’s like… I dunno, “bad”, for me, personaly | 13:33 |
| britthouser | So we could only have ubuntu binary in the stable branch, but not master, right? | 13:35 |
| SamYaple | right, but we arent merging all new code into a stable branch like that | 13:35 |
| SamYaple | kproskurin: its build from source once, then its packaged (docker container) | 13:36 |
| britthouser | Yeah that makes total sense. | 13:36 |
| *** akwasnie has quit IRC | 13:38 | |
| SamYaple | kproskurin: i dont know how much youve done with openstack from source, but when done right its far and away superior to the packaged versions of openstack. for all kinds of reasons | 13:41 |
| *** diogogmt has quit IRC | 13:49 | |
| SamYaple | britthouser: if you notice I actually _do_ include cloud-archive for packages | 13:50 |
| SamYaple | we take advantage of the newer qemu and rabbitmq and ceph etc | 13:50 |
| SamYaple | cananical has commited to maintaining security patches for those packages and they are newer than the old 14.04 packages | 13:51 |
| SamYaple | so its actually a pretty good balance | 13:51 |
| britthouser | Source for openstack, cloud-archive for the supporting packages? | 13:51 |
| SamYaple | correct | 13:51 |
| SamYaple | cloud-archive pulls from sid _or_ packages head of upstream for the project at the time of repo creation | 13:52 |
| SamYaple | so its actually really new stuff | 13:52 |
| *** tzn has joined #kolla | 13:52 | |
| *** diogogmt has joined #kolla | 13:53 | |
| britthouser | Makes sense. In the ops meetings the past couple of summits there has been a "lets get together and build packages together" initiative. Nobody has actually stepped up todo anything yet. But if that does get off teh ground, perhaps there would be packages for ubuntu prior to release. | 13:53 |
| britthouser | I'm not holding my breathe, but something to keep tabs on at least. | 13:54 |
| SamYaple | ops? build packages? you let me know when that happens! (I am an ops guy btw ;) | 13:54 |
| britthouser | I have a (probably dumb) questions about https://blueprints.launchpad.net/kolla/+spec/drop-root | 13:55 |
| SamYaple | shoot | 13:55 |
| britthouser | So I'm gonna make the container run as a different user | 13:55 |
| britthouser | different user = process I'm running (eg. keystone container runs as keystone) | 13:56 |
| britthouser | is that right? | 13:56 |
| SamYaple | yea | 13:56 |
| britthouser | but if my base OS has no keystone packages installed.....I have no keystone user? | 13:56 |
| britthouser | so this will require creating the user as well? | 13:56 |
| SamYaple | oh well we create the keysont/glance/neutron users already | 13:56 |
| SamYaple | although keystone is a bad example, since that runs as apache2 (and apache2 runs keystone process as user keystone) | 13:57 |
| SamYaple | britthouser: https://github.com/openstack/kolla/blob/master/docker/glance/glance-base/Dockerfile.j2#L39 | 13:57 |
| *** tzn has quit IRC | 13:57 | |
| britthouser | OH...so this is all within the container? | 13:58 |
| SamYaple | yes | 13:58 |
| SamYaple | the containers themselves run as root on the host (docker only runs as root) | 13:58 |
| britthouser | Ok I get it now. I was looking in /etc/passwd on the host for user keystone. | 13:58 |
| britthouser | I'm on the same page now I think. | 13:59 |
| britthouser | and I see what you're saying no in the keystone container. httpd running as apache (except for PID1), and keystone running as keystone. | 14:00 |
| britthouser | So the only diffrence I would expect to see when I add USER to the keystone container is that PID1 woudl change from root to apache, right? | 14:01 |
| *** absubram has joined #kolla | 14:01 | |
| SamYaple | for keystone containrer it will be different, not quite sure how it will look | 14:02 |
| SamYaple | from a glance perspective youll see pid 1 be running as glance | 14:02 |
| SamYaple | look at what i have set in the yaodu project | 14:02 |
| SamYaple | thats the final peice yoadu has to offer kolla | 14:02 |
| britthouser | Yeah I see the USER {{ container }} | 14:03 |
| britthouser | ok, when I login the glance container today, I see all the glance-api running as root. That's the part we're trying to fix. | 14:04 |
| * britthouser may try an easier container than keystone first. | 14:04 | |
| *** absubram_ has joined #kolla | 14:06 | |
| *** absubram has quit IRC | 14:06 | |
| *** absubram_ is now known as absubram | 14:06 | |
| *** slotti has quit IRC | 14:06 | |
| openstackgerrit | Proskurin Kirill proposed openstack/kolla: Few improvements in the quickstart doc https://review.openstack.org/242467 | 14:15 |
| kproskurin | britthouser: I hope it’s better now | 14:15 |
| *** sdake has joined #kolla | 14:16 | |
| sdake | morning | 14:17 |
| SamYaple | mornign sdake | 14:17 |
| sdake | wow sam stlil up? | 14:17 |
| SamYaple | infra approved our multigate before we got the patches backported :P | 14:17 |
| SamYaple | we lost | 14:17 |
| sdake | we los twhat | 14:18 |
| SamYaple | the race to approve a patch faster than infra | 14:19 |
| *** diogogmt has quit IRC | 14:21 | |
| SamYaple | btw sdake i put a whole bunch of backports in the queue | 14:24 |
| SamYaple | -1 now or forever hold your peace | 14:24 |
| sdake | cool did yu backport everything that is required for 1.0.1? | 14:25 |
| SamYaple | i dont know what was required | 14:25 |
| sdake | i guess anything with a backport tag | 14:26 |
| *** diogogmt has joined #kolla | 14:26 | |
| SamYaple | idk i only dug through a few patches | 14:27 |
| sdake | i've been saking people ot add a backport tag tothe patch if I think it is backport material | 14:27 |
| britthouser | ovsdb data container, rabbit clusterer, no-preempt for keepalived, pine ceph to specific release, write a version to /etc/kolla/release | 14:27 |
| SamYaple | my oneline cleared from my history :( | 14:27 |
| *** athomas has quit IRC | 14:27 | |
| SamYaple | britthouser: those are things for 1.1 | 14:27 |
| britthouser | oh right. | 14:27 |
| sdake | although I may not review every change | 14:27 |
| SamYaple | backporting does not boost commiting stats people | 14:28 |
| SamYaple | please do it | 14:28 |
| sdake | backporting is such a pain in teh ass if we rely on one person to do the job | 14:28 |
| sdake | because one person doens't know what another person should have backported | 14:29 |
| sdake | pleae take the time to backport folks | 14:29 |
| SamYaple | i backport everything non-docs and non-breaking | 14:29 |
| *** severion has joined #kolla | 14:31 | |
| sdake | SamYaple can you figure out what your one liner ws to backport and sbmit a backport patch into the repo so anyone can backport? | 14:34 |
| SamYaple | i have unfortunately lost that oneliner and must rebuild it | 14:35 |
| SamYaple | i may or may not have rekicked my nodes | 14:36 |
| SamYaple | turns out bash hsitory is not a great archive for things like that | 14:36 |
| sdake | roger - can yu rebuid it is whati was asking | 14:37 |
| SamYaple | probably | 14:38 |
| SamYaple | but this is more art than science | 14:38 |
| sdake | i woudl wirte one but i'm not sur eehow to get the hsh id based upon the ocntents of the change | 14:38 |
| SamYaple | no two oneliners are alike | 14:38 |
| britthouser | is it just a matter of cherry-picking to stable/liberty? | 14:38 |
| SamYaple | it is! | 14:38 |
| britthouser | gerrit has a button for that. | 14:39 |
| SamYaple | you go click through 200 patches | 14:39 |
| SamYaple | i dare you! | 14:39 |
| britthouser | I thought sdake was asking each person to backport as they patch? | 14:39 |
| *** athomas has joined #kolla | 14:39 | |
| britthouser | maybe I misunderstood | 14:39 |
| sdake | the foolrpoof way is for the person who +a's a backport change to be rewponsible for the backport | 14:40 |
| kproskurin | pbourke: Mind to disscuss your comment in https://review.openstack.org/#/c/242467/ ? | 14:42 |
| pbourke | kproskurin: yes does it not make sense? | 14:42 |
| kproskurin | We disscuss it with britthouser before, well thing is we tell to install gcc in *optional* section, which may be skiped | 14:43 |
| kproskurin | I would like to hear how to make it more clear in this docs, if you have any idea | 14:44 |
| dmsimard | rhallisey: just noticed I didn't actually merge that change from yesterday, just did | 14:44 |
| SamYaple | python-dev on ubuntu pulls in gcc fyi | 14:45 |
| kproskurin | Actually, we could make it mandatory and add some openstack cli examples… but it may make things more complicated | 14:45 |
| SamYaple | sdake: i like that | 14:45 |
| rhallisey | dmsimard, kk | 14:45 |
| SamYaple | sdake: +a must backport | 14:45 |
| *** achanda has joined #kolla | 14:46 | |
| pbourke | kproskurin: ok I guess it wasn't too obvious the first section was optional. No big deal I can +2 | 14:47 |
| *** rbradfor has joined #kolla | 14:47 | |
| kproskurin | pbourke: I agree it a bit confusing and we repeat ourselfs… I just not sure how to fix it | 14:48 |
| kproskurin | On the other hand it’s really not that critical | 14:48 |
| pbourke | kproskurin: yeah its not a big deal | 14:49 |
| pbourke | imo | 14:49 |
| kproskurin | ty | 14:49 |
| *** achanda has quit IRC | 14:51 | |
| *** ashishjain has joined #kolla | 14:59 | |
| *** dayou has quit IRC | 15:02 | |
| *** blahRus has joined #kolla | 15:24 | |
| *** cloudnautique has quit IRC | 15:24 | |
| sdake | SamYaple that aid, we need to sort ou tthe past | 15:29 |
| sdake | samyaple i | 15:29 |
| sdake | i'll float the idea of +a by the cr team individually | 15:29 |
| sdake | pbourke are you willing to participate? | 15:29 |
| sdake | rhallisey ^^ | 15:29 |
| sdake | jpeeler ^6 | 15:29 |
| sdake | mandre ^^ | 15:29 |
| sdake | harmw ^^ | 15:30 |
| pbourke | ? | 15:30 |
| sdake | the basic idea folks is that if you approve a change that has a backport tag, (meaning you +a it) you become responsible for backporting said change | 15:30 |
| pbourke | seems ok | 15:32 |
| jpeeler | uh | 15:32 |
| *** diogogmt has quit IRC | 15:32 | |
| jpeeler | i think maybe whoever +a should make sure the submitter has properly submitted the backport or something like that | 15:32 |
| sdake | that realisticallly just doesn't happen | 15:33 |
| sdake | right now nobody is backporting | 15:33 |
| sdake | except here and there | 15:33 |
| sdake | which is creating a chaotic situation where not all the stuff is being backported | 15:33 |
| jpeeler | it's a new practice, i'm sure if you hammer the point in the meetings people will get it | 15:34 |
| sdake | just want to make sur ecr team is game to participate | 15:34 |
| sdake | the idea that th person who submits the patch also backports - gets lost in teh shffule | 15:34 |
| ashishjain | hello. | 15:34 |
| ashishjain | I look forward to setting up my development environment for kolla. | 15:35 |
| sdake | the worst result it could have is people will be slow to approve backkports becuse they wont want to do the backport work | 15:35 |
| sdake | in which case it will end up on my plate | 15:35 |
| sdake | or pbourkes | 15:35 |
| ashishjain | I would prefer to use the bare metal deployment http://docs.openstack.org/developer/kolla/quickstart.html | 15:35 |
| sdake | or SamYaple since they have agreed to do the work :) | 15:35 |
| ashishjain | I got sufficient RAM & HDD but I got only one network card. | 15:35 |
| ashishjain | How can I go about having another interface? | 15:36 |
| ashishjain | Please advice. | 15:36 |
| sdake | SamYaple can you help ashishjain wit hthe magic incantation to create a virtual nic? | 15:36 |
| jpeeler | sdake: either way, this is a large enough change for the mailing list or a meeting at least | 15:36 |
| sdake | or pbourke and let me know too, so I'll document it | 15:36 |
| sdake | jpeeler will do mailing list agreed | 15:36 |
| sdake | just want ot contact devs indivdually | 15:36 |
| sdake | so you understand the requeted change | 15:37 |
| sdake | we can have a fuller discussion on the l | 15:37 |
| sdake | what it would mean is if yu dont agree to backporting, dont +a backports ;) | 15:37 |
| jpeeler | maybe i'm not understanding. a review goes into master and then if it's +a, that person submits a backport? | 15:38 |
| sdake | ya lets say sam submits a backport | 15:38 |
| *** diogogmt has joined #kolla | 15:38 | |
| sdake | and I +a his backport | 15:38 |
| sdake | rather sumbit a patch with backport: iberty | 15:38 |
| sdake | and i +a his patch | 15:38 |
| sdake | I do the backport | 15:38 |
| sdake | this unloads sam from doing the backport | 15:40 |
| sdake | and makes sure the backport actually gets done syncrhounously | 15:40 |
| sdake | if I dont have immediate time to do the backport I dont +a the change | 15:40 |
| sdake | then they pile up,adn I can complain all at once | 15:40 |
| sdake | and people can backport in chunks or whatever | 15:40 |
| rhallisey | kk | 15:41 |
| jpeeler | i guess the biggest thing is that the author is always going to be the expert in backporting. maybe since kolla branches are so similar it won't matter all that much | 15:41 |
| sdake | how about this, if the backport is a hugef cf the author is responsible for backporting | 15:44 |
| sdake | if its a simple cherrypick then the +a is responsible | 15:45 |
| sdake | and the person that +as is responsible or indicating to the developer and riding herd on him/her to do the backport if a siple cherry-pick doesn't get the job done | 15:45 |
| jpeeler | that sounds like a good compromise | 15:47 |
| sdake | rhallisey or jpeeler can one of you ack https://review.openstack.org/#/c/242343/ | 15:48 |
| sdake | this gets rhel rolling | 15:48 |
| sdake | i will backport it | 15:48 |
| sdake | this time | 15:48 |
| sdake | because it has a dependency that needs a backport | 15:48 |
| sdake | rhallisey or jpeeler do you guys know how to createa a virutal interface for use with kolla | 15:49 |
| sdake | or pbourke | 15:49 |
| jpeeler | it's documented isn't it? | 15:49 |
| pbourke | its not unfortunately | 15:49 |
| pbourke | im not sure | 15:49 |
| sdake | should be :( | 15:49 |
| jpeeler | where did i read it from then... | 15:50 |
| jpeeler | oh i must have looked at the gate patch | 15:50 |
| sdake | i think it was docmented | 15:50 |
| sdake | but maybe not | 15:50 |
| sdake | oh is it in the gate patch? | 15:50 |
| pbourke | SamYaple creates a dummy interface in the gate but that different I think | 15:50 |
| sdake | that is wht i want | 15:50 |
| jpeeler | oh i thought that's what you meant | 15:50 |
| jpeeler | https://github.com/openstack/kolla/blob/master/tests/deploy_aio.sh#L28 | 15:50 |
| *** cloudnautique has joined #kolla | 15:51 | |
| sdake | ashishjain there ya go ^ | 15:51 |
| sdake | can you document that with a change to quickstart.rst? | 15:51 |
| sdake | good way to learn about the workflow | 15:51 |
| pbourke | currently the doc says " In the case of a single interface on a machine, a veth pair may be used where one end of the veth pair is listed here and the other end is in a bridge on the system." | 15:51 |
| sdake | doc/quckstart.rst | 15:51 |
| *** cloudnautique has joined #kolla | 15:51 | |
| sdake | pbourke ya i dont know how to do that | 15:52 |
| pbourke | that's different to https://github.com/openstack/kolla/blob/master/tests/deploy_aio.sh#L28 | 15:52 |
| sdake | a dummy interface hsould be good enough for dev i think | 15:52 |
| sdake | ash also note the config option https://github.com/openstack/kolla/blob/master/tests/deploy_aio.sh#L24 | 15:53 |
| sdake | ashishjain | 15:53 |
| ashishjain | sdake: hi | 15:53 |
| sdake | ashishjain pbourke has a ink to how to creat the dummy interface | 15:54 |
| ashishjain | reading through | 15:54 |
| *** tzn has joined #kolla | 15:54 | |
| sdake | would recommend changing the doc he pointed out above to the dumy_interface line | 15:54 |
| sdake | other then that, Id ont nkow how to setup the bridge | 15:54 |
| pbourke | ive set up a bridge with pairs like that before but cant remember off the top of my head | 15:54 |
| pbourke | try the dummy interface, Im not sure what that gives you | 15:55 |
| ashishjain | sdake : sure I will make the doc change | 15:55 |
| openstackgerrit | Merged openstack/kolla: Make RHEL build properly https://review.openstack.org/242343 | 15:55 |
| *** vilobhmm11 has joined #kolla | 15:57 | |
| pbourke | ashishjain: sdake: id prefer you dont doc the dummy interface without advice from SamYaple | 15:58 |
| sdake | i thin we get that in a review | 15:58 |
| pbourke | ashishjain: that said feel free to put up a patch and he can comment on that | 15:58 |
| sdake | right | 15:58 |
| sdake | ashishjain set workflow -1 | 15:58 |
| *** tzn has quit IRC | 15:58 | |
| ashishjain | pbourke:agreed | 15:58 |
| sdake | i dont know if sam knows how to do the bridge thing either | 15:59 |
| pbourke | ok thanks! | 15:59 |
| pbourke | bet he does | 15:59 |
| sdake | i think we all sort of made that up :) | 15:59 |
| pbourke | i would try it now but my todo list is just too long :/ | 16:00 |
| sdake | rogeer | 16:01 |
| kproskurin | Guys, just got a “Build timed out (after 60 minutes). Marking the build as failed.” in gate-kolla-build-images-centos-source | 16:01 |
| *** sdake_ has joined #kolla | 16:03 | |
| *** cemmason1 has joined #kolla | 16:04 | |
| *** cemmason has quit IRC | 16:05 | |
| *** sdake has quit IRC | 16:06 | |
| sdake_ | kproskurin got a link? | 16:07 |
| sdake_ | i'd try a recheck | 16:07 |
| sdake_ | sometimes the builds just dont work | 16:07 |
| kproskurin | http://logs.openstack.org/67/242467/2/check/gate-kolla-build-images-centos-source/709c8e2/console.html | 16:07 |
| sdake_ | hesenbug ftw | 16:07 |
| *** gfidente|lunch is now known as gfidente | 16:07 | |
| sdake_ | this is worrisome | 16:08 |
| sdake_ | 2015-11-06 15:01:23.277 | INFO:kolla.cmd.build:nova-compute:Failed to get D-Bus connection: No connection to service manager. | 16:08 |
| sdake_ | oh that is part of glusterfs | 16:09 |
| sdake_ | I think that happens all the etime | 16:09 |
| sdake_ | kproskurin not sure just try a recheck | 16:09 |
| sdake_ | it looks like things were just too slow to make the 60 minute window | 16:09 |
| kproskurin | yep | 16:09 |
| kproskurin | just so you know... | 16:09 |
| sdake_ | i knwo the gate is flakey :) | 16:10 |
| sdake_ | what I'd like to know about is a degradation in teh gate | 16:11 |
| sdake_ | right now it should be failing about 1 in 20 times | 16:11 |
| sdake_ | if it fails like 1 in 3 times ,that would be something i'd like to know about | 16:11 |
| sdake_ | rather it shouldn't be failing at all | 16:11 |
| sdake_ | but doe so | 16:11 |
| sdake_ | does so | 16:11 |
| sdake_ | we got the gate all fixed up last night | 16:11 |
| sdake_ | mandre fixed one problem yesterday and i fixed another problem the day before | 16:12 |
| sdake_ | so the budl gates should be 1005 pass rate minus wierd 1 of ffailures | 16:12 |
| *** cemmason1 is now known as cemmason | 16:14 | |
| sdake_ | folks - please take a moment to take a couple work items from https://blueprints.launchpad.net/kolla/+spec/drop-root | 16:14 |
| sdake_ | these are easy patches - mostly meant to test how well distributed work operates with work-items | 16:15 |
| britthouser | I have tried adding user to both glance and keystone. But both deploys bombed with container logs that looks like: | 16:16 |
| * britthouser tries to figure out pastebin | 16:16 | |
| sdake_ | if you on centos run yum install fpaste | 16:16 |
| sdake_ | then docker logs xyz | fpaste | 16:16 |
| sdake_ | i haven tactually tried that USER thing | 16:17 |
| sdake_ | so ymmv :) | 16:17 |
| britthouser | http://paste.fedoraproject.org/287688/44682659 | 16:17 |
| sdake_ | i'll be getting to it in about an hour after this backport is done | 16:17 |
| britthouser | Ok...Yeah I'm scratching my head. Just don't know the code well enough yet. But I guess this is my opportunity! =) | 16:17 |
| sdake_ | ok, during the docker build you need to chown /etc/keystone to the keystone user | 16:17 |
| britthouser | yeah there is this line (above my USER statement) | 16:18 |
| britthouser | chown -R keystone: /etc/keystone /var/www/cgi-bin/keystone /var/log/apache2 /home/keystone | 16:18 |
| sdake_ | so add a chown keystone: /etc/keystone to docker/keystone/Dockerifle.j2 | 16:18 |
| sdake_ | can you paste your dockerfile.j2? | 16:18 |
| britthouser | http://paste.fedoraproject.org/287691/26762144 | 16:19 |
| britthouser | Oh...I guess b/c I said use apache (since that is what is running) | 16:19 |
| sdake_ | fpaste pretty handy huh? :)O | 16:19 |
| britthouser | but /etc/keystone is owned by keystone. | 16:19 |
| britthouser | yes it is! | 16:19 |
| *** achanda has joined #kolla | 16:20 | |
| sdake_ | your USer is in the wrong place | 16:20 |
| britthouser | I had same problem with glance thought (I was just trying that b/c I thought it woudl be more straighforward) | 16:20 |
| sdake_ | put on line 54 | 16:20 |
| sdake_ | ahead of run statemnt | 16:20 |
| britthouser | oh | 16:20 |
| britthouser | ok | 16:20 |
| britthouser | lemme give that a whirl and I'll report back | 16:21 |
| sdake_ | user has to run ahead of the run operatio essentially | 16:25 |
| sdake_ | makes ure not to put user ahead of the chown operation thoguh1 | 16:25 |
| *** sdake_ is now known as sdake | 16:28 | |
| openstackgerrit | Steven Dake proposed openstack/kolla: Fix RHEL builds https://review.openstack.org/242553 | 16:31 |
| openstackgerrit | Steven Dake proposed openstack/kolla: Make RHEL build properly https://review.openstack.org/242554 | 16:31 |
| sdake | rhallisey jpeeler can you guys start looking at the backports so I can cut a 1.0.1 release today - there a re about 10 in teh queue | 16:31 |
| sdake | pbourke as weol | 16:31 |
| sdake | they are all rubber stamps | 16:32 |
| sdake | britthouser btw I have glance assigned to me, did you want ot take eownership of it? | 16:33 |
| *** vilobhmm11 has quit IRC | 16:35 | |
| britthouser | No....I was just trying it in addition to see if I could get it to work easier than keystone. | 16:35 |
| *** vilobhmm11 has joined #kolla | 16:36 | |
| sdake | cool well your free to take it over if you want another easy one :) | 16:36 |
| sdake | britthouser I think the advice I offered earlier was wrong | 16:42 |
| britthouser | Yeah....now they don't build | 16:42 |
| sdake | what is happening is the base script is being run in the base container | 16:42 |
| sdake | this sometimes calls kolla_extend.sh | 16:42 |
| sdake | we need to somehow encode the user and pass it to the bae case | 16:43 |
| sdake | base script | 16:43 |
| sdake | https://blueprints.launchpad.net/kolla/+spec/drop-root | 16:44 |
| sdake | wrong paste | 16:44 |
| sdake | https://github.com/openstack/kolla/blob/master/docker/base/Dockerfile.j2#L173 | 16:44 |
| britthouser | so we need USER to be set before that CMD, right? | 16:44 |
| sdake | i am not quite sur ewhat we need | 16:45 |
| sdake | can you past eyour error gain? | 16:45 |
| *** achanda has quit IRC | 16:45 | |
| sdake | my scorllback is gone | 16:45 |
| rhallisey | anyone dealt with an issue when you deploy and somehow you images always try to pull | 16:45 |
| sdake | i ha to restart my client | 16:45 |
| rhallisey | even with the pull policy=missing | 16:45 |
| rhallisey | and the image existing | 16:45 |
| britthouser | http://paste.fedoraproject.org/287688/44682659 | 16:45 |
| sdake | i have noticed pull policy=missing always trie to pull | 16:45 |
| rhallisey | mine is stuck | 16:45 |
| rhallisey | it fails to pull | 16:46 |
| rhallisey | so blows up | 16:46 |
| rhallisey | when I have the image locally | 16:46 |
| pbourke | rhallisey: that is a known issue | 16:46 |
| rhallisey | pbourke, is this ansible? | 16:46 |
| rhallisey | 1.9.2 | 16:46 |
| pbourke | I dont think anyone got to the bottom of it | 16:46 |
| pbourke | its with the data image right? | 16:47 |
| rhallisey | yes | 16:47 |
| rhallisey | it's driving me crazy | 16:47 |
| pbourke | yeah :/ | 16:47 |
| sdake | britthouser can you hack your base script to ls -l /var/lib/kolla/config_files/keystone.conf | 16:47 |
| * rhallisey will hack around it | 16:47 | |
| sdake | you will have to do it inside kolla_set_configs | 16:47 |
| sdake | I want to see what permissions it has | 16:47 |
| sdake | i think your having errors on read not write | 16:47 |
| sdake | although it says permission denied on /etc/keystone | 16:48 |
| sdake | maybe its actually a perission error on read | 16:48 |
| sdake | nah its definately /etc/keystone | 16:48 |
| sdake | can you ls -l /etc/keystone as well | 16:48 |
| pbourke | sdake: do you want to wait for the gate before +a on these backports | 16:48 |
| sdake | might as well | 16:49 |
| pbourke | i'll look over them again in about 30 mins | 16:49 |
| *** Hui has joined #kolla | 16:49 | |
| *** vilobhmm11 has quit IRC | 16:49 | |
| *** vilobhmm11 has joined #kolla | 16:49 | |
| Hui | hi, is the kolla master branch installing mitaka openstack binary? | 16:50 |
| sdake | should be doing so hui | 16:50 |
| rhallisey | Hui, yes | 16:50 |
| sdake | not sure about from source | 16:50 |
| Hui | thanks, sdake and rhallisey | 16:50 |
| britthouser | ok sdake so I'll edit docker/base/set_configs.py and add 'ls' in there, and the rebuild the world, right? | 16:50 |
| rhallisey | omg what is going on with this pull lol.. | 16:50 |
| sdake | just rebuild keystone | 16:51 |
| sdake | build.py flags keystone will do it | 16:51 |
| Hui | how can I tell from the kolla code which rdo will be installed? mitaka or liberty? | 16:51 |
| sdake | hui black magic in the base file | 16:51 |
| sdake | diff stable/liberty from master | 16:51 |
| rhallisey | pbourke, do I need to downgrade ansible or something | 16:52 |
| rhallisey | or comment out everything in common | 16:53 |
| rhallisey | and start them by hand | 16:53 |
| Hui | sdake, I saw this line in base file RUN yum -y install http://rdoproject.org/repos/openstack-liberty/rdo-release-liberty.rpm | 16:53 |
| pbourke | rhallisey: I dont think so - I tried all sorts of combinations of ansible and docker-py and couldnt solve it | 16:53 |
| *** jmccarthy has quit IRC | 16:53 | |
| pbourke | rhallisey: in the end ended up pushing to a local registry and just using that | 16:53 |
| Hui | sdake, is this the magic? | 16:53 |
| rhallisey | :/ | 16:53 |
| sdake | that it | 16:53 |
| *** vilobhmm11 has quit IRC | 16:54 | |
| *** absubram has quit IRC | 16:55 | |
| sdake | hui thats it | 16:55 |
| Hui | sdake, thanks. | 16:55 |
| *** achanda has joined #kolla | 16:56 | |
| sdake | boy regv2 pile of shit - 90 minute push times totally unacceptable | 17:04 |
| *** unicell1 has quit IRC | 17:04 | |
| *** unicell has joined #kolla | 17:05 | |
| *** kproskurin has quit IRC | 17:08 | |
| diogogmt | i just noticed after enabling debug logs that almost half of the services don’t have a source url configured in the kola-build.conf file — is that because their sources are not being hosted in the openstack http://tarballs.openstack.org repo or something else? | 17:09 |
| sdake | diogogmt only base imgefs have uris associated | 17:10 |
| sdake | the other imges have info but not uris | 17:10 |
| diogogmt | sdake: so the other images are built from binaries instead? | 17:10 |
| sdake | provide example | 17:10 |
| *** mfalatic has quit IRC | 17:11 | |
| diogogmt | sdake: starting on line 94: http://pastebin.com/1ivKeBpW | 17:11 |
| diogogmt | it says some services don’t have a source url | 17:11 |
| diogogmt | but since i set the build flag to build from source how does it build if it can’t find the source file? | 17:12 |
| sdake | exampl enova-scheduler | 17:12 |
| sdake | has a nova-base | 17:12 |
| sdake | nova-base contains the uri | 17:12 |
| sdake | nova-sscheduler depends on nova-base | 17:12 |
| sdake | nova-base has the uri | 17:12 |
| sdake | the source file is inherited from the nova-base image in dockerfiles | 17:13 |
| sdake | the log notices are confusing - just ignore them | 17:13 |
| sdake | or file a wishlist bug and maybe someday we will clean that up a bit | 17:13 |
| diogogmt | sdake: kk, so if i understand correctly the nova-scheduler image will inherit from the nova-base image, right? | 17:14 |
| sdake | roger | 17:14 |
| diogogmt | and the nova-scheduler source is being referened by the nova-base image? | 17:14 |
| sdake | nova-scheduler is part of the nova tarballs | 17:14 |
| sdake | ec | 17:14 |
| diogogmt | oh ok, makes sense | 17:14 |
| sdake | https://github.com/openstack/kolla/blob/master/etc/kolla/kolla-build.conf#L122 | 17:15 |
| sdake | that file sets the uri | 17:15 |
| sdake | yuo see there is no nova-scheduler | 17:15 |
| sdake | https://github.com/openstack/kolla/blob/master/docker/nova/nova-scheduler/Dockerfile.j2#L1 | 17:15 |
| sdake | this line inherits from nova-base | 17:15 |
| sdake | https://github.com/openstack/kolla/blob/master/docker/nova/nova-base/Dockerfile.j2#L37 | 17:16 |
| sdake | this line is what installs the source tarball in nova-base | 17:16 |
| *** ssurana has joined #kolla | 17:16 | |
| *** daneyon has joined #kolla | 17:18 | |
| diogogmt | ok, got it — as far as the image building orchestration, is there a particular order the images are built, where in the code is that order defined? | 17:18 |
| dmsimard | Hui: the master branch installs trunk RDO (Mitaka), the stable liberty branch installs liberty RDO as of https://review.openstack.org/#/c/242229/ | 17:18 |
| Hui | dmsimard, thanks | 17:19 |
| Hui | the neutron server on the master branch keeps failing today | 17:19 |
| dmsimard | Hui: actually I might have sort of lied, hang on | 17:20 |
| Hui | ok | 17:20 |
| dmsimard | ok I lied | 17:22 |
| dmsimard | sdake: This bit is bad for the time being https://github.com/openstack/kolla/blob/master/docker/base/Dockerfile.j2#L52-L53 | 17:22 |
| sdake | what do you mean time being | 17:22 |
| dmsimard | We don't have proper CI in place for Mitaka yet so we don't actually promote packages right now. The current-passed-ci repo was the last tested repo before we cut liberty | 17:23 |
| dmsimard | Which was 09-28 | 17:23 |
| sdake | so this isn't master? | 17:23 |
| sdake | we want master in theere, preferrably a ci tested one but if not - master will do | 17:23 |
| dmsimard | We really only have one master, which is http://trunk.rdoproject.org/centos7/current/ - but, when CI is in place, we promote tested builds to current-passed-ci | 17:24 |
| dmsimard | We do not encourage using current as it can and will break (untested right now) | 17:24 |
| sdake | we expect our master to break | 17:24 |
| sdake | we are good wit hthat | 17:24 |
| sdake | can you submit a change to fix it to master? | 17:24 |
| dmsimard | By untested I mean we still go through unit/package/delorean tests but there is no integration tests so to speak so no tempest, installation, etc. | 17:24 |
| sdake | I've been around and aroudn bout the proper upstream repo to use | 17:24 |
| dmsimard | Yeah, and I'll point it back to the CI repo once we have the CI in place - should be two weeks or so | 17:25 |
| sdake | sounds good | 17:25 |
| dmsimard | Do you want a bug ? | 17:25 |
| dmsimard | or a good commit message is enough | 17:25 |
| sdake | Use "TrivialFix" flag | 17:25 |
| sdake | and good commit message | 17:25 |
| Hui | sdake, dmsimard, so you think using stable/liberty is more stable than master? | 17:26 |
| sdake | hui absolutely yes | 17:26 |
| dmsimard | sdake: hm, how do I set that flag ? | 17:26 |
| sdake | if your actually deploying please use stable/liberty | 17:26 |
| sdake | dmsimard add to the commit log TrivialFix | 17:26 |
| sdake | it just goes in the commit log - git log | grep TrivailFix for examples | 17:26 |
| dmsimard | sdake: like, below the Id ? | 17:26 |
| sdake | that way we aren't expecting a bu gid | 17:26 |
| sdake | ahedof change id | 17:27 |
| sdake | the reason we have a flag at all is we are not mind readers if the author thinks its a trivial fix or not | 17:27 |
| sdake | and sometimes it not trivial to the reviewer :) | 17:27 |
| dmsimard | ok got it | 17:27 |
| sdake | so to clarify we have the trivialfix flag which can always be used in palce a bug, but may notbe accepted by the reviewer at their preference | 17:27 |
| sdake | i haven't seen a trivialfix flag been rejected so far | 17:27 |
| sdake | people sem to be using it for the right things \o/ | 17:28 |
| sdake | pbourke do you know where hte continers are stored in the registery v1 code bsae in the registryv1 container? | 17:28 |
| sdake | or rhallisey or jpeeler ? | 17:28 |
| sdake | it isn't /var/lib/registry | 17:29 |
| openstackgerrit | David Moreau Simard proposed openstack/kolla: Switch to RDO untested master repository https://review.openstack.org/242579 | 17:30 |
| sdake | real26m32.687s | 17:30 |
| *** tzn has joined #kolla | 17:30 | |
| sdake | docker v1 push time - 89 containers | 17:30 |
| diogogmt | sdake: do you know who is the maintainer for the Vagrant file? can i ask you a question about it? | 17:31 |
| dmsimard | Hui: with that commit in, it'll really use the up-to-date master repositories ^ | 17:31 |
| dmsimard | Hui: but like sdake said, you probably really want to use the stable/liberty branch, especially as we're early in the mitaka cycle | 17:31 |
| Hui | which commit? sorry I miss some message, dmsimard | 17:32 |
| dmsimard | Hui: https://review.openstack.org/242579 | 17:32 |
| sdake | diogogmt hit up mandre or pbourke | 17:32 |
| jpeeler | dmsimard: when would CI be added for mitaka packages? | 17:32 |
| diogogmt | ok | 17:32 |
| Hui | dmsimard, i see. thanks | 17:33 |
| dmsimard | jpeeler: I'm hoping within the next 2 weeks | 17:33 |
| rhallisey | sdake, I don't think rhel is 100% working yet | 17:33 |
| sdake | well it builds properly ;-) | 17:33 |
| rhallisey | nova-compute image doesn't install the nova package atm | 17:33 |
| dmsimard | jpeeler: we need to finish some liberty CI refactor/migration but once that is done, Mitaka CI is largely a copy/paste | 17:33 |
| sdake | it worked when i tessted it with rhos a bit some time ago | 17:33 |
| rhallisey | eh ok | 17:33 |
| rhallisey | I will re try it | 17:34 |
| sdake | rhallisey eel free to please fix it | 17:34 |
| rhallisey | what flags did you use | 17:34 |
| rhallisey | rhos rhel | 17:34 |
| sdake | tools/build.py -b rhel -t rhos | 17:34 |
| rhallisey | kk | 17:34 |
| sdake | using master | 17:34 |
| sdake | my patch just hit the repo | 17:34 |
| sdake | which fixes that | 17:34 |
| sdake | if you don't have rhos license use -t rdo | 17:34 |
| sdake | we want rhos/rdo/source to work on rhel | 17:35 |
| rhallisey | sdake, ya it may just be rhos nova-compute | 17:39 |
| rhallisey | will look into it | 17:39 |
| sdake | what iw ould do is run the built container | 17:39 |
| sdake | docker run -it rhel-rhos-nova-compute bash | 17:39 |
| sdake | and have al ook around | 17:39 |
| rhallisey | it never installs nova | 17:39 |
| rhallisey | ya will do a yum in there | 17:39 |
| rhallisey | see what happens | 17:39 |
| sdake | what i mean by have a look around is try yum -y install nova-compute | 17:40 |
| *** mbound has quit IRC | 17:40 | |
| *** unicell has quit IRC | 17:42 | |
| rhallisey | 332 deps O.o | 17:44 |
| sdake | ya its a beast | 17:44 |
| sdake | surprised it works tbh ;-) | 17:44 |
| sdake | most of those deps are unneeded in a container deployment | 17:44 |
| sdake | but that is the beuty of containers - who cares ;) | 17:44 |
| *** mfalatic has joined #kolla | 17:44 | |
| sdake | hey mfalatic | 17:45 |
| rhallisey | [Errno -1] Package does not match intended download. Suggestion: run yum --enablerepo=rhel-7-server-htb-rpms | 17:45 |
| rhallisey | clean metadata | 17:45 |
| rhallisey | ya this is where I left off | 17:45 |
| sdake | wtf is htb-rpms? | 17:45 |
| rhallisey | I have no idea | 17:45 |
| sdake | this is what i get | 17:46 |
| sdake | INFO:__main__:openstack-base:https://cdn.redhat.com/content/dist/rhel/server/7/7Server/x86_64/os/repodata/productid: [Errno -1] Metadata file does not match checksum | 17:46 |
| sdake | INFO:__main__:openstack-base: | 17:46 |
| sdake | INFO:__main__:openstack-base:Trying other mirror. | 17:46 |
| sdake | INFO:__main__:openstack-base: | 17:46 |
| sdake | INFO:__main__:openstack-base: | 17:46 |
| rhallisey | ya there are a few of them | 17:47 |
| sdake | hui master works for me | 17:49 |
| sdake | STDAKE-M-J2VL:demo sdake$ nova list | grep ACTIVE | wc -l | 17:49 |
| sdake | 12 | 17:49 |
| sdake | init-runonce follows by demos/heat/launch | 17:49 |
| sdake | hui just built about 30 mins ago | 17:50 |
| sdake | and deployed in 3 node ha config | 17:50 |
| sdake | rhallisey i just built rhel + rhos and got following image name - perhaps this is the problem | 17:52 |
| sdake | kollaglue/rhel-binary-nova-compute latest 74329d59e709 2 minutes ago 984 MB | 17:52 |
| sdake | ile or directory | 17:52 |
| sdake | [root@08e54900f685 /]# rpm -qa | grep nova-compute | 17:52 |
| sdake | openstack-nova-compute-2015.1.1-3.el7ost.noarch | 17:52 |
| *** vilobhmm11 has joined #kolla | 17:54 | |
| Hui | sdake, (_mysql_exceptions.ProgrammingError) (1146, "Table 'neutron.ml2_vlan_allocations' doesn't exist") | 17:54 |
| Hui | sdake, I got this error from neutron server master | 17:54 |
| Hui | branch | 17:54 |
| sdake | Hui did you clean up your local environment and delete your database container data container? | 17:55 |
| sdake | we dont yet implement database migratiosn for update | 17:55 |
| Hui | I did clean-containers | 17:55 |
| *** vilobhmm11 has left #kolla | 17:55 | |
| Hui | sdake,do I need to remove /etc/kolla on the deploy node? | 17:55 |
| sdake | nope shouldn't need to unless yoru using ceph | 17:55 |
| sdake | then the answer is YES | 17:56 |
| Hui | I mean /etc/kolla/neutron-server, I do not use ceph | 17:56 |
| *** kproskurin has joined #kolla | 17:56 | |
| *** ashishjain has quit IRC | 17:59 | |
| sdake | hui shoouldn't need to do so | 18:00 |
| sdake | the problem you have is your database is not migrating | 18:01 |
| sdake | i may be experirinecing that exception and not seeing it | 18:01 |
| sdake | how do yo duplicate? | 18:01 |
| rhallisey | sdake, I think my subscription repos are just messed up | 18:02 |
| rhallisey | since rdo isn't working .. | 18:02 |
| sdake | make sure to use virtial repo pool id | 18:02 |
| sdake | not physical | 18:02 |
| sdake | virtual are unlimited pool subscriptions physical has imit | 18:02 |
| sdake | you don't want a physical in the build | 18:02 |
| sdake | that should probably go in the build docs | 18:02 |
| Hui | sdake, there is no existing database. I just install kolla from scratch today and neutron fails | 18:03 |
| sdake | hui are you doihng dev or production deployo/eval? | 18:04 |
| openstackgerrit | Merged openstack/kolla: Fix RHEL builds https://review.openstack.org/242553 | 18:04 |
| rhallisey | sdake, uh how do I do virtual | 18:04 |
| rhallisey | only see physical pool | 18:04 |
| sdake | ask the bossses for one | 18:04 |
| sdake | i asked my boss britthouser in this case ;-) | 18:05 |
| sdake | its just a separate pool id | 18:05 |
| sdake | need to hunt one down | 18:05 |
| Hui | sdake, I have a dev testbed of 4 nodes to build kolla everyday | 18:05 |
| sdake | hui nice! | 18:05 |
| openstackgerrit | Merged openstack/kolla: Make RHEL build properly https://review.openstack.org/242554 | 18:06 |
| Hui | if a build success, I upgrade the images in a larger testbed | 18:06 |
| jpeeler | rhallisey: mine has virtual if you need to use it again | 18:06 |
| sdake | nice canary deploooyment -that would be a huge killer feture | 18:06 |
| Hui | sdake, now as you pointed out, I may consider pin to stable/liberty for the stable reason | 18:06 |
| rhallisey | jpeeler, how do I get to it | 18:07 |
| rhallisey | I'm using yours | 18:07 |
| sdake | ya ounds more like your doing eval use csae hui | 18:07 |
| sdake | rather then dev | 18:07 |
| jpeeler | i always have to relook it up because i use it like once a year... | 18:07 |
| Hui | sdake, yes | 18:07 |
| sdake | hui what is your max scale size your gunning for? | 18:09 |
| Hui | 5 compute nodes + 2 network nodes + 1 controller as of now | 18:10 |
| sdake | hui 'd recommend trying out scale teting of ha if that is in your future :) | 18:11 |
| Hui | sdake, of course | 18:11 |
| Hui | sdake, is there any official doc about kolla HA? | 18:11 |
| sdake | just the spec but its dated | 18:11 |
| sdake | most fo the good info is floating about in devs brains | 18:11 |
| *** achanda has quit IRC | 18:12 | |
| Hui | ok | 18:12 |
| sdake | we need an architecture.rst document | 18:12 |
| *** unicell has joined #kolla | 18:12 | |
| britthouser | rhallisey: if you are on baremetal, the virtual pools won't show up | 18:13 |
| britthouser | but if you are in a VM, you should see both virtual and physical pools. | 18:13 |
| britthouser | make sure the pick the virtual pool. | 18:13 |
| rhallisey | is it possible to attach to a virtual pool on baremetal | 18:16 |
| rhallisey | not sure if it was accpeted.. | 18:16 |
| rhallisey | pool is restricted to virtual guests | 18:17 |
| britthouser | I think subscription manager is smart enough not to let you do that. =) | 18:18 |
| rhallisey | :'( | 18:18 |
| britthouser | Or else your paycheck would be very very small. =P | 18:18 |
| rhallisey | ya | 18:19 |
| *** cemmason has quit IRC | 18:22 | |
| sdake | britthouser see this http://paste.fedoraproject.org/287756/44683451 | 18:29 |
| sdake | http://paste.fedoraproject.org/287757/83456214 | 18:30 |
| britthouser | Yeah...that is what keystone container looks liek today | 18:30 |
| sdake | nm aparently i didn't build keystone correctly | 18:30 |
| sdake | i had added user in but didn't build it propery | 18:31 |
| britthouser | Ok. | 18:31 |
| britthouser | I guess what we're expecting to see is that the inital HTTPd and the logging processes owned by apache? | 18:31 |
| sdake | ya although i'm not sure that is viable | 18:31 |
| sdake | it may work if we change permissiosn on /dev/log or whatever it s called | 18:32 |
| britthouser | *seems* like we're mostly doing the right thing here alrady. | 18:32 |
| *** absubram has joined #kolla | 18:32 | |
| britthouser | but I'm not sure what the goal is. 100% or where its just flagrantly wrong. | 18:33 |
| *** achanda has joined #kolla | 18:34 | |
| *** achanda has quit IRC | 18:35 | |
| sdake | britthouser here is what I get with user at the end | 18:41 |
| sdake | http://paste.fedoraproject.org/287765/14468352 | 18:41 |
| sdake | did you ge tthat ls jammed in there? | 18:41 |
| severion | hello everyone. have a question regarding the kolla heat template for testing kolla. stack is in the creation process right now, but i noticed the images in the "stack view" details are broken. they seem to work with other stack templates i have running (like magnum). is this a kolla thing or do i need to reach out to the heat team about this? | 18:43 |
| severion | haven't really ever seen this before...so new to me | 18:43 |
| sdake | severion just a heads up, the heat template is currently totally and completely broken | 18:43 |
| severion | lol | 18:44 |
| sdake | severion someone needs to moidfy it to match ansible vs compose where it is at now | 18:44 |
| severion | well, man... | 18:44 |
| sdake | i'd recommend bare metal personally | 18:44 |
| sdake | but alternatively vagrant works | 18:44 |
| severion | crap...my test environment is all OS at the moment | 18:44 |
| britthouser | not yet sdake....getting pulled into other things this afternoon. | 18:44 |
| severion | i tried vagrant...but ran into issues and couldn't find answers on that one either. | 18:44 |
| sdake | you can use os - just treat the vm like bare metal | 18:44 |
| sdake | roger | 18:44 |
| sdake | severion I highly doubt you can use vagrant on top of openstack | 18:46 |
| severion | no | 18:46 |
| sdake | and if you do it would be painfully slow sine it would be virt in virt | 18:46 |
| severion | the vagrant option i was building in virtualbox | 18:46 |
| severion | on mac | 18:46 |
| sdake | roger | 18:46 |
| severion | and yes...it was slow anyway :D | 18:46 |
| sdake | ya that is tessted and works | 18:46 |
| severion | very | 18:46 |
| sdake | pbourke and mandre are points of contact there | 18:46 |
| dmsimard | sdake: you can with an openstack provider | 18:46 |
| sdake | i'd recommend just spinning up a 16 gig vm and treating it like bare metal | 18:47 |
| dmsimard | sdake: say you use openstack instead of virtualbox | 18:47 |
| severion | so my issues...hmmm...let me look back | 18:47 |
| severion | never got past this: "unable to connect, check login credentials (login_user, and login_password, which can be defined in ~/.my.cnf), | 18:47 |
| severion | under creating keystone | 18:47 |
| sdake | and following the quickstart guide there | 18:47 |
| sdake | which version ar eyou using? | 18:47 |
| sdake | i just deployed master 1 hour ago | 18:47 |
| severion | grabbed github a couple of days ago. | 18:47 |
| severion | did that fix some issues? | 18:48 |
| severion | meaning, your resent push to master... | 18:48 |
| sdake | we have 4-6 patches hitting the repo every day | 18:48 |
| sdake | i'd recommend a fresh pull | 18:48 |
| severion | oye | 18:48 |
| severion | ok | 18:48 |
| severion | i'll try again | 18:48 |
| *** achanda has joined #kolla | 18:48 | |
| severion | thanks for the tip. i'll try again...follow instructions exactly...and then report back. | 18:49 |
| sdake | severion if your doing an eval, checkout stable/liberty | 18:50 |
| severion | if that doesn't work, i'll do the OS bare metal attempt. will probably run faster that way anyway. | 18:50 |
| sdake | it is more stable then master ;-) | 18:50 |
| sdake | clone the repo then git checkout stable/liberty and use that | 18:50 |
| severion | ah, good call. is magnum included or does that need to be baked in after turn-up? | 18:50 |
| sdake | we have not implmeneted ansible playbooks for magnum - it is in the review queue atm as WIP | 18:51 |
| severion | the inception since containers is wild...container kolla, running caas magnum! haha | 18:51 |
| sdake | i have reviewed it - code looks ready for mergef imo | 18:51 |
| severion | as is magnum in general anyway...understandable entirely | 18:51 |
| severion | last time i checked (maybe last week) kubernetes started working, mesos had additional requirements and swarm...well, that was just plain broken from what i could tell. | 18:52 |
| *** absubram has quit IRC | 18:53 | |
| Hui | sdake, I found why my neutron fails. The bootstrap node fails due to empty /etc/neutron/neutron.conf in the container | 19:05 |
| *** gfidente has quit IRC | 19:05 | |
| sdake | idea why that file is empty? | 19:05 |
| Hui | volumes: "{{ node_config_directory }}/glance-api/:{{ container_config_directory }}/:ro", container_config_directory /var/lib/kolla/config_files/neutron.conf | 19:06 |
| Hui | do you know which line of code copy /var/lib/kolla/config_files/neutron.conf to /etc/neutron/ in the boostrap container? | 19:07 |
| sdake | set_configs.py does the job | 19:07 |
| sdake | it is based upon config.json | 19:07 |
| sdake | it reads config.json and does its thing | 19:07 |
| sdake | are you running ansible as sudo? | 19:07 |
| sdake | (you should be imo:) | 19:07 |
| rhallisey | Hui, wasn't this fixed in sam's patch? | 19:08 |
| sdake | oh right | 19:08 |
| Hui | no | 19:08 |
| sdake | hui how old is your repo? | 19:08 |
| Hui | updated to the latest HEAD | 19:08 |
| sdake | are you running as sudo? | 19:09 |
| sdake | I suspect you hae a permissions problem on one of the files sansible is using | 19:09 |
| Hui | I am running as a root | 19:09 |
| sdake | out of disk space? | 19:10 |
| sdake | so I am running as root, but I didn't clean up my confi dirs and whatnot | 19:10 |
| *** mbound has joined #kolla | 19:10 | |
| sdake | so its possible master is broken | 19:10 |
| dmsimard | Hey so I have a kind of a stupid question, how do you see Kolla different than OSAD ? From an untrained eye they overlap a lot. Using very similar means to get things running (ansible, containers). Perhaps the main difference I see is that they do not provide support for binary installs. | 19:13 |
| sdake | dms moment | 19:13 |
| * britthouser gets some popcorn | 19:13 | |
| Hui | sdake, I saw the neutron-server.json file. It should copy the file in the container | 19:14 |
| rhallisey | dmsimard, they use container like vms | 19:14 |
| rhallisey | and they don't containerize everything | 19:14 |
| rhallisey | it's a completely different model | 19:14 |
| rhallisey | we use the micro service approach | 19:14 |
| dmsimard | Sorry for the (somewhat on purpose) trick question :) I just want to see where you guys are coming from and ensure there's no wheel re-invention going on | 19:15 |
| sdake | hang tight | 19:15 |
| sdake | dmsimard https://review.openstack.org/#/c/206789/ | 19:16 |
| rhallisey | dmsimard, this is a completely different methodology. It's a fair question that we had to pass to get into the big tent | 19:16 |
| sdake | dmsimard see my comment on patch set 5 | 19:17 |
| sdake | britthouser enjoy the salt-free popcorn ;-) | 19:17 |
| sdake | the tc had a similar question - how is it different | 19:17 |
| sdake | that comment answers that quesstion | 19:18 |
| sdake | there are a slew more differences | 19:18 |
| sdake | but those are the big ones | 19:18 |
| sdake | and to be fair, kolla started first on the concept ;-) | 19:18 |
| sdake | so if anyone is wheel reinventing its osad ;) | 19:18 |
| rhallisey | hell ya | 19:18 |
| dmsimard | sdake: I'm happy with your explanation in the patchset, you should keep that link handy for people who ask that :) | 19:21 |
| sdake | its more fun to go see all the drama in my inbox re that topic ;) | 19:22 |
| *** achanda has quit IRC | 19:24 | |
| *** achanda has joined #kolla | 19:29 | |
| *** openstackstatus has joined #kolla | 19:34 | |
| *** ChanServ sets mode: +v openstackstatus | 19:34 | |
| -openstackstatus- NOTICE: Gerrit will be offline at 20:00-20:15 UTC today (starting 20 minutes from now) for scheduled project rename maintenance | 19:39 | |
| openstackgerrit | Michal Rostecki proposed openstack/kolla: Add Ansible support for Magnum https://review.openstack.org/236223 | 19:47 |
| dmsimard | oh yeah support for magnum in kolla, we've gone full circle, now let's go full triple-o inception | 19:48 |
| * dmsimard runs away | 19:49 | |
| *** cloudnau_ has joined #kolla | 19:50 | |
| nihilifer | no, before that we have to go in deeper inception | 19:50 |
| nihilifer | Kolla which runs LXD instances on which Magnum runs containers | 19:51 |
| nihilifer | that's the way to go | 19:51 |
| *** cloudnautique has quit IRC | 19:53 | |
| *** tzn has quit IRC | 19:53 | |
| *** tzn has joined #kolla | 19:59 | |
| -openstackstatus- NOTICE: Gerrit is offline until 20:15 UTC today for scheduled project rename maintenance | 20:00 | |
| *** ChanServ changes topic to "Gerrit is offline until 20:15 UTC today for scheduled project rename maintenance" | 20:00 | |
| *** tzn has quit IRC | 20:04 | |
| *** tzn has joined #kolla | 20:06 | |
| *** tzn has quit IRC | 20:11 | |
| *** tzn has joined #kolla | 20:11 | |
| *** tzn has quit IRC | 20:12 | |
| *** tzn has joined #kolla | 20:12 | |
| *** achanda has quit IRC | 20:15 | |
| *** nihilifer has quit IRC | 20:23 | |
| *** nihilifer has joined #kolla | 20:25 | |
| *** ChanServ changes topic to "Kolla IRC meetings on Wednesday 16:30 UTC (Day light savings doesn't apply) - Agenda @ https://wiki.openstack.org/wiki/Meetings/Kolla - IRC channel is *LOGGED* @ http://eavesdrop.openstack.org/irclogs/%23kolla/" | 20:35 | |
| *** dims is now known as dimsum__ | 20:37 | |
| openstackgerrit | Merged openstack/kolla: Switch to RDO untested master repository https://review.openstack.org/242579 | 20:58 |
| *** openstackgerrit has quit IRC | 21:01 | |
| *** openstackgerrit has joined #kolla | 21:01 | |
| sdake | britthouser ive got glance rolling once i post patche you can bse keystone off hat | 21:04 |
| *** absubram has joined #kolla | 21:13 | |
| *** kproskurin has quit IRC | 21:25 | |
| *** kproskurin has joined #kolla | 21:25 | |
| britthouser | Thanks sdake! | 21:37 |
| britthouser | I needed someone to blaze that trail for me. =) | 21:37 |
| SamYaple | sdake: keystone is probably going to be a speical case | 21:38 |
| SamYaple | recommend all things other than keystone first | 21:39 |
| SamYaple | and acutally its going to be tricky for all of them... | 21:39 |
| SamYaple | like nova nad neutron need special sudo files | 21:39 |
| sdake | yup | 21:39 |
| SamYaple | theyll still have some commands they need to run as root | 21:39 |
| sdake | the infra is down so i can't post my patches | 21:39 |
| sdake | i hve to run to pick u pdaugther bbiaf | 21:39 |
| SamYaple | im going back to sleep | 21:40 |
| SamYaple | dmsimard: OSAD isn't a thing! It's OSA now. Maybe they'll have another rename WHO CAN KNOW THESE THINGS??!! | 21:41 |
| dmsimard | SamYaple: it used to be called oasd before big tent | 21:48 |
| dmsimard | I guess it kind of stuck | 21:48 |
| dmsimard | :) | 21:48 |
| *** jasonsb has joined #kolla | 21:57 | |
| openstackgerrit | Merged openstack/kolla: Finish implementation of spice https://review.openstack.org/242320 | 21:57 |
| SamYaple | dmsimard: i know i supported it for over a year | 21:58 |
| SamYaple | dmsimard: many name changes | 21:58 |
| dmsimard | :) | 21:59 |
| dmsimard | Lots of good things in there, if there's one thing I was surprised about is how rackspace managed the Swift rings | 21:59 |
| dmsimard | cool and clever | 22:00 |
| SamYaple | dmsimard: having to declare every disk on every node in a yaml file is not clever in my book | 22:01 |
| SamYaple | pain in the ass, sure. | 22:01 |
| SamYaple | and it was. | 22:01 |
| *** blahRus has quit IRC | 22:02 | |
| dmsimard | SamYaple: huh ? Last I checked didn't they iterate over disks with ansible, generate a config file based on that (csv?) and then actually parse that file and build the ring files with a python script ? | 22:02 |
| dmsimard | memory is fuzzy and friday laziness is kicking in | 22:02 |
| SamYaple | dmsimard: they do... but the operator had to first declare those disks in a big blob of yaml | 22:03 |
| sdake | vs a label, sounds painful | 22:04 |
| sdake | i hope pbourke adds labeing for swift | 22:04 |
| sdake | it would be a sweet feature | 22:04 |
| SamYaple | dmsimard: https://github.com/openstack/openstack-ansible/blob/master/doc/source/install-guide/configure-swift-devices.rst | 22:05 |
| *** jasonsb has quit IRC | 22:05 | |
| SamYaple | pretty awful in my opinion (and my experince....) | 22:05 |
| SamYaple | and dmsimard they also require setting up the ring settings in a yaml file | 22:07 |
| SamYaple | all of them | 22:07 |
| SamYaple | https://github.com/openstack/openstack-ansible/blob/master/doc/source/install-guide/configure-swift-config.rst | 22:08 |
| dmsimard | oh man from my perspective that was all automated | 22:08 |
| SamYaple | none of it is | 22:08 |
| dmsimard | I didn't read the fine print :p | 22:08 |
| SamYaple | like none of it | 22:08 |
| SamYaple | it was always a pain | 22:08 |
| *** achanda has joined #kolla | 22:08 | |
| SamYaple | dmsimard: have you seen my ceph implementation for Kolla? | 22:09 |
| sdake | glance works | 22:10 |
| sdake | yay | 22:10 |
| sdake | going to eat for a bit - bbiaf | 22:10 |
| *** daneyon has quit IRC | 22:10 | |
| dmsimard | SamYaple: I haven't dug into how each project has been implemented (yet) but I am a core for puppet-ceph fwiw so I can have a look eventually :) | 22:11 |
| SamYaple | oh youll like it i think | 22:11 |
| SamYaple | completely automated | 22:11 |
| SamYaple | operator sets a marker in the form of a disk label on the disks they want to bootstrap | 22:12 |
| SamYaple | thats it | 22:12 |
| SamYaple | automated cache tier setup | 22:12 |
| SamYaple | erasure pool | 22:12 |
| dmsimard | well to be fair setting a simple cluster automatically isn't so hard, it's about managing the crush map where it gets tricky (e.g, cache pool, erasure coding and their settings) | 22:12 |
| SamYaple | ^^ | 22:12 |
| SamYaple | see above | 22:12 |
| dmsimard | oh i'll have to look then :) | 22:12 |
| sdake | ya ew got al that automated with tags | 22:12 |
| SamYaple | :D | 22:12 |
| sdake | rather labels | 22:12 |
| sdake | it rocks | 22:12 |
| sdake | it would be better if I could share a disk | 22:12 |
| * sdake flogs SamYaple a bit get to work! | 22:13 | |
| SamYaple | yea its got a few limitations dmsimard, im not saying its perfect | 22:13 |
| SamYaple | but it does do erasure and automated cache tiers | 22:13 |
| SamYaple | i deploy exclusively with erasure and cahce tiers | 22:13 |
| dmsimard | does it spin one container for each osd or what ? | 22:13 |
| SamYaple | yup | 22:13 |
| sdake | the ceph code in kolla is relaly the gem of koll aimo | 22:13 |
| sdake | that is something nobody else could possibly hope to hae ;-) | 22:14 |
| SamYaple | its also basically stand alone | 22:14 |
| sdake | anyway my mortal form grows weak | 22:14 |
| dmsimard | no kidding, i'm still so jetlagged | 22:15 |
| SamYaple | me too i should be sleep | 22:16 |
| SamYaple | anyway, i would love to have some additional ceph verification and code support dmsimard | 22:16 |
| SamYaple | this has been basically 100% code from me | 22:17 |
| SamYaple | (its from an older project of mine) | 22:17 |
| dmsimard | SamYaple: historically there was enovance's puppet-ceph module https://github.com/redhat-cip/puppet-ceph and we started a new module from scratch just to ensure we wouldn't use exported resources - just that was kind of challenging | 22:18 |
| dmsimard | I'll look at the stuff in kolla when I have a chance :) | 22:19 |
| SamYaple | yea but ceph in docker is.... harder :) | 22:19 |
| SamYaple | i was the first that I know of to get seperate osds to work with ipv4 in docker containers | 22:20 |
| SamYaple | theres some interprocess communication that requires the processes to be in the same pid namespace to not break all the things | 22:20 |
| dmsimard | Oooh wait, didn't you work with Sebastien Han for that | 22:24 |
| dmsimard | I remember asking you guys "Why?" on Twitter or something :p | 22:25 |
| SamYaple | yea thats leseb | 22:27 |
| SamYaple | i had a project going in parallel | 22:27 |
| *** achanda has quit IRC | 22:27 | |
| SamYaple | we do mix company and share ideas, but ive always built it targeting openstack while he builds stuff thats meant to be general ceph | 22:28 |
| SamYaple | to be honest though, i really hate teh ceph-dockerstuff | 22:28 |
| SamYaple | (the other ceph in docker containers project) | 22:28 |
| SamYaple | its pretty bad in my opinion | 22:28 |
| *** alisonh has quit IRC | 22:32 | |
| rhallisey | I hate it too | 22:38 |
| *** kproskurin has quit IRC | 22:38 | |
| rhallisey | their container solution wasn't the best | 22:38 |
| *** alisonh has joined #kolla | 22:50 | |
| *** achanda has joined #kolla | 23:08 | |
| *** achanda has quit IRC | 23:08 | |
| *** Hui has quit IRC | 23:24 | |
| *** achanda has joined #kolla | 23:29 | |
| kfox1111 | how far out is 1.1? | 23:32 |
| rhallisey | kfox1111, I think steak id building | 23:32 |
| rhallisey | if* | 23:33 |
| rhallisey | is* | 23:33 |
| rhallisey | :/ | 23:33 |
| kfox1111 | steak? | 23:33 |
| rhallisey | sdake | 23:33 |
| kfox1111 | ah. | 23:33 |
| rhallisey | or stake | 23:34 |
| rhallisey | I'm just hungry atm | 23:34 |
| kfox1111 | :) | 23:34 |
| kfox1111 | for ha, why was keepalived choosen over pacemaker? | 23:37 |
| *** achanda has quit IRC | 23:37 | |
| *** dimsum__ has quit IRC | 23:52 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!
