| *** Mudpuppy_ has joined #openstack-ansible | 00:00 | |
| *** mahito has joined #openstack-ansible | 00:04 | |
| *** Mudpuppy has quit IRC | 00:04 | |
| *** mahito has quit IRC | 00:08 | |
| *** mahito has joined #openstack-ansible | 00:10 | |
| *** daneyon_ has joined #openstack-ansible | 00:15 | |
| *** daneyon has quit IRC | 00:18 | |
| *** daneyon_ has quit IRC | 00:21 | |
| *** mahito has quit IRC | 00:28 | |
| *** dkalleg has quit IRC | 00:45 | |
| openstackgerrit | Merged stackforge/os-ansible-deployment: Perform sort when creating hostnames_ip file https://review.openstack.org/186059 | 00:47 |
|---|---|---|
| *** jimi|ansible has joined #openstack-ansible | 00:48 | |
| *** javeriak has quit IRC | 00:53 | |
| openstackgerrit | Merged stackforge/os-ansible-deployment: Avoid keeping re-downloading container caches https://review.openstack.org/185911 | 01:19 |
| cloudnull | Sigmavirus24 that's awesome! | 01:21 |
| cloudnull | What was the secrete little bit? | 01:21 |
| cloudnull | Nvrmd read #ansible | 01:24 |
| cloudnull | Which is good stuff | 01:24 |
| cloudnull | =) | 01:24 |
| *** mahito has joined #openstack-ansible | 01:26 | |
| *** sdake_ has quit IRC | 01:28 | |
| *** markvoelker has quit IRC | 01:30 | |
| *** markvoelker_ has joined #openstack-ansible | 01:30 | |
| sigmavirus24 | Yeah. I gave up at the end and just asked | 01:37 |
| *** dkalleg has joined #openstack-ansible | 01:45 | |
| sigmavirus24 | odyssey4me: go to sleep | 01:49 |
| *** dkalleg has quit IRC | 01:50 | |
| *** sigmavirus24 is now known as sigmavirus24_awa | 01:50 | |
| *** mahito has quit IRC | 02:02 | |
| *** mahito has joined #openstack-ansible | 02:04 | |
| *** fawadkhaliq has quit IRC | 02:09 | |
| cloudnull | Sleep is for the weak. | 02:26 |
| *** mahito has quit IRC | 02:50 | |
| openstackgerrit | Steve Lewis proposed stackforge/os-ansible-deployment: Set Rabbit recovery method for network partitions https://review.openstack.org/186243 | 02:53 |
| openstackgerrit | Steve Lewis proposed stackforge/os-ansible-deployment: Set Rabbit recovery method for network partitions https://review.openstack.org/186244 | 03:07 |
| openstackgerrit | Steve Lewis proposed stackforge/os-ansible-deployment: Set Rabbit recovery method for network partitions https://review.openstack.org/186245 | 03:14 |
| openstackgerrit | Steve Lewis proposed stackforge/os-ansible-deployment: Set Rabbit recovery method for network partitions https://review.openstack.org/186244 | 03:18 |
| *** jmccrory has joined #openstack-ansible | 03:28 | |
| *** jmccrory has quit IRC | 03:56 | |
| *** JRobinson__ is now known as JRobinson__afk | 04:12 | |
| *** sdake has joined #openstack-ansible | 04:39 | |
| *** openstackgerrit has quit IRC | 04:50 | |
| *** sdake_ has joined #openstack-ansible | 04:50 | |
| *** openstackgerrit has joined #openstack-ansible | 04:50 | |
| *** javeriak has joined #openstack-ansible | 04:53 | |
| *** JRobinson__afk is now known as JRobinson__ | 04:53 | |
| *** sdake has quit IRC | 04:54 | |
| *** fawadkhaliq has joined #openstack-ansible | 04:59 | |
| *** cloudm2 has quit IRC | 05:02 | |
| *** stevemar has joined #openstack-ansible | 05:06 | |
| *** mahito has joined #openstack-ansible | 05:23 | |
| *** dkehn__ has joined #openstack-ansible | 05:24 | |
| *** mahito has quit IRC | 05:27 | |
| *** dkehn has quit IRC | 05:27 | |
| *** mahito has joined #openstack-ansible | 05:46 | |
| *** Mudpuppy_ has quit IRC | 05:49 | |
| *** radek__ has joined #openstack-ansible | 06:16 | |
| *** stevemar has quit IRC | 06:25 | |
| *** javeriak has quit IRC | 06:40 | |
| *** JRobinson__ has quit IRC | 07:13 | |
| *** igorr1 has joined #openstack-ansible | 07:35 | |
| *** mahito has quit IRC | 07:44 | |
| *** mahito has joined #openstack-ansible | 07:45 | |
| *** Mudpuppy has joined #openstack-ansible | 07:50 | |
| *** Mudpuppy has quit IRC | 07:55 | |
| *** mahito has quit IRC | 08:13 | |
| * svg thinks cloudnull went to sleep | 08:25 | |
| openstackgerrit | Matt Thompson proposed stackforge/os-ansible-deployment: add option to use dd for swapfile creation https://review.openstack.org/186051 | 08:28 |
| *** dkalleg has joined #openstack-ansible | 08:43 | |
| *** dkalleg has quit IRC | 08:48 | |
| *** sdake has joined #openstack-ansible | 08:58 | |
| *** sdake_ has quit IRC | 09:01 | |
| *** sdake_ has joined #openstack-ansible | 09:04 | |
| *** sdake has quit IRC | 09:07 | |
| *** markvoelker_ has quit IRC | 09:16 | |
| svg | how does an image flow, when uploaded through horizon? | 09:25 |
| svg | I got mine getting stuck in status queued, and no clue from glance logs (even with debug true) | 09:26 |
| odyssey4me | hmm, sigmavirus24_awa is probably best positioned to answer that properly - but at a high level with my perhaps outdated knowledge it goes a little something like this | 09:26 |
| odyssey4me | file upload started -> image registered in the glance registry -> file uploaded to horizon web server -> horizon server uploads image to glance store | 09:28 |
| odyssey4me | the whole image has to be uploaded to the horizon server first, which is why it's recommended that you disable the feature | 09:28 |
| odyssey4me | or make your horizon containers really large to handle a large quantity of large images | 09:29 |
| svg | hm ok, so that sounds like it might be stuck on the horizon container | 09:29 |
| odyssey4me | that was how it worked in Grizzly AFAIK, and I don't think it's changed | 09:30 |
| odyssey4me | as I recall there was talk of implementing an upload stream to Glance instead, but I doubt that's actually been done | 09:30 |
| svg | I read sometehing about that too | 09:32 |
| svg | of course, horizon logs are basically succinct apache logs | 09:32 |
| mattt | svg: you should try a manual image upload through glance client to rule horizon out | 09:32 |
| mattt | unless you did that already, at which point you're free to ignore me :P | 09:33 |
| svg | I would never ignore you Matt | 09:34 |
| odyssey4me | svg check the apache error logs - the python trace events should show there | 09:36 |
| svg | checked those, nothing there | 09:37 |
| mattt | you're checking /var/log/horizon and /var/log/apache ? | 09:39 |
| mattt | (can't remember what goes where, but best to check both locations) | 09:39 |
| svg | the horizon one | 09:43 |
| svg | hm, found a little stacktrace on one of the apache/error, but was sleeping at that time | 09:44 |
| odyssey4me | mattt do you happen to have a kilo environment running? | 09:53 |
| odyssey4me | in my AIO it would seem that swift is configured, but not running | 09:53 |
| openstackgerrit | Merged stackforge/os-ansible-deployment: Backported some bootstrap processes for gating https://review.openstack.org/186069 | 09:56 |
| odyssey4me | oh, odd - swift-init says it's not running - but a process list shows it running | 09:56 |
| mattt | odyssey4me: yep but deployed a good # of days ago | 10:01 |
| mattt | you want your key on it ? | 10:01 |
| odyssey4me | nope - I'm just not seeing logs for it and digging into why | 10:01 |
| mattt | pretty sure swift is working because it's multi node and i use it for glance backend | 10:01 |
| odyssey4me | fair enough - it does appear to be running, but I see no logs | 10:02 |
| odyssey4me | interestingly I am seeing logs for nova compute in syslog - not sure how it's getting there! | 10:02 |
| mattt | swift logs go into an odd location iirc | 10:03 |
| odyssey4me | swift logs to syslog on LOCAL2 | 10:03 |
| odyssey4me | rsyslog is meant to drop them into /var/log/swift -> /openstack/logs/aio1-swift | 10:04 |
| odyssey4me | but only the proxy is actually producing logs | 10:04 |
| mattt | yeah that's what i was referring to | 10:05 |
| mattt | keep in mind stuff runs on the host itself so you have logs in /openstack/logs/aio1 for swift | 10:05 |
| odyssey4me | ah, in fact swift should be logging to local1/local2/local3/local4 | 10:05 |
| odyssey4me | ah yes I see that now - silly me | 10:06 |
| mattt | gets me every time :P | 10:06 |
| odyssey4me | considering that there's a symlink, perhaps we should just be logging to /var/log/swift/ and we won't get caught with this nonsense any more | 10:07 |
| *** markvoelker has joined #openstack-ansible | 10:17 | |
| *** markvoelker has quit IRC | 10:23 | |
| svg | hm, so I tested with the glance client, and at first I didnt issue a container format (as I didnt see that option in horizon either), and the result was "Invalid container format 'None' for image.", with the image staying queued too | 10:30 |
| svg | adding that container format worked now | 10:30 |
| *** sdake_ has quit IRC | 10:38 | |
| openstackgerrit | Merged stackforge/os-ansible-deployment: Add alarm for rabbitmq_max_channels_per_conn https://review.openstack.org/186001 | 10:51 |
| openstackgerrit | Merged stackforge/os-ansible-deployment: fix bash/curl command in development-stack.rst https://review.openstack.org/186091 | 10:51 |
| openstackgerrit | Merged stackforge/os-ansible-deployment: Backported some bootstrap processes for gating https://review.openstack.org/186102 | 11:03 |
| mattt | svg: do you mind if i push a change to your master ceph review ? | 11:27 |
| svg | no, of course not | 11:29 |
| svg | how do you track that afterwards? | 11:30 |
| * mattt gets ready to break stuff | 11:30 | |
| svg | I'm keeping my original commits, as i need to track them on a test-deploy branch | 11:30 |
| svg | and squash tem before git reviewing them | 11:31 |
| mattt | svg: they're all there in gerrit anyway | 11:31 |
| mattt | you can view previews revisions | 11:31 |
| mattt | so if you don't agree w/ this change we can revert | 11:31 |
| svg | I'd need to be able to cherry pick your change onto my local master-ceph-devel branch | 11:31 |
| *** openstackgerrit has quit IRC | 11:39 | |
| *** openstackgerrit has joined #openstack-ansible | 11:39 | |
| svg | What could be the cause of all compue_hosts having a state down? | 11:43 |
| mattt | nova-compute not running? :P | 11:43 |
| openstackgerrit | Matt Thompson proposed stackforge/os-ansible-deployment: [WIP] Ceph/RBD support https://review.openstack.org/181957 | 11:43 |
| mattt | svg: https://review.openstack.org/#/c/181957/15..16/playbooks/roles/os_nova/templates/nova.conf.j2 and https://review.openstack.org/#/c/181957/15..16/playbooks/roles/ceph_client/vars/main.yml | 11:45 |
| mattt | the latter being super nasty :( | 11:45 |
| mattt | but cinder-volume pukes if ceph package isn't installed | 11:45 |
| mattt | (at least when you do something like nova boot --flavor 2 --image Trusty --block-device source=image,id=0374cb7f-4d0c-4cbf-848e-5d703465bea2,dest=volume,size=20,shutdown=preserve,bootindex=0 test | 11:47 |
| mattt | ) | 11:47 |
| svg | ceph package: where did yo notice? | 11:47 |
| mattt | svg: in the cinder-volume container | 11:47 |
| mattt | svg: https://gist.githubusercontent.com/mattt416/87f4e3d02d7ee5a413fa/raw/83c4dd7abea53e905c8cb36903a10eb137485303/gistfile1.txt | 11:48 |
| mattt | it does a ceph mon dump but that fails because ceph_argparse python mod isn't installed | 11:49 |
| svg | and that module is parte of the ceph package? | 11:50 |
| mattt | yep but it shoudl be part of ceph-common, it's a packaging bug | 11:51 |
| svg | root@dc1-rk2-ch1-bl6:~# dpkg -L python-ceph | grep argparse | 11:51 |
| svg | /usr/lib/python2.7/dist-packages/ceph_argparse.py | 11:51 |
| mattt | svg: you must be using firefly ? | 11:51 |
| mattt | i'm using the default (hammer) | 11:51 |
| svg | still on giant | 11:52 |
| mattt | maybe i should add some sort of conditional to only do it when using hammer | 11:54 |
| mattt | cuz that pulls in a lot of unneeded packages | 11:54 |
| svg | so on hammer you con;t have the python-ceph package? | 11:54 |
| svg | oh, yes, | 11:55 |
| svg | mentioned in that bug report | 11:55 |
| svg | confused with your update in the nova.conf template; I actually thought I made that change | 11:56 |
| mattt | the nova.conf template change allows you to have all nova instances backed by rbd, or to allow nova to boot from rbd cinder volumes (this is the piece you were missing) | 11:57 |
| mattt | if you want to be able to do both, you only want to set rbd_user and rbd_secret_uuid once, so i've moved those two options out of the existing if statement | 11:58 |
| svg | yes, sure, I get it, I just thought I implemented that; must have forgotten :) | 12:01 |
| svg | now, how could I cherry pick just your update? | 12:01 |
| mattt | svg: git fetch https://review.openstack.org/stackforge/os-ansible-deployment refs/changes/57/181957/16 && git cherry-pick FETCH_HEAD | 12:01 |
| *** markvoelker has joined #openstack-ansible | 12:02 | |
| svg | hm, doesnt apply cleanly | 12:05 |
| mattt | oh derp | 12:05 |
| mattt | sorry | 12:05 |
| mattt | you need to cherry-pick that individual change | 12:05 |
| mattt | hrrmph | 12:05 |
| mattt | that would cherry-pick the entire review | 12:06 |
| svg | well, the funny thing is, it does only show that change, but in a broken way | 12:06 |
| svg | AA playbooks/roles/ceph_client/vars/main.yml | 12:06 |
| svg | UU playbooks/roles/os_nova/templates/nova.conf.j2 | 12:06 |
| svg | http://sprunge.us/AWBL | 12:08 |
| svg | I don't see why he doesnt want to apply that | 12:08 |
| mattt | you def. don't have some local changes in the working dir ? | 12:11 |
| svg | nope | 12:16 |
| svg | perhaps caused by the master rebasing somewhere | 12:20 |
| mattt | ah that is possible | 12:23 |
| svg | do you have your change in a single commit somewhere? | 12:25 |
| svg | I could resolve the merge error, but it would be more cleanly to get it form you | 12:25 |
| mattt | svg: sec, let me try | 12:27 |
| svg | Wondering what kind of local work flow people here use | 12:28 |
| *** fawadkhaliq has quit IRC | 12:30 | |
| mattt | derp | 12:31 |
| mattt | looking still | 12:32 |
| mattt | svg: https://gist.github.com/mattt416/dae5e14931660f714fe0 | 12:38 |
| *** evrardjp has joined #openstack-ansible | 12:39 | |
| svg | done | 12:48 |
| svg | how do you usually handle such things? | 12:48 |
| mattt | svg: my git knowledge is extremely limited, i created two branches, in the one i pulled revision 15, in the second i pulled revision 16 | 12:49 |
| mattt | then i did a git diff <brancha> <branchb> | 12:49 |
| svg | sorry, I meant when someone does an update to a patch set of you, how do you backport that patch in local tree? | 12:52 |
| evrardjp | I'm not sure to understand what you mean, but do you mean http://www.git-scm.com/docs/git-apply ? | 12:57 |
| svg | more in general, when working on a patch set, against gerrit, which local work flow do you use? | 12:58 |
| svg | in the case of ceph, I have a branch to master, one to kilo, and I merge the latter into a customer-test-setup specific branch | 12:59 |
| svg | I also keep all commits in history, and only squash them into a master based separate branch before doing a git review | 12:59 |
| svg | So I like to keep track of different changes, which nakes it easier en less error prone to propagete them between branches | 13:00 |
| svg | (sometimes i might do a change on the master branch, sometime firt on the testsetup branch) | 13:00 |
| svg | now when mattt did an update on my patch set, I need to have that change set as a seperate commit, I can cherry-pick locally | 13:01 |
| svg | all in all a bit of work | 13:01 |
| svg | so just wondering what different ways you guys use | 13:01 |
| *** metral is now known as metral_zzz | 13:03 | |
| vincent_vdk | so if i'm not mistaken the hypervisor with a AIO is the host itself. right? | 13:12 |
| mattt | vincent_vdk: yeah | 13:13 |
| vincent_vdk | ok, the reason why i ask this is that cinder expects a cinder-volumes VG | 13:13 |
| vincent_vdk | but it's not created | 13:13 |
| vincent_vdk | though the aio init script should take care of this | 13:13 |
| vincent_vdk | not sure why it's not created... | 13:14 |
| mattt | vincent_vdk: i have a cinder-volumes VG on my AIO | 13:18 |
| mattt | (which i'm building as we speak) | 13:18 |
| vincent_vdk | weird | 13:18 |
| vincent_vdk | if i'm not mistaken it uses a /dev/loop3 | 13:18 |
| vincent_vdk | (in my case) | 13:18 |
| mattt | mine is /dev/loop0 | 13:18 |
| vincent_vdk | when i manually tru the steps in the init script the pv fails | 13:19 |
| vincent_vdk | device not found.... | 13:19 |
| mattt | i have loop0-loop3 | 13:20 |
| mattt | loop1-3 being swift disks | 13:20 |
| mattt | vincent_vdk: when did you deploy btw ? and what are you deploying on ? | 13:21 |
| vincent_vdk | mine are swift=0,1,2 | 13:21 |
| vincent_vdk | i see this in lvm config | 13:21 |
| vincent_vdk | filter = [ "a/loop0/", "r/.*/" ] | 13:21 |
| mattt | vincent_vdk: wonder if you need this? https://github.com/stackforge/os-ansible-deployment/commit/ccb16763234a13fc204b91943e3d0c197545f2e9 | 13:21 |
| vincent_vdk | which might explain why i cant create a pv on top of it | 13:22 |
| mattt | vincent_vdk: are you using ext3 ? | 13:22 |
| mattt | on your root drive | 13:22 |
| vincent_vdk | ext4 | 13:22 |
| mattt | i see that same filter | 13:23 |
| vincent_vdk | i'm going to hack a bit | 13:24 |
| mattt | k | 13:26 |
| vincent_vdk | i can pv loop3 when i change the filter to accept all loop devices | 13:27 |
| vincent_vdk | now let's see if i can deploy an instance | 13:27 |
| *** KLevenstein has joined #openstack-ansible | 13:37 | |
| mattt | vincent_vdk: any luck ? | 13:42 |
| vincent_vdk | nah, | 13:44 |
| vincent_vdk | thing is that i'm brand new to openstack | 13:45 |
| vincent_vdk | i'm like 1 week in | 13:45 |
| vincent_vdk | i do see that the cinder volume container complains about a cinder-volumes vg that does not exist | 13:45 |
| vincent_vdk | should it be defined in the container as well? | 13:46 |
| mattt | vincent_vdk: you can boot instances w/out cinder btw | 13:46 |
| mattt | vincent_vdk: but yeah, you should be able to vgdisplay cinder-volumes in the cinder-volumes container | 13:47 |
| vincent_vdk | yeah, but i need to understand how cinder works for a use case we have here | 13:47 |
| vincent_vdk | ah, that is not the case | 13:47 |
| mattt | vincent_vdk: perhaps you can stop/start container | 13:48 |
| vincent_vdk | i'll give it a try | 13:54 |
| vincent_vdk | mattt: didn't help | 14:00 |
| *** sigmavirus24_awa is now known as sigmavirus24 | 14:02 | |
| *** Mudpuppy has joined #openstack-ansible | 14:02 | |
| svg | So he did try turning it opff and on again. | 14:07 |
| vincent_vdk | yes i did :) | 14:09 |
| *** sdake has joined #openstack-ansible | 14:09 | |
| vincent_vdk | btw, i created the vg manually and restarted the cinder service | 14:09 |
| vincent_vdk | now i can create volume without errors | 14:09 |
| vincent_vdk | not sure why th vg wasn't there though | 14:10 |
| *** sdake_ has joined #openstack-ansible | 14:10 | |
| sigmavirus24 | svg: you resolve the glance problems you were having? | 14:13 |
| *** sdake has quit IRC | 14:14 | |
| *** fawadkhaliq has joined #openstack-ansible | 14:19 | |
| *** fawadkhaliq has quit IRC | 14:24 | |
| svg | sigmavirus24: yes, I had to deactivete the VPN providerm, which was not installed, but was somehow triggered as I enabled lbaas | 14:25 |
| sigmavirus24 | o_O | 14:26 |
| svg | euh | 14:26 |
| svg | sorry, ixing things up | 14:26 |
| svg | I managed to uplad my image manually, with the cli client, but had to provide the container format 'bare' | 14:26 |
| svg | I found a bug that could be related, though can't seem to find it anymore | 14:34 |
| odyssey4me | svg ther emay be a Horizon setting which enables/disables inputting that field, or setting a default. | 14:35 |
| odyssey4me | vincent_vdk the AIO setup is largely focused on being used in a p1-8 instance in RAX or HP cloud... if your hardware is different then it may require some tweaking. | 14:36 |
| odyssey4me | That said, if you've setup a vg for cinder prior to bootstrapping the AIO then it'll respect it. It only sets stuff up which isn't already there.... mostly :p | 14:37 |
| svg | I didn't find any in the provided file at least, only reference is OPENSTACK_IMAGE_BACKEND | 14:37 |
| sigmavirus24 | svg let me know what glance bugs you find and I'll be happy to track them down | 14:39 |
| sigmavirus24 | If they're osad problems or glance problems I can fix either | 14:39 |
| odyssey4me | sigmavirus24 I think it may be a Horizon bug. Perhaps stevelle or miguelgrinberg should poke into it. | 14:40 |
| *** metral_zzz is now known as metral | 14:40 | |
| odyssey4me | Or not a bug, but rather a 'feature'. | 14:40 |
| *** sdake has joined #openstack-ansible | 14:40 | |
| sigmavirus24 | hehe | 14:40 |
| sigmavirus24 | good volunteering odyssey4me | 14:40 |
| odyssey4me | :p | 14:41 |
| svg | might be related to this one: https://bugs.launchpad.net/glance/+bug/1268680 | 14:42 |
| openstack | Launchpad bug 1268680 in Glance "Creating an image without container format queues image and fails with 400" [Low,In progress] - Assigned to Mike Fedosin (mfedosin) | 14:42 |
| *** sdake_ has quit IRC | 14:44 | |
| sigmavirus24 | svg: I'll bug Mike then ;) | 14:45 |
| svg | but i can't find any reference on what might be missing in horizon | 14:46 |
| sigmavirus24 | "proper usage of glanceclient" is what I'd guess | 14:46 |
| *** metral is now known as metral_zzz | 14:47 | |
| svg | yes, seems like this answer is deprecated: https://openstack.nimeyo.com/30539/openstack-specify-container-format-creating-image-horizon | 14:49 |
| svg | something else, my kilo setup had marked all compute_hosts's status down, and i struggle to find any relevant errors on why that is | 14:50 |
| *** stevemar has joined #openstack-ansible | 14:51 | |
| mattt | svg: going to remove the WIP from https://review.openstack.org/#/c/181957/ | 14:58 |
| mattt | so we can get some more eyes on it | 14:59 |
| svg | ok | 14:59 |
| openstackgerrit | Matt Thompson proposed stackforge/os-ansible-deployment: Ceph/RBD support https://review.openstack.org/181957 | 14:59 |
| svg | What happened to the logging infra in kilo? | 14:59 |
| mattt | your compute nodes still marked down ? | 14:59 |
| svg | yes | 14:59 |
| mattt | why does changing topic in gerrit wipe all reviews :( | 14:59 |
| mattt | s/topic/commit message/ | 15:00 |
| mattt | svg: and nova-compute is running ? | 15:00 |
| sigmavirus24 | mattt: because the commit message is part of the review | 15:01 |
| sigmavirus24 | So it wipes reviews because they were for a different commit msg | 15:01 |
| * mattt cries | 15:01 | |
| svg | mattt yes compute is running | 15:01 |
| mattt | svg: where are you seeing that, in horizon ? | 15:03 |
| mattt | svg: if you do a ps and grep for nova-compute, is pid constantly changing ? | 15:04 |
| *** galstrom_zzz is now known as galstrom | 15:04 | |
| *** galstrom is now known as galstrom_zzz | 15:05 | |
| svg | in hoorizon, but also nova hypervisor-list | 15:06 |
| svg | + pid is stable | 15:06 |
| svg | (what is the meaning of workflow +1?) | 15:07 |
| svg | mattt ^^ | 15:08 |
| *** dkalleg has joined #openstack-ansible | 15:10 | |
| *** igorr1 has quit IRC | 15:12 | |
| odyssey4me | mattt it submits it as a new patch set | 15:12 |
| odyssey4me | it won't wipe +2's though, only -w and +1's | 15:12 |
| odyssey4me | oh, that last comment related to a rebase | 15:13 |
| odyssey4me | svg workflow +1 is required to signal gerrit to go ahead with a merge if it has enough votes | 15:13 |
| odyssey4me | (and has been verified) | 15:14 |
| *** dkalleg has quit IRC | 15:14 | |
| svg | o, so that is a manual step | 15:15 |
| odyssey4me | yep, someone has to give the go ahead | 15:19 |
| odyssey4me | often it's just a core member - but sometimes if the decision is big enough the PTL will ask for more votes than usual and do the +w | 15:20 |
| odyssey4me | when satisfied that consensus has been reached | 15:20 |
| *** jwagner_away is now known as jwagner | 15:26 | |
| svg | I'd expect to not be able to do a w+1, but seems I can | 15:32 |
| svg | *could* | 15:32 |
| openstackgerrit | Merged stackforge/os-ansible-deployment: Add warning and critical thresholds checks https://review.openstack.org/184110 | 15:53 |
| *** fawadkhaliq has joined #openstack-ansible | 15:54 | |
| *** sdake_ has joined #openstack-ansible | 15:58 | |
| *** sdake has quit IRC | 15:59 | |
| *** sdake has joined #openstack-ansible | 16:00 | |
| cloudnull | meeting time cloudnull, mattt, andymccr, d34dh0r53, hughsaunders, b3rnard0, palendae, Sam-I-Am, odyssey4me, serverascode, rromans, mancdaz, dolphm, _shaps_, BjoernT, claco, echiu, dstanek, jwagner | 16:00 |
| d34dh0r53 | o\ | 16:00 |
| dstanek | \o | 16:00 |
| d34dh0r53 | /0\ | 16:01 |
| cloudnull | in #openstack-meeting-4 | 16:01 |
| cloudnull | https://wiki.openstack.org/wiki/Meetings/openstack-ansible | 16:01 |
| openstackgerrit | Merged stackforge/os-ansible-deployment: Set Rabbit recovery method for network partitions https://review.openstack.org/186244 | 16:02 |
| *** sdake_ has quit IRC | 16:03 | |
| *** metral_zzz is now known as metral | 16:04 | |
| *** jmccrory has joined #openstack-ansible | 16:28 | |
| *** metral is now known as metral_zzz | 16:36 | |
| *** dkalleg has joined #openstack-ansible | 17:17 | |
| svg | Do I see it right, the logstash-elastics-kibana setup was torn out in kilo, and moved to rpc-extras? | 17:17 |
| sigmavirus24 | svg: you are seeing it correctly | 17:19 |
| *** javeriak has joined #openstack-ansible | 17:25 | |
| *** daneyon has joined #openstack-ansible | 17:32 | |
| *** metral_zzz is now known as metral | 17:33 | |
| *** daneyon_ has joined #openstack-ansible | 17:33 | |
| *** sdake_ has joined #openstack-ansible | 17:35 | |
| *** daneyon has quit IRC | 17:36 | |
| *** sdake has quit IRC | 17:39 | |
| *** javeriak has quit IRC | 17:45 | |
| svg | ok, I struggle to find the review about that removal | 17:47 |
| svg | Gerrit searches in mysterious ways | 17:47 |
| *** galstrom_zzz is now known as galstrom | 17:51 | |
| *** javeriak has joined #openstack-ansible | 17:56 | |
| openstackgerrit | Javeria Khan proposed stackforge/os-ansible-deployment-specs: Add PLUMgrid Neutron Plugin to neutron playbooks https://review.openstack.org/184726 | 18:34 |
| *** KLevenstein has quit IRC | 19:05 | |
| sigmavirus24 | svg: it does | 19:11 |
| sigmavirus24 | svg: I suspect it's one of Kevin's patches and it was a mega patch | 19:11 |
| openstackgerrit | Javeria Khan proposed stackforge/os-ansible-deployment: Modularizing Neutron playbooks https://review.openstack.org/184665 | 19:17 |
| odyssey4me | svg, it was within my de-rax patch I think | 19:20 |
| odyssey4me | quite some time ago | 19:21 |
| svg | How is that kibana stuff rax specific? | 19:21 |
| odyssey4me | svg it's not that it's necessarily rax specific, but that it's not part of a standard openstack deployment - it's an opinionated deployment | 19:22 |
| odyssey4me | we very possibly may split it out from rpc-extras into its own roles and plays somewhere, or if there's enough demand from the community we could roll it into OSAD | 19:22 |
| odyssey4me | the trouble we have is defining the scope of OSAD | 19:22 |
| odyssey4me | how far do we go before it becomes too much to manage and maintain | 19:23 |
| svg | sounds legit | 19:23 |
| odyssey4me | why should the project focus on maintaining code for non-openstack things when those things could be external roles consumed by the project? | 19:24 |
| *** galstrom is now known as galstrom_zzz | 19:24 | |
| *** KLevenstein has joined #openstack-ansible | 19:24 | |
| odyssey4me | this is the dilemma - we can do all the things, but that carries a cost which we don't have enough contributors to maintain... so we need to choose what we include wisely | 19:25 |
| odyssey4me | we're busy doing tons of work on the ELK stack for RPC11's release - what we had before was OK, but didn't scale... we hope that our mods will scale much better and work better | 19:26 |
| *** sdake has joined #openstack-ansible | 19:26 | |
| odyssey4me | and, of course, rpc-extras is fully open and consumable by anyone who wants to | 19:26 |
| odyssey4me | you can also be selective with what you consume - no need to go the whole hog | 19:26 |
| svg | yeah, being able to easily consume external stuff is nice | 19:27 |
| svg | but, so you say you are doing lot of work on ELK for RPC11 - you mean rackspace and not osad there? | 19:28 |
| *** sdake_ has quit IRC | 19:29 | |
| *** radek__ has quit IRC | 19:34 | |
| *** galstrom_zzz is now known as galstrom | 19:35 | |
| *** radek__ has joined #openstack-ansible | 19:35 | |
| evrardjp | This is exactly the same conversation as what happened at the meeting... | 19:49 |
| sigmavirus24 | svg yes | 19:49 |
| sigmavirus24 | svg: rpc-extras is working hard on ELK and OS-A-D is just a way of installing /just/ openstack | 19:49 |
| evrardjp | isn't it fine that way? | 19:50 |
| sigmavirus24 | os-a-d icehouse/juno branches are Rackspace tech debt | 19:50 |
| evrardjp | I mean, shouldn't we focus on OSAD? | 19:50 |
| odyssey4me | svg yeah, RPC consumes OSAD as an upstream product | 19:50 |
| sigmavirus24 | evrardjp: yeah, but it breaks compatibility between those technical debt branches and kilo | 19:50 |
| sigmavirus24 | and svg has been using juno iirc | 19:50 |
| odyssey4me | so the team that builds RPC work on OSAD as an upstream product | 19:50 |
| evrardjp | oh, now I get the issue | 19:50 |
| sigmavirus24 | evrardjp: yeah, sorry, that was apparent to everyone else in this conversation (context: it helps) | 19:51 |
| evrardjp | I've never used OSAD only | 19:51 |
| evrardjp | but that's another topic | 19:51 |
| evrardjp | breaking backwards compatibility isn't fine? Because there wasn't any OSAD at first... so it seems not to be a large issue at first | 19:52 |
| stevelle | better to get it done earlier than later, for sure | 19:53 |
| sigmavirus24 | evrardjp: correct. The first truly community version of OSAD is on the kilo branch | 19:55 |
| odyssey4me | svg just got into the game a little early :p | 19:56 |
| odyssey4me | but, svg, your voice is part of this ecosystem - you can lobby in the group meetings for re-inclusion if you like :) | 19:56 |
| evrardjp | I don't understand HOW the breakup between the RPC product and OSAD happened, but IMHO it's perfectly fine the way it currently is: OSAD as upstream and rackspace specifics in their own repositories... It could be either ELK, or whatever plugin/backend they want on a non-"reference" implementation | 19:58 |
| evrardjp | just my 2 cents as a noob of this chan | 19:59 |
| evrardjp | with my view of consumer | 19:59 |
| odyssey4me | evrardjp yeah, so a discussion we need to have is what that reference implementation is | 20:00 |
| odyssey4me | that would help us with the scoping of what belongs in the project or noto | 20:00 |
| svg | I have a juno test-cluster, but not plans to go in prod with that, will be kilo | 20:01 |
| odyssey4me | Right now we're doing a lot of things in OSAD which are available in other ansible roles, or which should perhaps be regarded as out of scope. | 20:01 |
| sigmavirus24 | For example, do you as a consumer want the installation to work even if you have a corporate proxy between your DC and the outside world that likes to break stuff? | 20:01 |
| openstackgerrit | Javeria Khan proposed stackforge/os-ansible-deployment-specs: Add PLUMgrid Neutron Plugin to neutron playbooks https://review.openstack.org/184726 | 20:02 |
| odyssey4me | d34dh0r53 swift logs now done - if you find anything not coming out as expected, let me know | 20:03 |
| d34dh0r53 | odyssey4me: will do, thank you sir! | 20:03 |
| odyssey4me | I do wish that all projects had one of these pages: http://docs.openstack.org/developer/swift/logs.html | 20:04 |
| *** daneyon has joined #openstack-ansible | 20:05 | |
| evrardjp | before giving my opinion on scope, which is a tricky question, I wonder something... | 20:05 |
| evrardjp | at some point, rackspace have to make money on their product, and I guess they would prefer releasing their next version of RPC (v12?) as soon as possible. If there are delays in the OSAD splitting, they won't be able to release anything stable, right? | 20:05 |
| evrardjp | so if they take more time to release their work, it could/will affect negatively the OSAD project? | 20:05 |
| evrardjp | I'm maybe mistaken, I don't know the complete story, but I'm interested by knowing it | 20:06 |
| *** yaya has joined #openstack-ansible | 20:07 | |
| evrardjp | in other terms: isn't is easier to wait for rackspace to release their next version and make them split in whatever they seem fit? This would be better for them, and easier to go forward... | 20:07 |
| evrardjp | But I'm maybe out of line | 20:07 |
| odyssey4me | evrardjp RPC consumes OSAD - so the first focus is OSAD, once that releases then RPC's dev/docs/QE teams step in and work on the RPC update | 20:07 |
| evrardjp | ok | 20:07 |
| odyssey4me | so OSAD will always lead RPC | 20:08 |
| evrardjp | now I get it | 20:08 |
| *** daneyon_ has quit IRC | 20:08 | |
| openstackgerrit | Javeria Khan proposed stackforge/os-ansible-deployment: Modularizing Neutron playbooks https://review.openstack.org/184665 | 20:08 |
| odyssey4me | OSAD released kilo on the same day (or day after - I was on leave) as upstream kilo was tagged | 20:08 |
| odyssey4me | we had been working on OSAD v11/kilo for two weeks ahead of that tag | 20:08 |
| odyssey4me | RPC11 has yet to be released while we fix bugs we identify in OSAD while working on RPC extras/docs | 20:09 |
| svg | odyssey4me: proxy support is a must imho | 20:09 |
| odyssey4me | so RPC11 will be released based on a minor tag of OSAD which includes bugs RPC is ready to support with the SLA promised to customers | 20:10 |
| odyssey4me | there is no conspiracy here :) | 20:10 |
| odyssey4me | FYI svg if you want to see the up-to-the-minute ELK updates we're working on, it's happening here: https://github.com/odyssey4me/rpc-extras/tree/log_shipping | 20:11 |
| odyssey4me | I'm hoping to issue the PR to rpc-extras this evening | 20:11 |
| openstackgerrit | Javeria Khan proposed stackforge/os-ansible-deployment-specs: Add PLUMgrid Neutron Plugin to neutron playbooks https://review.openstack.org/184726 | 20:11 |
| evrardjp | didn't know you were using maas ;D | 20:12 |
| evrardjp | but it's another story | 20:12 |
| odyssey4me | evrardjp if your maas the same as our maas? | 20:13 |
| odyssey4me | ie Monitoring as a Service from RAX | 20:13 |
| evrardjp | nope | 20:13 |
| *** yaya has left #openstack-ansible | 20:13 | |
| odyssey4me | ah, you're thinking Ubuntu Metal as a Service? | 20:13 |
| evrardjp | I read MaaS on the github repo, I was mistaken. but I said "it's another story" | 20:13 |
| evrardjp | idd | 20:14 |
| odyssey4me | :) yeah, I was confused when I first started too | 20:14 |
| evrardjp | to come back to the proxy stuff, is it really a part of openstack deployment? I see that as "infrastructure" | 20:15 |
| evrardjp | more like something linked with bare metal... but, ofc, it impacts the deployment | 20:15 |
| odyssey4me | actually our monitoring stuff, while designed to be consumed by RAX Monitoring, are usable for nagios and other similar check-type monitoring systems | 20:15 |
| odyssey4me | evrardjp I would agree, but we have grey areas... | 20:16 |
| odyssey4me | for example, we already drop a pip config and would overwrite it if something else edited it - so we must cater for a proxy there | 20:16 |
| evrardjp | odyssey4me: that would be good to document somewhere (the nagios thingy) | 20:16 |
| evrardjp | although I won't ever use it | 20:16 |
| odyssey4me | evrardjp it's more of an interesting thing to know conceptually :) | 20:17 |
| evrardjp | ofc, I'll compare with what I've done with our zabbix system, to make sure we have enough checks | 20:18 |
| evrardjp | more checks are good, right? :D | 20:18 |
| odyssey4me | evrardjp Zabbix is powerful, but a bit of a mess. I used it where I worked previously. | 20:19 |
| odyssey4me | I wrote a few checks which need a lot of love. :p | 20:19 |
| evrardjp | I'm using nagios style checks with our zabbix for the openstack, maybe I'll compare when I'll have time ;) | 20:20 |
| evrardjp | so I didn't give my opinion earlier today, but I think neutron is a super mess so if there are every plugin in it, it's gonna be difficult... | 20:21 |
| evrardjp | I may have an idea: we settle for an arbitrary number, for example 90%. | 20:22 |
| odyssey4me | evrardjp there're several different groups working on monitoring for openstack | 20:22 |
| odyssey4me | many use nagios, and of course zabbix consumes nagios checks happily | 20:23 |
| evrardjp | We then check on the last openstack survey the deployments, and everything that is used more than this arbitrary number of percents has to be in the OSAD | 20:23 |
| odyssey4me | hahaha | 20:24 |
| odyssey4me | bah, it really comes down to how many people are prepared to develop feature x and maintain it afterwards | 20:24 |
| evrardjp | it would be resumed to a (maybe chicken and egg thing) math calculation | 20:24 |
| odyssey4me | we're actively seeking non-RAX cores | 20:25 |
| evrardjp | if they don't maintain, nobody will deploy it -> decrease in the polls | 20:25 |
| odyssey4me | we need more diversity | 20:25 |
| evrardjp | I didn't got the chance to see the talk @Vancouver with Auro | 20:25 |
| sigmavirus24 | evrardjp: it's online already! | 20:26 |
| *** sdake_ has joined #openstack-ansible | 20:26 | |
| evrardjp | I know... I could also have gone to that room | 20:26 |
| evrardjp | :p | 20:26 |
| evrardjp | always these damn priorities! | 20:26 |
| sigmavirus24 | evrardjp: ah okay | 20:27 |
| sigmavirus24 | https://www.youtube.com/watch?v=mQTS_n_nxAs for anyone else curious | 20:27 |
| evrardjp | I'll check on my side with our team what would be possible to do for this diversity | 20:28 |
| evrardjp | anyway, I'm off, it's beginning to be late in Europe... | 20:28 |
| evrardjp | thanks for the interesting talk | 20:29 |
| odyssey4me | evrardjp :) I'm in the UK and it's late, but deadlines :p | 20:29 |
| *** sdake has quit IRC | 20:30 | |
| svg | HA! In belgium still an hour laterz | 20:30 |
| evrardjp | svg, you're belgian? | 20:30 |
| svg | yup | 20:30 |
| *** galstrom is now known as galstrom_zzz | 20:41 | |
| *** javeriak has quit IRC | 20:43 | |
| *** metral is now known as metral_zzz | 20:47 | |
| palendae | Wow, channel's grown since summit, good | 20:47 |
| *** sdake_ has quit IRC | 20:47 | |
| d34dh0r53 | zomg palendae, welcome back sir | 20:48 |
| *** daneyon has quit IRC | 20:48 | |
| palendae | d34dh0r53, Not fully back, but bored | 20:49 |
| *** sdake has joined #openstack-ansible | 20:49 | |
| palendae | was the repo-builder pulled completely out of osad? https://github.com/rcbops/rpc-extras/commit/ef5881399e43e74072eb83e78425229237447a91 | 20:52 |
| odyssey4me | palendae nope, but it's being added to rpc-extras in order to build python packages for rpc-extras | 20:56 |
| palendae | Ok, that's what I thought. Just confirming | 20:56 |
| odyssey4me | it builds its own repository on the same repo servers, but under a different path | 20:56 |
| odyssey4me | this is necessitated due to the recent pip conf lockdown changes | 20:56 |
| odyssey4me | (with pip 7.x) | 20:56 |
| palendae | Ah | 20:56 |
| odyssey4me | palendae even if you try to override the index used on the cli with pip 7 and our lockdown config - it no longer works! | 20:57 |
| odyssey4me | you have to use the built repo | 20:57 |
| odyssey4me | something of a blessing and curse :p | 20:57 |
| odyssey4me | anyway, I'm out for the night | 20:58 |
| *** javeriak has joined #openstack-ansible | 21:06 | |
| *** sdake_ has joined #openstack-ansible | 21:11 | |
| *** sdake has quit IRC | 21:14 | |
| *** Sam-I-Am has joined #openstack-ansible | 21:35 | |
| Sam-I-Am | huh, how the heck did this channel disappear | 21:36 |
| *** openstackgerrit has quit IRC | 21:36 | |
| *** openstackgerrit has joined #openstack-ansible | 21:37 | |
| svg | Sam-I-Am: sounds mor elike you disappeared :) | 21:44 |
| vincent_vdk | 8G RAM for the aio seems a bit on the low side | 21:50 |
| vincent_vdk | just OOM'ed one of the galera nodes | 21:50 |
| *** fawadkhaliq has quit IRC | 21:53 | |
| openstackgerrit | Merged stackforge/os-ansible-deployment: Restart Apache when Keystone configuration changes https://review.openstack.org/185440 | 21:53 |
| openstackgerrit | Merged stackforge/os-ansible-deployment: Restart Apache when Keystone configuration changes https://review.openstack.org/185231 | 21:53 |
| palendae | vincent_vdk, It is, but that's what we're limited to with the openstack infra for testing | 21:54 |
| palendae | Well, gating | 21:55 |
| palendae | If you have access to more, it's certainly not going to be wasted | 21:55 |
| cloudnull | vincent_vdk: in our gate job we scale back the galera cache and mem usage so that itll fit within the 8GB ram that OS infra gives us. https://github.com/stackforge/os-ansible-deployment/blob/master/scripts/bootstrap-aio.sh#L302-L305 | 21:57 |
| vincent_vdk | i see | 21:58 |
| vincent_vdk | not an issue since i can get more for testing | 21:58 |
| *** sdake has joined #openstack-ansible | 21:59 | |
| vincent_vdk | good night | 21:59 |
| cloudnull | have a good one. | 22:00 |
| *** sdake_ has quit IRC | 22:03 | |
| *** metral_zzz is now known as metral | 22:05 | |
| *** KLevenstein has quit IRC | 22:07 | |
| *** jwagner is now known as jwagner_away | 22:09 | |
| svg | I'm off too, have fun! | 22:11 |
| *** hoppers99 has joined #openstack-ansible | 22:17 | |
| openstackgerrit | Satheesh Kumar Ulaganathan proposed stackforge/os-ansible-deployment: Allow configuration of Nova SQLAlchemy options https://review.openstack.org/186583 | 22:22 |
| *** sdake_ has joined #openstack-ansible | 22:23 | |
| *** Mudpuppy_ has joined #openstack-ansible | 22:25 | |
| *** sdake has quit IRC | 22:27 | |
| *** jimi|ansible has quit IRC | 22:28 | |
| *** Mudpuppy has quit IRC | 22:28 | |
| *** darrenc is now known as darrenc_afk | 22:28 | |
| *** Mudpuppy_ has quit IRC | 22:29 | |
| *** galstrom_zzz is now known as galstrom | 22:29 | |
| *** darrenc_afk is now known as darrenc | 22:37 | |
| *** galstrom is now known as galstrom_zzz | 22:49 | |
| openstackgerrit | Jimmy McCrory proposed stackforge/os-ansible-deployment: Perform sort when creating hostnames_ip file https://review.openstack.org/186591 | 23:00 |
| *** JRobinson__ has joined #openstack-ansible | 23:34 | |
| *** jmccrory has quit IRC | 23:35 | |
| *** radek__ has quit IRC | 23:59 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!