| mrstanwell | multi-projection | 00:02 |
|---|---|---|
| *** markvoelker has quit IRC | 00:10 | |
| *** mrstanwell has quit IRC | 00:13 | |
| *** shoutm has quit IRC | 00:18 | |
| *** shoutm has joined #openstack-ansible | 00:21 | |
| *** markvoelker has joined #openstack-ansible | 00:31 | |
| *** shoutm_ has joined #openstack-ansible | 00:38 | |
| *** shoutm has quit IRC | 00:40 | |
| openstackgerrit | Merged stackforge/os-ansible-deployment: Update tempest configuration https://review.openstack.org/220559 | 00:44 |
| *** BjoernT has quit IRC | 00:44 | |
| *** abitha has quit IRC | 00:45 | |
| *** spotz_zzz has quit IRC | 00:47 | |
| *** lbragstad has quit IRC | 00:48 | |
| *** palendae has quit IRC | 00:48 | |
| *** eglute_s has quit IRC | 00:48 | |
| *** gus has quit IRC | 00:48 | |
| *** neillc has quit IRC | 00:48 | |
| *** darrenc has quit IRC | 00:48 | |
| *** Guest75828 has quit IRC | 00:48 | |
| *** mattoliverau has quit IRC | 00:48 | |
| *** bgmccollum has quit IRC | 00:49 | |
| *** darrenc has joined #openstack-ansible | 00:49 | |
| *** sigmavirus24_awa has quit IRC | 00:49 | |
| *** dolphm has quit IRC | 00:49 | |
| *** shoutm_ has quit IRC | 00:50 | |
| *** shoutm has joined #openstack-ansible | 00:50 | |
| *** d34dh0r53 has quit IRC | 00:50 | |
| *** jroll has quit IRC | 00:50 | |
| *** neillc has joined #openstack-ansible | 00:50 | |
| *** mgagne has quit IRC | 00:50 | |
| *** mattoliverau has joined #openstack-ansible | 00:51 | |
| *** gus has joined #openstack-ansible | 00:51 | |
| *** metral is now known as metral_zzz | 00:51 | |
| *** blewis has joined #openstack-ansible | 00:53 | |
| *** spotz_zzz has joined #openstack-ansible | 00:53 | |
| *** jroll has joined #openstack-ansible | 00:53 | |
| *** d34dh0r53 has joined #openstack-ansible | 00:53 | |
| *** eglute has joined #openstack-ansible | 00:53 | |
| *** blewis is now known as Guest18499 | 00:53 | |
| *** bgmccollum has joined #openstack-ansible | 00:54 | |
| *** mgagne has joined #openstack-ansible | 00:54 | |
| *** palendae has joined #openstack-ansible | 00:54 | |
| *** lbragstad has joined #openstack-ansible | 00:54 | |
| *** dolphm has joined #openstack-ansible | 00:54 | |
| *** metral_zzz is now known as metral | 00:55 | |
| *** sigmavirus24_awa has joined #openstack-ansible | 00:57 | |
| *** metral is now known as metral_zzz | 00:57 | |
| *** metral_zzz is now known as metral | 01:08 | |
| *** alop has quit IRC | 01:32 | |
| openstackgerrit | Miguel Grinberg proposed stackforge/os-ansible-deployment: Fix the heat stack user create https://review.openstack.org/218184 | 01:34 |
| *** darrenc is now known as darrenc_afk | 01:42 | |
| *** sdake_ is now known as sdake | 02:06 | |
| *** darrenc_afk is now known as darrenc | 02:17 | |
| *** sdake_ has joined #openstack-ansible | 02:25 | |
| *** sdake has quit IRC | 02:29 | |
| *** markvoelker has quit IRC | 02:31 | |
| *** sdake has joined #openstack-ansible | 02:32 | |
| *** sdake_ has quit IRC | 02:35 | |
| *** Apsu has quit IRC | 02:37 | |
| *** Apsu has joined #openstack-ansible | 02:37 | |
| *** Apsu has quit IRC | 02:39 | |
| *** Apsu has joined #openstack-ansible | 02:40 | |
| *** shoutm has quit IRC | 02:55 | |
| *** shoutm has joined #openstack-ansible | 03:06 | |
| *** bgmccollum has quit IRC | 03:29 | |
| *** markvoelker has joined #openstack-ansible | 03:31 | |
| *** dolphm has quit IRC | 03:33 | |
| *** sdake_ has joined #openstack-ansible | 03:34 | |
| *** sdake has quit IRC | 03:35 | |
| *** sdake has joined #openstack-ansible | 03:35 | |
| *** markvoelker has quit IRC | 03:36 | |
| openstackgerrit | Kevin Carter proposed stackforge/os-ansible-deployment: Create config_template plugin https://review.openstack.org/220212 | 03:37 |
| *** dolphm has joined #openstack-ansible | 03:38 | |
| *** bgmccollum has joined #openstack-ansible | 03:38 | |
| openstackgerrit | Kevin Carter proposed stackforge/os-ansible-deployment: Replaced the copy_update module https://review.openstack.org/216790 | 03:38 |
| openstackgerrit | Kevin Carter proposed stackforge/os-ansible-deployment: Adds the config_template to ceilometer https://review.openstack.org/217030 | 03:39 |
| *** sdake_ has quit IRC | 03:39 | |
| openstackgerrit | Kevin Carter proposed stackforge/os-ansible-deployment: Create config_template plugin https://review.openstack.org/220212 | 03:41 |
| coolj | <3 cloudnull | 03:42 |
| cloudnull | say what ? | 03:42 |
| coolj | ^ comfig_template proposals | 03:42 |
| coolj | config even | 03:42 |
| cloudnull | ha. ya, hopfully coming soon. | 03:45 |
| openstackgerrit | Kevin Carter proposed stackforge/os-ansible-deployment: Replaced the copy_update module https://review.openstack.org/216790 | 03:45 |
| cloudnull | if you have some spare cycles I'd love some review on that | 03:45 |
| coolj | relatives coming to town and am gonna be out a couple days, but i might have some time at night. i would love to check it out | 03:46 |
| cloudnull | please do and feel free to ping me if you have questions. | 03:52 |
| coolj | cool, will do! | 03:58 |
| openstackgerrit | Kevin Carter proposed stackforge/os-ansible-deployment: Replaced the copy_update module https://review.openstack.org/216790 | 04:00 |
| openstackgerrit | Kevin Carter proposed stackforge/os-ansible-deployment: Replaced the copy_update module https://review.openstack.org/216790 | 04:02 |
| openstackgerrit | Kevin Carter proposed stackforge/os-ansible-deployment: Adds the config_template to OpenStack Services https://review.openstack.org/217030 | 04:09 |
| openstackgerrit | Kevin Carter proposed stackforge/os-ansible-deployment: Added in keystone reserved port https://review.openstack.org/196702 | 04:17 |
| *** shoutm has quit IRC | 05:01 | |
| *** shoutm has joined #openstack-ansible | 05:05 | |
| *** tlian2 has quit IRC | 05:12 | |
| *** shoutm has quit IRC | 05:22 | |
| *** markvoelker has joined #openstack-ansible | 05:32 | |
| *** markvoelker has quit IRC | 05:37 | |
| openstackgerrit | Miguel Grinberg proposed stackforge/os-ansible-deployment: Fix the heat stack user create https://review.openstack.org/218184 | 05:56 |
| *** shoutm has joined #openstack-ansible | 06:13 | |
| openstackgerrit | Merged stackforge/os-ansible-deployment: Changed certificate order for pem generation with CA files https://review.openstack.org/221361 | 06:19 |
| openstackgerrit | Merged stackforge/os-ansible-deployment: Set the Ubuntu mirror used based on the environment https://review.openstack.org/218611 | 06:19 |
| *** cristicalin has joined #openstack-ansible | 06:24 | |
| *** javeriak has joined #openstack-ansible | 06:39 | |
| *** javeriak has quit IRC | 06:40 | |
| *** shausy has joined #openstack-ansible | 06:55 | |
| *** javeriak has joined #openstack-ansible | 07:09 | |
| *** sdake_ has joined #openstack-ansible | 07:23 | |
| *** javeriak has quit IRC | 07:25 | |
| *** sdake has quit IRC | 07:27 | |
| *** shoutm has quit IRC | 07:28 | |
| *** shoutm has joined #openstack-ansible | 07:29 | |
| *** shoutm_ has joined #openstack-ansible | 07:31 | |
| *** shoutm has quit IRC | 07:33 | |
| *** markvoelker has joined #openstack-ansible | 07:33 | |
| *** sdake has joined #openstack-ansible | 07:36 | |
| *** markvoelker has quit IRC | 07:38 | |
| *** gparaskevas has joined #openstack-ansible | 07:39 | |
| *** sdake_ has quit IRC | 07:39 | |
| *** javeriak has joined #openstack-ansible | 07:58 | |
| *** shoutm has joined #openstack-ansible | 08:21 | |
| *** shoutm_ has quit IRC | 08:24 | |
| *** javeriak has quit IRC | 08:26 | |
| openstackgerrit | Jesse Pretorius proposed stackforge/os-ansible-deployment: Create config_template plugin https://review.openstack.org/220212 | 08:30 |
| openstackgerrit | Jesse Pretorius proposed stackforge/os-ansible-deployment: Fix of haproxy ssl misconfiguration with nova_console https://review.openstack.org/221386 | 08:30 |
| evrardjp | good morning | 08:30 |
| openstackgerrit | Jesse Pretorius proposed stackforge/os-ansible-deployment: Replaced the copy_update module https://review.openstack.org/216790 | 08:31 |
| evrardjp | I'm deploying a new version of our cloud, but I have weird issues | 08:31 |
| evrardjp | the flush net cache fails for the memcached containers in setup infrastructure | 08:31 |
| evrardjp | if I'm manually connecting to ssh on the hosts with the ipv4 or ipv6, it doesn't seem to be an issue | 08:32 |
| evrardjp | has anyone experienced this? | 08:32 |
| evrardjp | (I even restarted setup-hosts to make sure my containers are fine, no change) | 08:33 |
| evrardjp | adding verbosity just show me "ssh error: data could not be sent to the remote host" | 08:33 |
| odyssey4me | morning evrardjp | 08:37 |
| odyssey4me | evrardjp that's not the first time I'm hearing that in the last few days | 08:38 |
| odyssey4me | are you using run-playbooks, or are you running plays directly? | 08:38 |
| evrardjp | setup-* playbooks | 08:38 |
| evrardjp | here it's the setup-infrastructure | 08:39 |
| odyssey4me | ok, is that with or without the repo-sync? | 08:39 |
| evrardjp | I'll check with the first one if it | 08:39 |
| odyssey4me | ie are you using repo-build instead of repo-sync? | 08:39 |
| evrardjp | I'm even to that point | 08:39 |
| odyssey4me | our kilo branch currently does repo-sync, and I think our upstream repo needs an update with the most recent merges | 08:39 |
| evrardjp | I'm at memcached-install | 08:40 |
| evrardjp | I'm not even to that point* | 08:40 |
| openstackgerrit | Jesse Pretorius proposed stackforge/os-ansible-deployment: Shift irqbalance package from lxc_hosts to openstack_hosts https://review.openstack.org/218354 | 08:40 |
| odyssey4me | odd, ok have you set FORKS at all, or are you using the default value? | 08:41 |
| evrardjp | default | 08:41 |
| evrardjp | but | 08:42 |
| evrardjp | I added ssh_delay | 08:42 |
| evrardjp | I'll remove it, just to see what it changes | 08:42 |
| evrardjp | doesn't change anything... It's maybe a network issue at my side, but then I don't understand why I don't have issues connecting with ssh to the host | 08:43 |
| odyssey4me | ssh_delay just changes the amount of time it waits before testing for the ssh connection | 08:43 |
| evrardjp | yeah I removed it, just in case | 08:43 |
| odyssey4me | what did you have it set to? | 08:44 |
| mattt | does anyone want to comment on https://review.openstack.org/#/c/220050/ ? | 08:45 |
| evrardjp | it was 5, but it doesn't change anything | 08:46 |
| evrardjp | mattt: I'll check | 08:46 |
| mattt | if there is enough agreement w/ kevin we may need to unwind the commits in juno/master | 08:46 |
| mattt | evrardjp: thanks! | 08:47 |
| mattt | it _seems_ easier to let global-requirements do the work, but perhaps there may be issues resulting from that | 08:48 |
| evrardjp | mattt: can we rely on these upstream changes? or if not, is there a way to override ? | 08:48 |
| odyssey4me | mattt so in my opinion allowing global requirements to manage it still puts us in a situation where it is a known version - because it's a sha fixed version of global requirements | 08:48 |
| evrardjp | it's true | 08:48 |
| evrardjp | but it doesn't mean that we want to consume exactly the same version | 08:48 |
| mattt | evrardjp: of course there will be situations where global-requirements will be wrong | 08:49 |
| mattt | evrardjp: but we can certainly work around these situations by pinning a specific client | 08:49 |
| odyssey4me | this mechanism puts us in a position where we have to manage less - note that I proposed that mechanism to solve another issue which we got into because we weren't doing this | 08:49 |
| evrardjp | what I don't see in the commit, is the impact on the other containers, like the utility containers | 08:49 |
| odyssey4me | evrardjp the only impact is the version of the client installed | 08:49 |
| evrardjp | I mean, does it work directly? | 08:50 |
| odyssey4me | directly? | 08:50 |
| odyssey4me | when repo-build is run, it collects all the requirements and builds the repo based on the most common versions required | 08:51 |
| mattt | odyssey4me: yeah, i'm not opposed to this, i think the less moving pieces on our part the better | 08:51 |
| odyssey4me | so every client (except in the tempest venv) is the same version | 08:51 |
| evrardjp | mattt: I agree | 08:51 |
| evrardjp | I just think an override of clients could be nice | 08:51 |
| mattt | i just figured it would be good to get more input incase people disagreed with what has been done in juno/master, so we can hold this review going through and unwind the other branches | 08:52 |
| evrardjp | I'll check the repo-build again | 08:52 |
| odyssey4me | evrardjp we can easily override the global requirements to include a specific client or library version if we need to - we do that from time to time to work around upstream issues or to ensure that we have a version of something that includes a fix we need. | 08:53 |
| evrardjp | odyssey4me: yeah, I remember that happened in the past... I just wanted to make sure the opportunity still exists, because it's really good :) | 08:54 |
| mattt | odyssey4me: i've not tested and don't know the repo building stuff well enough, but can you just drop the necessary entries into your user_variables.yml and have that work? | 08:54 |
| evrardjp | mattt: I'm not expert enough to give my review on it... but I guess it's ok | 08:55 |
| mattt | evrardjp: i have a feeling that's not going to work, since the repo building stuff happens inside a container and wouldn't know of those variables outside | 08:55 |
| odyssey4me | mattt no, the repo is built from in-tree values at this stage - we could perhaps include some way of allowing user-based overrides | 08:55 |
| odyssey4me | gotta run for a hair cut - bbiab | 08:56 |
| mattt | njoy | 08:56 |
| evrardjp | it's not that I require this | 08:56 |
| evrardjp | have fun! | 08:56 |
| mattt | evrardjp: it sounds like a handy feature, but perhaps no point crafting solutions until we know it's actually necessary! | 08:57 |
| evrardjp | yeah | 08:57 |
| evrardjp | I remember a time we had a messup with the versions | 08:57 |
| evrardjp | so I thought an override of clients versions is always good | 08:57 |
| evrardjp | I trust the kilo build we are making, so I guess it's not really an issue for me | 08:58 |
| evrardjp | anyway, I'll continue on my issue | 08:58 |
| evrardjp | (sorry to be selfish ;) ) | 08:59 |
| *** markvoelker has joined #openstack-ansible | 09:04 | |
| *** markvoelker has quit IRC | 09:08 | |
| evrardjp | how come the teardown.sh is working fine, while the memcached-install can't even connect to my hosts when doing the flush net cache??? | 09:12 |
| odyssey4me | teardown only deals with the hosts | 09:33 |
| odyssey4me | so the flush net cache is causing connections to hosts to fail? | 09:34 |
| odyssey4me | that's odd, because the flush net cache action is delegated to the host | 09:34 |
| odyssey4me | evrardjp can you please backport https://review.openstack.org/221361 to kilo? | 09:35 |
| openstackgerrit | Jesse Pretorius proposed stackforge/os-ansible-deployment: Set the Ubuntu mirror used based on the environment https://review.openstack.org/221666 | 09:35 |
| evrardjp | if you allow me to merge that one before | 09:36 |
| evrardjp | https://review.openstack.org/#/c/215525/ | 09:36 |
| openstackgerrit | Jean-Philippe Evrard proposed stackforge/os-ansible-deployment: Adds the ability to provide user certificates to HAProxy https://review.openstack.org/221667 | 09:39 |
| odyssey4me | hughsaunders can you please backport https://review.openstack.org/217310 to kilo now that its dependent changes are merged | 09:39 |
| hughsaunders | yep | 09:40 |
| odyssey4me | evrardjp ah, you need https://review.openstack.org/221667 in before https://review.openstack.org/221361 | 09:41 |
| evrardjp | yes, I'm looking again at what you explained me before | 09:41 |
| evrardjp | the dependencies | 09:41 |
| openstackgerrit | Hugh Saunders proposed stackforge/os-ansible-deployment: Add variable for cirros url https://review.openstack.org/221669 | 09:42 |
| evrardjp | let me try it again | 09:42 |
| odyssey4me | if you do 'git review -d 221667', then 'git checkout -b kilo_bug/1493421', then cherry pick -x https://review.openstack.org/221361 then git review that, it'll mark the patches as dependent | 09:42 |
| evrardjp | ok | 09:46 |
| evrardjp | I was there: https://www.mediawiki.org/wiki/Gerrit/Advanced_usage#Create_a_dependency | 09:46 |
| evrardjp | but I have an issue, I can't do the git review -d... FETCH_HEAD being ambiguous | 09:56 |
| evrardjp | it's maybe because 221667 already has a dependency ? | 09:57 |
| odyssey4me | you get an error when doing 'git review -d 221667' ? | 09:58 |
| odyssey4me | or when doing the cherry-pick? | 09:58 |
| evrardjp | when doing the git review | 10:00 |
| evrardjp | should I be on some branch? | 10:00 |
| evrardjp | for example a new branch, before doing the git review -d? | 10:01 |
| odyssey4me | evrardjp nope, it'll effectively do a checkout with all the extra metadata | 10:01 |
| evrardjp | ok | 10:01 |
| evrardjp | that's what I thought | 10:01 |
| odyssey4me | you should probably do a 'git fetch gerrit' first though, or checkout to master and git pull | 10:01 |
| evrardjp | I did a git fetch --all | 10:02 |
| evrardjp | and did that too | 10:03 |
| odyssey4me | evrardjp odd | 10:04 |
| odyssey4me | try a fresh clone to a different folder, run git review -s to setup gerrit, then try again? | 10:04 |
| evrardjp | git pull tells me "already up to date" | 10:04 |
| evrardjp | yeah will try that | 10:05 |
| *** javeriak has joined #openstack-ansible | 10:05 | |
| evrardjp | seems better | 10:06 |
| evrardjp | don't know why | 10:06 |
| evrardjp | no pending changes on my environments | 10:06 |
| openstackgerrit | Jean-Philippe Evrard proposed stackforge/os-ansible-deployment: Changed certificate order for pem generation with CA files https://review.openstack.org/221682 | 10:09 |
| evrardjp | I like gerrit :) | 10:09 |
| odyssey4me | evrardjp it may be that you have another branch with the same change id or something | 10:10 |
| evrardjp | should I kill my old folder? | 10:11 |
| *** arbrandes1 has joined #openstack-ansible | 10:24 | |
| *** logan2 has quit IRC | 10:24 | |
| odyssey4me | evrardjp that's up to you - I do that often :) | 10:25 |
| odyssey4me | fyi evrardjp https://bugs.launchpad.net/openstack-ansible/+bug/1489451 | 10:25 |
| openstack | Launchpad bug 1489451 in openstack-ansible "Not able to connect memcached_container while running setup-infrastructure.yml " [Undecided,New] | 10:25 |
| *** logan2 has joined #openstack-ansible | 10:28 | |
| openstackgerrit | Jesse Pretorius proposed stackforge/os-ansible-deployment: Fix the heat stack user create https://review.openstack.org/218184 | 10:29 |
| *** markvoelker has joined #openstack-ansible | 11:04 | |
| *** markvoelker has quit IRC | 11:09 | |
| openstackgerrit | Matt Thompson proposed stackforge/os-ansible-deployment: Ensure cinder-backup-related variables are defined https://review.openstack.org/221705 | 11:14 |
| evrardjp | odyssey4me: thanks, but I think I found my issue(s) | 11:21 |
| evrardjp | first, the number of forks | 11:21 |
| evrardjp | reduced to 5 it works flawlessly (so far less than the amount of my destination available cpus) | 11:21 |
| evrardjp | second, I've regenerated ssh_keys on each of my physical_hosts | 11:22 |
| evrardjp | (if you follow the documentation, they key generation happen automatically when installing openssh server) | 11:23 |
| evrardjp | third, I've found that my inventory had an issue with used_ips: used_ips doesn't count the host that deploys the cloud ;) | 11:23 |
| odyssey4me | evrardjp ah, nice - perhaps you can add those notes to the bug I pointed out? | 11:26 |
| evrardjp | It looks unrelated... it looks a python issue, mine is just ... my mess up ;) | 11:28 |
| evrardjp | but I could add a comment on the doc | 11:28 |
| odyssey4me | evrardjp I've seen this reported more than once recently, so everyone's experience and solutions can help us figure it out | 11:28 |
| evrardjp | mentionning that you should put the deploy hosts in the used_ips (although this is obvious) could be good in the doc | 11:29 |
| evrardjp | what do you think? | 11:29 |
| odyssey4me | agreed :) | 11:30 |
| evrardjp | odyssey4me: it's not because the result is the same that the cause is the same... so I'm not sure commenting unrelated stuff on the bug would help the others in the future | 11:30 |
| evrardjp | ok I'll write something on the doc | 11:30 |
| odyssey4me | evrardjp so while doing the python thing worked for the bug author, it's not something that others seeing the same issue have been able to replicate - but others are seeing the same result without the same solution | 11:31 |
| evrardjp | yeah, I think we should pay more attention the causes, and mentionning them somewhere... probably in the doc | 11:32 |
| evrardjp | 'known issues' | 11:33 |
| evrardjp | if it's not really known, then it's a bu | 11:33 |
| evrardjp | bug* | 11:33 |
| odyssey4me | good idea | 11:33 |
| evrardjp | new bugs are fine, right? | 11:33 |
| evrardjp | :D | 11:33 |
| *** logan2 has quit IRC | 11:38 | |
| openstackgerrit | Merged stackforge/os-ansible-deployment: Updated the configuration to current dynamic inventory standards https://review.openstack.org/221402 | 11:40 |
| *** logan2 has joined #openstack-ansible | 11:40 | |
| openstackgerrit | Jean-Philippe Evrard proposed stackforge/os-ansible-deployment: Be more clear about used_ips, mostly in the example file https://review.openstack.org/221713 | 11:43 |
| javeriak | hey guys, odyssey4me. I'm about to port these changes to kilo: https://review.openstack.org/#/c/184665/8 . Wanted to ask if there were any other things I should take into consideration since kilo is the main active OSAD branch? | 11:46 |
| mgariepy | good morning everyone ! | 11:47 |
| javeriak | I can see that kilo already provides the capability to enable/disable neutron agent services, I'd like to apply that option to the neutron agent deploying playbooks as well. | 11:47 |
| evrardjp | good morning mgariepy | 11:49 |
| *** logan2 has quit IRC | 11:50 | |
| mgariepy | how are you doing evrardjp ? | 11:50 |
| odyssey4me | javeriak I would suggest that the work be done to get that included in master first. The playbooks and roles are almost the same at this stage - but it's important that any new feature work is done against master first, before backporting to kilo. | 11:51 |
| evrardjp | mgariepy: fine, deploying to prod now, and you? | 11:51 |
| javeriak | odyssey4me, okay sounds good | 11:52 |
| odyssey4me | javeriak the only complexity is that master is running Liberty-3 code, so you might encounter some implementation differences | 11:52 |
| javeriak | odyssey4me, are there any deployment complexities for master? as in can i use that branch to setup a new cluster? | 11:53 |
| mgariepy | i'm doing fine. | 11:53 |
| mgariepy | i have issue with keepalived tho ;) | 11:53 |
| evrardjp | oh? | 11:53 |
| evrardjp | what happened? | 11:53 |
| mgariepy | when a server is in backup mode why does haproxy does have to be stopped ? | 11:54 |
| evrardjp | in order to reduce backend checks | 11:54 |
| odyssey4me | javeriak yes, you can use it to deploy a new cluster - if you find any issues we'd like to know! | 11:54 |
| evrardjp | mgariepy: it's not mandatory | 11:54 |
| evrardjp | you can use another notification script | 11:54 |
| *** logan2 has joined #openstack-ansible | 11:54 | |
| javeriak | cool :). btw do i need to update my relevant BP? | 11:54 |
| odyssey4me | javeriak this is all part of the preparation work to release liberty, so any issues discovered would be great to know (bugs) and great to have patches for :) | 11:54 |
| mgariepy | yeah but if it's stopped the killall -0 haproxy result in faulty state ;) | 11:55 |
| evrardjp | it does, but normally my script checks the state | 11:55 |
| evrardjp | previous state | 11:55 |
| evrardjp | so at some point you can restore from fault state | 11:56 |
| odyssey4me | javeriak considering that your blueprint was implemented some time ago, I think it'd be best to register a new one for the liberty cycle, assuming you'll be able to get the work completed within the next few weeks? | 11:56 |
| evrardjp | and understand it's a backup status | 11:56 |
| evrardjp | but if you prefer using a simpler approach, it would also work... you just have to let haproxy run all the time | 11:56 |
| evrardjp | (which is perfecly fine) | 11:56 |
| javeriak | odyssey4me sure, will do. btw any ETA when kilo becomes rackspace11 officially? | 11:57 |
| odyssey4me | javeriak I expect that you could largely copy the old spec and just update it - this time you won't need to refer to branch names... just focus on the planning to implement it - the decision to backport the feature to kilo is something that comes after it's implemented into master | 11:57 |
| mgariepy | yeah but the issue i have is if my server is backup + haproxy stopped, when it need to switch to master, it fails because haproxy is not running. | 11:57 |
| evrardjp | mgariepy: that shouldn't happen | 11:58 |
| mgariepy | i might be missing something tho.. | 11:58 |
| odyssey4me | javeriak RPC11's release is imminent, so I expect that with your feature it'd come in a minor point release rather than initial release | 11:58 |
| evrardjp | mgariepy: if he needs to switch to master, it checks the old state when transitionning | 11:58 |
| evrardjp | go through faulty, and then immediately master | 11:59 |
| evrardjp | that's what I see here | 11:59 |
| javeriak | odyssey4me, also you guys seem to have taken out the CloudFiles backend setup doc from kilo; are the rackspace_cloud_* parameters still relevant for uservariables.yml ? | 11:59 |
| evrardjp | but I can check with you if you want | 11:59 |
| odyssey4me | javeriak it would be great if you guys could setup some sort of External CI for commit tests to verify whether patches in the project affect the code paths for PlumGrid | 11:59 |
| odyssey4me | javeriak yeah, afaik it's still possible to use them - the docs just need revision | 12:00 |
| odyssey4me | the variable names may be different though | 12:00 |
| evrardjp | my clean install just went from fault (after installing keepalived without haproxy) to master/backup (after the completion of the haproxy playbook) | 12:02 |
| evrardjp | I'm killing now master | 12:02 |
| mgariepy | evrardjp, haproxy is running ? | 12:02 |
| evrardjp | state changed from backup to master | 12:02 |
| evrardjp | mgariepy: it wasn't running on my backup | 12:02 |
| *** markvoelker has joined #openstack-ansible | 12:02 | |
| evrardjp | mmm | 12:03 |
| mgariepy | when i start keepalived, if haproxy is not running it's going into fault state. | 12:03 |
| mgariepy | when i start haproxy it goes to backup | 12:03 |
| javeriak | odyssey4me, yes work should complete latest by two week, its being chased actively. As far as the CI is concerned, yes we've started an internal effort for setting up a jenkins pipeline for OSAD. where its going to be run is being discussed with your folks too | 12:04 |
| mgariepy | if haproxy is stopped , it's stays backup, but when the master dies, it goest to fault. | 12:04 |
| evrardjp | ok will try | 12:04 |
| odyssey4me | javeriak good news :) | 12:05 |
| javeriak | will get back to you guys with some more questions I'm going to have for kilo PLUMgrid support. Our package requirements change slightly for kilo.. | 12:08 |
| evrardjp | mgariepy: sadly someone toyed with my LB server just right now | 12:08 |
| evrardjp | so the only thing I can tell you, is that the backup went to master ;) | 12:08 |
| mgariepy | well i think i found the issue ;) | 12:08 |
| evrardjp | yup? | 12:09 |
| evrardjp | what's that? | 12:09 |
| *** woodard has joined #openstack-ansible | 12:09 | |
| mgariepy | the if statement to check if the state was Fault, is not right ;) | 12:12 |
| mgariepy | if []; then, after this , -ne is supposed to be used to compare int not strings ;) | 12:12 |
| mgariepy | but it's not working anyway:s | 12:13 |
| mgariepy | the only way i have to switch to backup to master is to have haproxy running. | 12:14 |
| *** logan2 has quit IRC | 12:17 | |
| *** woodard has quit IRC | 12:20 | |
| *** logan2 has joined #openstack-ansible | 12:20 | |
| *** woodard has joined #openstack-ansible | 12:22 | |
| *** javeriak has quit IRC | 12:28 | |
| evrardjp | we should maybe simplify this | 12:31 |
| evrardjp | let me get my load balancer back from maintenance and I'll check with you | 12:31 |
| evrardjp | some stuff always happen at the wrong time :/ | 12:31 |
| mgariepy | haha indeed | 12:31 |
| evrardjp | good disk was switched with replacement disk instead of the bad disk :D | 12:33 |
| *** shausy has quit IRC | 12:33 | |
| *** shausy has joined #openstack-ansible | 12:33 | |
| mgariepy | evrardjp, jinja lstrip : http://paste.ubuntu.com/12320472/ | 12:35 |
| mgariepy | this way jinja2 block won't produce whitespace in the rendered file | 12:35 |
| evrardjp | does this change something? | 12:37 |
| mgariepy | not for the behavior , just make the keepalived.conf file readable. | 12:37 |
| evrardjp | or it's because the jinja is ugly? | 12:37 |
| evrardjp | file* | 12:37 |
| evrardjp | yeah ok | 12:37 |
| mgariepy | every " {% if..." will print 2 spaces in the file. | 12:38 |
| odyssey4me | mgariepy you can also do " {%- if ..." | 12:38 |
| *** javeriak has joined #openstack-ansible | 12:38 | |
| odyssey4me | that strips the whitespace to the left | 12:38 |
| mgariepy | odyssey4me, yeah but when you have 25 blocks.. is better a new line in top of the line ;) | 12:39 |
| odyssey4me | mgariepy does it respect the whitespace you actually want though? | 12:39 |
| mgariepy | it only lstrip whitespace in front of block. | 12:39 |
| odyssey4me | good to know :) | 12:40 |
| mgariepy | it the equivalent of putting - in all the blocks. | 12:41 |
| *** logan2 has quit IRC | 12:41 | |
| *** logan2 has joined #openstack-ansible | 12:42 | |
| evrardjp | mgariepy: you were right, haproxy needs to run at all times | 12:44 |
| evrardjp | at least that's what it seems | 12:45 |
| evrardjp | I'll check because I think this was fixed with the state checking | 12:48 |
| evrardjp | it's maybe because we are trying to simulate too fast | 12:48 |
| mgariepy | maybe. | 12:49 |
| mgariepy | the script is triggered only when a check script state changes ? | 12:50 |
| evrardjp | yup | 12:50 |
| evrardjp | what would be best is to have the old state | 12:50 |
| evrardjp | in this notify script | 12:51 |
| evrardjp | instead of only the new state | 12:51 |
| evrardjp | what I did is just a workaround | 12:51 |
| openstackgerrit | Major Hayden proposed stackforge/os-ansible-deployment: Add profiling for Ansible tasks https://review.openstack.org/216849 | 12:52 |
| mgariepy | haproxy_check script should look in the state file and check haproxy only if it is in master state ? | 12:53 |
| mhayden | just rebased and killed the conflict ^^ | 12:53 |
| evrardjp | mgariepy: that would be a solution | 12:55 |
| mgariepy | or another hack haha | 12:55 |
| evrardjp | :) | 12:56 |
| evrardjp | the fact is, I didn't try that someone would kill haproxy with no reason | 12:56 |
| mgariepy | i didn't do it for no reason, juste because when it goes from master to backup the notify script stops it. | 12:57 |
| mgariepy | and haproxy is noisy | 12:57 |
| evrardjp | maybe I'm dreaming but I don't see the status going from backup to FAULT | 13:02 |
| mgariepy | i have it if haproxy is stopped, then i start keepalived, and i kill the master. | 13:03 |
| evrardjp | haproxy is stopped on both? | 13:04 |
| mgariepy | nop | 13:04 |
| evrardjp | just on the backup | 13:04 |
| evrardjp | how do you kill the master? killing keepalived/haproxy/both ? | 13:04 |
| mgariepy | stopping keepalive | 13:05 |
| openstackgerrit | Major Hayden proposed stackforge/os-ansible-deployment: Add profiling for Ansible tasks https://review.openstack.org/216849 | 13:08 |
| *** tlian has joined #openstack-ansible | 13:11 | |
| *** Mudpuppy has joined #openstack-ansible | 13:12 | |
| evrardjp | mgariepy: Ok I've got the issue too | 13:15 |
| mgariepy | ok | 13:15 |
| evrardjp | I thought all the cases were tested | 13:16 |
| evrardjp | apparently this simple one isn't | 13:16 |
| evrardjp | I'll check on that later mgariepy, if you don't mind | 13:20 |
| evrardjp | do you have a deadline? | 13:20 |
| mgariepy | i'll continue poking it. | 13:25 |
| mgariepy | well not for now ;) haha | 13:25 |
| mgariepy | i have another month or so to play around. | 13:25 |
| *** scarlisle has joined #openstack-ansible | 13:31 | |
| *** cloudtrainme has joined #openstack-ansible | 13:41 | |
| cloudnull | morning | 13:41 |
| *** cloudtrainme has quit IRC | 13:41 | |
| mgariepy | good morning cloudnull | 13:42 |
| mattt | cloudnull: regarding https://review.openstack.org/#/c/202822/ | 13:52 |
| mattt | cloudnull: what is the benefit of having a role that has no dependencies ? | 13:52 |
| *** woodard has quit IRC | 13:55 | |
| openstackgerrit | Javeria Khan proposed stackforge/os-ansible-deployment-specs: Add modularize-neutron-liberty specification https://review.openstack.org/221777 | 13:57 |
| *** shoutm has quit IRC | 13:58 | |
| *** woodard has joined #openstack-ansible | 13:59 | |
| *** mrstanwell has joined #openstack-ansible | 14:15 | |
| openstackgerrit | Javeria Khan proposed stackforge/os-ansible-deployment-specs: Add PLUMgrid Neutron Plugin to neutron playbooks https://review.openstack.org/221788 | 14:22 |
| *** sigmavirus24_awa is now known as sigmavirus24 | 14:23 | |
| openstackgerrit | Kevin Carter proposed stackforge/os-ansible-deployment: Updated MariaDB to the new release version (10.0) https://review.openstack.org/178259 | 14:25 |
| cloudnull | mattt: say what ? | 14:26 |
| cloudnull | you want to add deps to the role ? | 14:26 |
| mattt | cloudnull: "There is duplicate code in the various roles however those bits ensure that the role can be independently deployed without the need for additional dependencies." | 14:26 |
| cloudnull | ah. | 14:27 |
| mattt | i was just wondering what the benefit of that is | 14:27 |
| *** javeriak has quit IRC | 14:27 | |
| cloudnull | a role dep is defined as another role. which means to use a given role with deps you must have all of the roles on your deployment host. | 14:28 |
| cloudnull | essentially I want to keep as few roles as possible and not go back to having this https://github.com/stackforge/os-ansible-deployment/tree/juno/rpc_deployment/roles | 14:29 |
| cloudnull | in which we had single purpose roles and magic variables. | 14:29 |
| cloudnull | all of that role nightmare was created in an effort to stay DRY, however with ansible not being code but rather a simple config management system written in YAML staying DRY is not really a thing we can or should do. IMHO | 14:31 |
| mattt | yeah, that was a lot of roles :P | 14:32 |
| cloudnull | and most of them were single purpose roles using the ansible standard lib. we've dropped that structure in favor of this https://github.com/stackforge/os-ansible-deployment/tree/master/playbooks/roles which is far simpler to understand. again IMHO. | 14:36 |
| palendae | Interjecting here, but I like the post-kilo break up | 14:37 |
| palendae | Much more focused, though there is a loss of 'DRY' | 14:37 |
| cloudnull | ++ | 14:38 |
| *** phalmos has joined #openstack-ansible | 14:38 | |
| palendae | Ansible's got some sharp edges if you want to try it as a programming language | 14:38 |
| mattt | palendae: yeah, i've not really noticed before how unDRY we are until i saw this: https://review.openstack.org/#/c/202822/ | 14:38 |
| palendae | Yeah :( | 14:39 |
| cloudnull | ALL if you have a chance today please review the following items https://review.openstack.org/#/q/starredby:cloudnull+status:open,n,z and https://review.openstack.org/#/q/starredby:odyssey4me+status:open,n,z | 14:39 |
| palendae | If we really want to break up into independent roles, it'll have to be not DRY though :( | 14:39 |
| cloudnull | opps https://review.openstack.org/#/q/starredby:"Jesse+Pretorius"+status:open,n,z | 14:40 |
| cloudnull | what palendae said. | 14:40 |
| odyssey4me | well, Ansible lends itself in style to not be DRY - but instead to be simple | 14:41 |
| git-harry | Role dependencies strike me as perfect for not repeating oneself and having independent roles. | 14:41 |
| *** cloudtrainme has joined #openstack-ansible | 14:42 | |
| git-harry | As I understand it roles are for collections of commonly repeated ordered tasks - that sounds DRY to me | 14:42 |
| *** cloudtrainme has quit IRC | 14:43 | |
| cloudnull | git-harry: +1 roles are the ansible interface into attempting to stay DRY | 14:43 |
| cloudnull | however creating a new role to simply create a rabbitmq user and vhosts seems like a waste. | 14:43 |
| cloudnull | should we also create a common role for directory creates and system users? | 14:43 |
| cloudnull | IMO i say no. | 14:44 |
| palendae | cloudnull: Mmm, I haven't looked at this change in depth, but that seems a little crazy on the surface | 14:44 |
| palendae | A role that creates all the common stuff for OSAD/OA is moving away from independent roles | 14:44 |
| palendae | Or those roles would have to try to create the files/dirs on their own anyway | 14:45 |
| openstackgerrit | Major Hayden proposed stackforge/os-ansible-deployment: Install auditd on the host https://review.openstack.org/221805 | 14:45 |
| mattt | so what is the point of a standalone role? i thought it was so i could deploy (for example) glance without needing keystone, etc. | 14:47 |
| palendae | mattt: Yeah. But our project started all inter-connected | 14:47 |
| mattt | i didn't think it was to avoid pulling in dependencies | 14:47 |
| git-harry | cloudnull: Personally I think that this code should actually be part of a separate play. I think the attempt to make individual service roles means trying to for the orchestration layer, by which I mean plays, into roles. For the general concept of putting a few tasks in a role and the including that as a dependency, I think that is a good idea because there is now only one location to manage that config and it means | 14:48 |
| git-harry | s/to for the/to force the/ | 14:49 |
| openstackgerrit | Major Hayden proposed stackforge/os-ansible-deployment: Install auditd on the host https://review.openstack.org/221805 | 14:50 |
| mattt | git-harry: yeah, i tend to agree that perhaps this shouldn't even be in the roles at all ... | 14:50 |
| mattt | git-harry: like the way the stackforge chef stuff operated ... no openstack service role managed db creates, rabbit user creates, etc. ... it was assumed that was handled elsewhere | 14:51 |
| cloudnull | we could move those tasks into the plays and out of the roles. however it would be the same tasks in the various plays which would still not be a standalone role. | 14:51 |
| odyssey4me | if I think of someone wanting to consume, say, the keystone role - would it not be fair to think that when consuming the role that it should do all the things needed to run keystone? | 14:51 |
| odyssey4me | I think this is much more along the lines of where we draw the line | 14:52 |
| cloudnull | however: what odyssey4me said | 14:52 |
| odyssey4me | should that keystone role setup the database server? | 14:52 |
| odyssey4me | should it setup the database on the database server? | 14:53 |
| git-harry | odyssey4me: I think that is a misunderstanding of a role | 14:53 |
| odyssey4me | should it setup the user for itself? | 14:53 |
| openstackgerrit | Merged stackforge/os-ansible-deployment: Fix of haproxy ssl misconfiguration with nova_console https://review.openstack.org/221386 | 14:53 |
| git-harry | I don't thing everything can or should be shoehorned into a role | 14:53 |
| odyssey4me | git-harry agreed, but I think that as a project we need to decide where that line is and stick to it as a standard | 14:53 |
| mattt | odyssey4me: well if we want to make roles standalone then having glance create rabbit users, etc. breaks that design goal | 14:54 |
| git-harry | odyssey4me: I think the 'role' that you want is actually a collection of playbooks and roles | 14:54 |
| odyssey4me | there is a place for the use of dependent roles, group vars, and playbook orchestration - but we need to balance over engineering against ease of use | 14:54 |
| mattt | a concern with having all that boiler plate code through the roles is when we break them up and have different projects consume different roles, they're going to become an absolute mess | 14:55 |
| git-harry | odyssey4me: yes, I think stuffing everything into a single service role is over-engineering and not what ansible was designed to do | 14:55 |
| cloudnull | mattt: i disagree. i think it promotes that goal. I mean to deploy glance you need all those parts and removing them makes it impossible to use without first doing what the role already accomplishes. | 14:56 |
| odyssey4me | it may make sense, for instance, to have a rabbitmq_admin role - then for the glance playbook to consume that role to setup the users/vhosts/etc - then the glance role is built with the assumption that the rabbit stuff is all there | 14:56 |
| cloudnull | git-harry: i disagree there too. | 14:56 |
| mattt | cloudnull: deploying something and something being useable are not necessarily the same thing :) | 14:56 |
| odyssey4me | but I think these are things we need to work on standardising in the Mitaka cycle and it should be a summit discussion | 14:56 |
| cloudnull | mattt: true | 14:56 |
| mattt | yeah i think these are good discussions to have | 14:57 |
| odyssey4me | working within the current conventions for current patches is the best we can do for the delivery of liberty | 14:57 |
| odyssey4me | For the Mitaka cycle we need to have a strong focus on reducing technical debt build up - and standards is one of many ways of doing that | 14:57 |
| mattt | dunno if the ansible guys will be at the summit again, but it would be great to pull them into discussions if at all possible | 14:58 |
| mattt | s/guys/folk/ :) | 14:58 |
| palendae | Honestly we're in a little bit of uncharted territory - most ansible roles and playbooks seem to sit in private repos, not public, as seen with Galaxy's poor adoption | 14:58 |
| cloudnull | git-harry: as defined by ansible "Roles are just automation around ‘include’ directives"..."and really don’t contain much additional magic beyond some improvements to search path handling for referenced files" | 14:58 |
| cloudnull | so in the play or in the role its really the same thing. | 14:59 |
| git-harry | not if the role is making decisions about which host the role applies to | 14:59 |
| cloudnull | of course it is . | 15:00 |
| cloudnull | you can use delegate in play tasks just like you can in a role. | 15:00 |
| git-harry | I'm not thinking about delegate_to here so much as conditionals based on group membership | 15:01 |
| palendae | Didn't think roles were supposed to know about group membership | 15:01 |
| palendae | If not by ansible standards, then ours | 15:01 |
| mattt | i do also think that roles knowing about specific hosts feels slightly wrong too | 15:01 |
| git-harry | I think roles should be like modules from that perspective | 15:02 |
| cloudnull | svg might have some thoughts on that but in reviewing plenty of roles out in the wild many of them have groups defined within them. | 15:02 |
| *** jwagner_away is now known as jwagner | 15:02 | |
| mattt | we ultimately need to agree on some patterns and document them | 15:02 |
| mattt | then enforce them via the review process | 15:02 |
| odyssey4me | mattt ++ | 15:03 |
| odyssey4me | what is or isn't best practise is subjective | 15:03 |
| palendae | Yeah, project standards would help | 15:03 |
| odyssey4me | we need to decide what our patterns and conventions are | 15:03 |
| odyssey4me | then to apply them properly | 15:03 |
| odyssey4me | they can adjust over time as we learn to do betterer, and as ansible gains more features | 15:03 |
| palendae | From what I know of the current state of Ansible 2.0, gaining features will be a while | 15:04 |
| mattt | dangit wish i were going to tokyo now :P | 15:05 |
| palendae | afaik, Ansible 2.0 was "Rewrite it so the code is how we want it, then add support back in for existing stuff" | 15:06 |
| palendae | They're on the latter part now | 15:06 |
| odyssey4me | evrardjp can you backport https://review.openstack.org/221386 and https://review.openstack.org/221402 please :) | 15:06 |
| *** scarlisle has quit IRC | 15:07 | |
| openstackgerrit | Jesse Pretorius proposed stackforge/os-ansible-deployment: Shift irqbalance package from lxc_hosts to openstack_hosts https://review.openstack.org/218354 | 15:08 |
| cloudnull | mattt odyssey4me: I agree that we need to agree on a standard however I think we default to ansible best practices and define our own where needed. http://docs.ansible.com/ansible/playbooks_best_practices.html | 15:09 |
| svg | cloudnull: what's the specific question? | 15:10 |
| cloudnull | svg: whats your stance on using group names within a role. | 15:10 |
| cloudnull | 's/./?/' | 15:11 |
| svg | avoid | 15:11 |
| svg | but thats not always easy | 15:11 |
| openstackgerrit | Jesse Pretorius proposed stackforge/os-ansible-deployment: Add galera_address to all playbooks installing galera_client https://review.openstack.org/221301 | 15:11 |
| svg | I read sth ^^ about include paths - I once wrote some patch to have the full role_path availabla as a var to help with that | 15:12 |
| svg | but yes, that basically sucks | 15:12 |
| svg | the difference between files/ and templates/ is bad, as different modules, lookup plugins etc are not consequent | 15:13 |
| svg | cloudnull: is there a more specific use case you are discussing here? | 15:14 |
| cloudnull | so in our case where a role executes a particular code path on group_name[0] and another on != group_name[0] would the "best practice" be to break the role into multiple roles, move those tasks into the plays, or to currently carry on knowing that the current situation "basically sucks"? | 15:14 |
| cloudnull | svg: being that you've been in the anisble community / ecosystem a lot longer than most of us, your expertise would be appreciated here. | 15:15 |
| svg | hm, that's hard one indeed | 15:16 |
| cloudnull | IMO it would help direct us at the desing summit going into M | 15:16 |
| svg | I think splitting up a role for that would be the baddest thing actually | 15:17 |
| *** alop has joined #openstack-ansible | 15:17 | |
| svg | I'd rather stick to hardcoding then | 15:17 |
| svg | I guess that's also a part where one needs more information from inventory, how hosts relate to parent groups, into the exectution (playbooks/roles) | 15:18 |
| *** richoid has joined #openstack-ansible | 15:19 | |
| palendae | cloudnull: Also worth noting - we've been inconsistent in reaching out to the Ansible community, rather than just the OpenStack one | 15:19 |
| palendae | We have one foot in each really, or should | 15:19 |
| cloudnull | svg: indeed the inventory group membership is then tightly to what the role requires. | 15:19 |
| *** richoid has quit IRC | 15:19 | |
| git-harry | I'm not sure what benefit roles give the project over using playbooks because the roles are so tied to the project they'll never be used anywhere else. | 15:20 |
| svg | btw, I noticed you guys often talk about making rols more generic, putting them in their own repo, publishing hem on galaxy... I'm not sure why that is so important. Especially for very specific deploys like this project. Even the best stuff on galaxy will probably never suit what a project as osad needs. IMHO. | 15:20 |
| svg | heh, so, what git-harry said | 15:20 |
| palendae | svg: I think that comes from trying to accomodate more deployer needs more easily | 15:20 |
| *** richoid has joined #openstack-ansible | 15:20 | |
| svg | (I never was a big fan of galaxy) | 15:20 |
| palendae | Instead of having extension points all over in this big repo | 15:21 |
| palendae | But I'm starting to wonder the same | 15:21 |
| palendae | The concept, or the implementation? | 15:21 |
| odyssey4me | yeah, so the general idea is that if someone wants to just deploy keystone and swift then they can only pull those roles down (and the infra roles they need to make it go) | 15:21 |
| odyssey4me | it's not necessarily to tie it to galaxy itself, but rather to allow more selective use of the roles we work on | 15:22 |
| *** gparaskevas has quit IRC | 15:23 | |
| odyssey4me | it also makes it more obvious for someone who wants to put together their own playbooks instead of using the ones we have, which are very specifically tied to the architecture we use - someone could deploy using whatever architecture they prefer | 15:23 |
| cloudnull | git-harry: thats not true. presently there are teams using our os_keystone and galera_server roles as standalone for deployments on baremetal. which we know works because they're testing it right now. so having the roles gives people like that an inteface to use our things without deploying OSAD specifically. | 15:24 |
| odyssey4me | technically that's possible now, but it's not obvious that it's possible | 15:24 |
| palendae | cloudnull: Can you say which teams? It'd be great to have them in here and discussing their experience | 15:24 |
| jbweber_ | outsider perspective, we use ansible to deploy among other things openstack and ceph and tend to never be able to reuse directly anyones roles mostly because of operating specific choices or environment issues. the way stuff is setup right now even with the split you're discussing it still wouldn't be directly usable | 15:24 |
| richoid | hey all - any plans on multi region compute being prioritized? | 15:24 |
| cloudnull | lbragstad: and crew | 15:24 |
| palendae | jbweber_: Yeah, I think that's what we're going to find | 15:25 |
| jbweber_ | our stuff is osad based (kinda) but uses RDO + RHEL instead of source and ubuntu | 15:25 |
| palendae | Making them generic is going to be a long and painful process. That's mostly conjecture, though, since I've not tried | 15:25 |
| palendae | I'd love to be wrong! | 15:25 |
| odyssey4me | jbweber_ right now they're heavily tied into the architecture - but the idea of breaking them out is to be more disciplined at making them more reusable outside of the same architecture | 15:25 |
| cloudnull | palendae: ++ | 15:25 |
| cloudnull | long and painful. :) | 15:26 |
| jbweber_ | I've been lurking because I've been wondering if the coming changes will make anything easier for us or not and we can dump our custom stuff. So this is an interesting topic to me | 15:26 |
| odyssey4me | jbweber_ we'd be very interested in more collaboration around making it more OS independent | 15:27 |
| git-harry | cloudnull: my point is that by tying the roles to the inventory etc the project forces the roles consumers to use absorb a some of osad's structure in the same way that consuming playbooks would. Therefore putting the code in roles doesn't provide any additional benefit. | 15:27 |
| odyssey4me | some research has already been done which can make it less painful with very little adjustment... at least for packages and templates | 15:28 |
| git-harry | and I think is a detriment because playbooks aren't made best use of | 15:28 |
| svg | I think roles should do one thing and to it well. If they become too bug or too complex, then something is wrong. | 15:29 |
| svg | e.g. if one has a role to deploy a webapp, don't code things into it that deploy and configure the $httpd for it | 15:30 |
| svg | ideally, a consumer should be able to choose his $httpd of choice | 15:30 |
| odyssey4me | palendae cloudnull while I agree that breaking out the current roles may be a long and painful process, agreeing that independent role repositories can form part of the openstack-ansible umbrella would be a great first step to being more inclusive of work that others have done for projects not currently included in the architecture | 15:31 |
| palendae | IIS | 15:31 |
| cloudnull | palendae: this is a serious converstaion ;) | 15:31 |
| palendae | cloudnull: I'm switching to PowerShell DSC | 15:31 |
| odyssey4me | if we can agree that it's ok for there to be more role repositories, then others can independently put together roles for optional use and we don't have to be tied into the review process for them if we don't want to be | 15:31 |
| *** arbrandes1 has quit IRC | 15:31 | |
| *** phalmos has quit IRC | 15:32 | |
| * svg installs Windows 10 on palendae laptop | 15:33 | |
| * cloudnull takes ball and goes home... | 15:33 | |
| openstackgerrit | Hugh Saunders proposed stackforge/os-ansible-deployment: Install nfs-common with nova-compute https://review.openstack.org/221844 | 15:33 |
| odyssey4me | svg agreed - for example we have several roles deploying apache for openstack services... we could probably break that out into a role and have the openstack service playbooks do the apache bits | 15:33 |
| palendae | svg: Oh, you're not aware of my laptop shopping righ tnow.. | 15:33 |
| git-harry | odyssey4me: I do agree with that concept. The project is sort of reversing big tent if a project needs to be in osad to be deployed using it | 15:33 |
| palendae | odyssey4me: It all sounds good in theory, sure | 15:33 |
| svg | palendae: considering a tablet? | 15:33 |
| palendae | svg: Upgrading my current X1 Carbon to be a Linux/Windows workstation for work | 15:34 |
| palendae | We'll see though | 15:34 |
| svg | palendae: in what sense 'upgrading'? | 15:34 |
| cloudnull | git-harry: +1 I agree. | 15:34 |
| palendae | svg: I have an X1 with 8GB RAM/256GB SSD. Would double those | 15:35 |
| * cloudnull runs knowing that lightning is about to strike due to git-harry and cloudnull agreeing. | 15:35 | |
| svg | palendae: I thought RAM upgrade was not possible with the X1? of perhaps that's an older model? | 15:35 |
| odyssey4me | git-harry cloudnull that's why https://review.openstack.org/213779 was proposed - to allow independent repositories and to consider breaking out the existing roles | 15:36 |
| *** skamithi13 has quit IRC | 15:36 | |
| palendae | svg: No, I'd be stupid and buy a whole new machine :) | 15:36 |
| *** skamithi13 has joined #openstack-ansible | 15:36 | |
| palendae | When I got mine 8GB was the highest option | 15:36 |
| *** sdake_ has joined #openstack-ansible | 15:36 | |
| odyssey4me | the existing roles would still require their own blueprint/spec as it's a complex affair | 15:36 |
| cloudnull | palendae: if you get another machine and end up selling the x1 let me know :) | 15:36 |
| palendae | cloudnull: You may be in luck! | 15:36 |
| svg | same was with my x220, officially no support for more than 8g, but doubling it just works | 15:36 |
| cloudnull | i have an x230 which i love | 15:37 |
| cloudnull | but the res is kinda a drag | 15:37 |
| palendae | That x230 is nice | 15:37 |
| palendae | I just replaced my Thunderbolt monitor with a Dell 27" 4k one | 15:37 |
| palendae | So I can finally use something other than macbooks on my large screen | 15:37 |
| openstackgerrit | Jesse Pretorius proposed stackforge/os-ansible-deployment: Fix keystone shibboleth template to detect ssl usage correctly https://review.openstack.org/221347 | 15:38 |
| *** arbrandes has joined #openstack-ansible | 15:38 | |
| *** sdake__ has joined #openstack-ansible | 15:39 | |
| *** sdake has quit IRC | 15:40 | |
| openstackgerrit | Jesse Pretorius proposed stackforge/os-ansible-deployment: Add profiling for Ansible tasks https://review.openstack.org/216849 | 15:40 |
| *** sdake_ has quit IRC | 15:43 | |
| *** phalmos has joined #openstack-ansible | 15:45 | |
| *** shoutm has joined #openstack-ansible | 15:46 | |
| *** sdake has joined #openstack-ansible | 15:47 | |
| evrardjp | too bad I missed the nice discussion | 15:48 |
| evrardjp | I'll give my opinion anyway :p | 15:49 |
| *** sdake__ has quit IRC | 15:49 | |
| *** cristicalin has quit IRC | 15:49 | |
| evrardjp | I think that separating in different independant roles is good. It doesn't force the deployer to use the inventory as it currently is. (if the deployer has other dynamic inventories, it's far more helpful to not impose him that) | 15:49 |
| evrardjp | however the orchestration of roles is difficult | 15:50 |
| evrardjp | but that's why there are tools in ansible that can help us: roles dependencies, tags for skipping parts of a role ... | 15:50 |
| openstackgerrit | Matt Thompson proposed stackforge/os-ansible-deployment: Allow testing of cinder multi backends w/ tempest https://review.openstack.org/221851 | 15:51 |
| palendae | evrardjp: The inventory thing is orthagonal, I think - it needs to be revisited, too | 15:51 |
| openstackgerrit | Jean-Philippe Evrard proposed stackforge/os-ansible-deployment: Fix of haproxy ssl misconfiguration with nova_console https://review.openstack.org/221853 | 15:51 |
| openstackgerrit | Matthew Thode proposed stackforge/os-ansible-deployment-specs: Add spec for IPv6 support for projects https://review.openstack.org/221516 | 15:52 |
| openstackgerrit | Jean-Philippe Evrard proposed stackforge/os-ansible-deployment: Updated the configuration to current dynamic inventory standards https://review.openstack.org/221854 | 15:52 |
| evrardjp | palendae: I don't know what you mean by that, but I understand that it's also a keystone (pun intended) of the ansible deployment | 15:52 |
| palendae | evrardjp: It is, but our in-tree dynamic inventory script is kind of a snowflake compared to most ansible ones, afaik | 15:53 |
| evrardjp | we are using dynamic inventory here | 15:53 |
| palendae | It uses the /etc/*_deploy/*_user_config.yml file to generate and populate container IPs and names | 15:53 |
| evrardjp | so it's not a standard inventory either | 15:53 |
| evrardjp | palendae: yeah I know how it works :) | 15:54 |
| evrardjp | more or less | 15:54 |
| odyssey4me | evrardjp you probably know better than I do at this point :) | 15:55 |
| cloudnull | I would like inventory and the generation of inventory to be better. i just cant put my finger on how yet. | 15:55 |
| evrardjp | I don't think it needs to be changed | 15:55 |
| palendae | evrardjp: Yeah - for most of the upstream examples from Ansible don't have the dynamic inventory scripts generate inventory, just query for it | 15:55 |
| evrardjp | right now | 15:55 |
| palendae | I think it needs to be refactored and have some tests wrapping it | 15:55 |
| evrardjp | I think the focus would be to be as modular as possible, to use other inventories, and still deploy without problems | 15:55 |
| palendae | Lots of things to be done, just time :( | 15:56 |
| evrardjp | palendae: in upstream example that's often what happens: json generation, caching and reuse | 15:57 |
| evrardjp | in upstream examples for dynamic inventories* | 15:57 |
| palendae | evrardjp: Right, but (again, afaik) they don't then create the machines | 15:57 |
| palendae | Our LXC vs AWS, from what I could tell was the AWS ones simply use what's in AWS's db | 15:57 |
| palendae | Maybe this is all ok and i'm chasing ghosts | 15:57 |
| *** sdake has quit IRC | 15:58 | |
| cloudnull | everyone should adopt the standard Venus time. then there wouldn't be an issue with time. | 15:58 |
| evrardjp | lxc is just a tool, a convenient one | 15:58 |
| odyssey4me | as far as I can tell from the little bit of reading I've worked through, everything we currently do in our inventory can be expressed in the ini format... it's standard ansible | 15:59 |
| lbragstad | cloudnull: o/ | 15:59 |
| evrardjp | creating a lxc container just bring something to your inventory, that's just something you could skip if not using lxc | 15:59 |
| palendae | odyssey4me: Yeah | 15:59 |
| palendae | evrardjp: Sure | 15:59 |
| palendae | I'm probably chasing ghsots then :) | 15:59 |
| *** Mudpuppy has quit IRC | 15:59 | |
| odyssey4me | the only trouble we have with it is that the way we put it together is a bit obscure | 15:59 |
| cloudnull | o/ lbragstad | 15:59 |
| palendae | odyssey4me: I would like to move our defaults out of the scripts themselves and into an ini file in /etc/ansible | 15:59 |
| palendae | odyssey4me: Yes | 15:59 |
| lbragstad | cloudnull: you pinged earlier? | 15:59 |
| odyssey4me | so you're right palendae - we should document how it's put together and have some tests on it | 16:00 |
| evrardjp | I'm mean, yes it's not only reading from inventory source A, sometimes you should also write to an inventory ;) | 16:00 |
| *** Mudpuppy has joined #openstack-ansible | 16:00 | |
| cloudnull | just talking about the use of the os_keystone roles. | 16:00 |
| palendae | lbragstad: He pinged cause I was asking about groups using independent roles | 16:00 |
| palendae | Evidently you guys are | 16:00 |
| lbragstad | palendae: ahh | 16:00 |
| lbragstad | palendae: independent roles? | 16:01 |
| palendae | Would love feedback on what works and doesn't as we try to figure out separating our repo | 16:01 |
| palendae | lbragstad: So we understood that you're 'just' using the os_keystone role? | 16:01 |
| odyssey4me | palendae well, it would be ideal if we could move away from using -e overrides, and instead use group vars better | 16:01 |
| lbragstad | palendae: oh, yes! sorry I was thinking groups and roles in the terms of keystone entities | 16:01 |
| palendae | odyssey4me: That'd be good too | 16:01 |
| *** alextricity has joined #openstack-ansible | 16:01 | |
| evrardjp | odyssey4me: +1 | 16:01 |
| palendae | odyssey4me: Like I said, lots and lots of room for improvement :) | 16:01 |
| evrardjp | group_vars are powerfulz | 16:02 |
| evrardjp | ansible variable merging would also be great if it could come from the ansible guys | 16:02 |
| palendae | lbragstad: It was really more of a 'we should sit down sometime and get focused feedback', nothing to solve right now | 16:02 |
| evrardjp | (on a task/play basis) | 16:02 |
| palendae | lbragstad: Maybe at summit if you can make time | 16:02 |
| lbragstad | palendae: we are using the standalone keystone stuff the cloudnull pushed to his repository. brianl or blewis is the guy who is working one that from our team. | 16:03 |
| palendae | Ahh, so the os_cloud stuff? | 16:03 |
| palendae | Or os-cloud | 16:03 |
| *** skamithi14 has joined #openstack-ansible | 16:03 | |
| cloudnull | palendae: test play for stand alone keystone https://github.com/cloudnull/osad_standalone_keystone | 16:04 |
| lbragstad | palendae: yep https://github.com/cloudnull/osad_standalone_keystone | 16:04 |
| *** skamithi13 has quit IRC | 16:04 | |
| cloudnull | that one :) | 16:04 |
| odyssey4me | lbragstad what held you guys back from using the osad repo as it is? | 16:04 |
| lbragstad | odyssey4me: we did, i spent a weekend diving into it and we deployed keystone with it | 16:04 |
| *** skamithi13 has joined #openstack-ansible | 16:04 | |
| cloudnull | richoid: sorry totally missed the message around multi region compute. . | 16:05 |
| *** skamithi15 has joined #openstack-ansible | 16:05 | |
| evrardjp | quick question, not related, any doc of the complete list of the variables I should set to bring all the service endpoints to SSL (and work with ssl, not just have https:// in its url)? | 16:05 |
| cloudnull | richoid: we had a spec for this https://review.openstack.org/#/c/192421/ | 16:06 |
| cloudnull | but it was abandoned due to priorities shifting to federation and other things | 16:06 |
| cloudnull | if thats something that you'd be interested in we can revive the spec | 16:06 |
| cloudnull | ideally we'd need some input on making it go. | 16:07 |
| cloudnull | technically multi region work, you simply have to define variables to make it work but shared infra pieces need to be updated to better accomodate the deployment process. | 16:08 |
| *** skamithi14 has quit IRC | 16:08 | |
| *** phalmos has quit IRC | 16:08 | |
| cloudnull | heres an example of making multi-region work with kilo https://github.com/cloudnull/osad-regions | 16:08 |
| cloudnull | simple test play but it does the job of illustrating how it works. | 16:09 |
| *** skamithi13 has quit IRC | 16:09 | |
| odyssey4me | richoid cloudnull yeah, most of the work is already done - the only part you can't do right now is use the same ansible inventory to deploy a seperate global galera cluster | 16:09 |
| odyssey4me | if you build your global galera cluster seperately, the rest is just variable changes | 16:10 |
| odyssey4me | evrardjp documented - no, not to my knowledge | 16:10 |
| lbragstad | odyssey4me: we got it to work fine with the osad-keystone role but cloudnull pretty much did the same thing we did, just better. | 16:10 |
| odyssey4me | evrardjp assuming that you have your LB setup to do the ssl offloading and don't want to use ssl internally, it should just be a case of modifying the public endpoints in the service catalogues after you've deployed | 16:11 |
| odyssey4me | evrardjp if you actually want to have all the services use ssl endpoints internally, I'm not sure we have all the bits there to make that go | 16:12 |
| evrardjp | that's what I thought odyssey4me | 16:12 |
| cloudnull | evrardjp: at last check not all of the OS services correctly support ssl | 16:12 |
| evrardjp | Maybe I should focus on improving haproxy to use ssl | 16:12 |
| evrardjp | for those services | 16:12 |
| odyssey4me | evrardjp only keystone is instrumented for that, primarily because I don't think anyone has thought that anything other than identity traffic securing is important | 16:13 |
| evrardjp | to terminate* | 16:13 |
| *** kukacz has joined #openstack-ansible | 16:13 | |
| evrardjp | I think horizon already works with https too | 16:13 |
| odyssey4me | yeah, another reason is that much of the time when you try to make clients use ssl, then break | 16:13 |
| evrardjp | in our playbooks | 16:13 |
| odyssey4me | *they | 16:13 |
| odyssey4me | ah yes, horizon does work too | 16:13 |
| evrardjp | that's my issue | 16:13 |
| evrardjp | I want to give my users an opportunity to use their client | 16:14 |
| evrardjp | but it miserably fails | 16:14 |
| odyssey4me | my general recommendation has been to only do ssl from the outside, so that it only affects the end-users who would occasionally use the CLI (if that's your use-case) | 16:14 |
| odyssey4me | but definitely don't do it on the inside for all services, because then your back-end breaks | 16:14 |
| evrardjp | ok | 16:15 |
| evrardjp | that's also what I thought | 16:15 |
| openstackgerrit | Hugh Saunders proposed stackforge/os-ansible-deployment: Remove auth_url from auth_token config https://review.openstack.org/221869 | 16:21 |
| *** cloudtrainme has joined #openstack-ansible | 16:26 | |
| *** sdake has joined #openstack-ansible | 16:30 | |
| openstackgerrit | Jesse Pretorius proposed stackforge/os-ansible-deployment: Be more clear about used_ips, mostly in the example file https://review.openstack.org/221713 | 16:35 |
| *** CheKoLyN has joined #openstack-ansible | 16:35 | |
| odyssey4me | lbragstad so the idea of cloning the whole osad repo was not a viable option for you guys | 16:36 |
| odyssey4me | ? | 16:37 |
| lbragstad | odyssey4me: i could have been, i was just assisting our ops team with the initial spike to get things moving | 16:39 |
| Sam-I-Am | hughsaunders: ping | 16:41 |
| hughsaunders | Sam-I-Am: sup | 16:41 |
| Sam-I-Am | hughsaunders: i see your auth_url patch | 16:42 |
| Sam-I-Am | for a long time, even though middleware doesnt seem to read it, every service spewed out some crazy warning when it wasnt there | 16:42 |
| Sam-I-Am | are those warnings gone in liberty? | 16:42 |
| hughsaunders | oh :( | 16:42 |
| hughsaunders | not sure, will check the logs once jenkins has gone though | 16:43 |
| Sam-I-Am | ok. if it works i need to incorporate those changes into the upstream install guide | 16:43 |
| hughsaunders | I'll add docimpact in the commit message if the logs are clean | 16:43 |
| richoid | cloudnull odyssey4me thanks - that last bit is what I would care about the most, similar to tricircle | 16:46 |
| cloudnull | ah is tricircle the successor to the cascading project proposed by huawei? | 16:48 |
| richoid | yea | 16:48 |
| cloudnull | I wasnt aware of tricircle. | 16:48 |
| * cloudnull reading | 16:48 | |
| richoid | its very incomplete | 16:48 |
| *** shausy has quit IRC | 16:49 | |
| richoid | what you explained isn't really similar to tricircle though | 16:49 |
| richoid | thanks for link to osad-regions | 16:49 |
| Sam-I-Am | hughsaunders: actually, its not a docimpact for osad... but it sort of is for upstream upstream :) | 16:50 |
| palendae | Sam-I-Am: So for like, somethinig openstack projects depend on? | 16:50 |
| palendae | Or do you mean downstream? | 16:50 |
| Sam-I-Am | palendae: there's no docs for keystone middleware | 16:50 |
| palendae | Oh, ok | 16:50 |
| Sam-I-Am | so i sort of navigate by feel for every release | 16:50 |
| palendae | Yeah, upstream | 16:50 |
| odyssey4me | richoid so it is possible to deploy portions of osad using a seperate inventory - eg: you could deploy galera on a special cluster for a globally synchronised db, then each localised compute cluster would simply refer to that db for keystone (or whichever services you want a globally synced db for) | 16:51 |
| Sam-I-Am | jamie lennox usually updates the gate jobs with the new config directives, but doesnt document them anywhere else | 16:51 |
| Sam-I-Am | so around this time in the cycle i'll poke around a bit and often just try stuff | 16:51 |
| Sam-I-Am | sometimes i fix things upstream and then apply those changes to osad... other times it goes the other way | 16:52 |
| Sam-I-Am | depends on who catches what when | 16:53 |
| odyssey4me | richoid you can then set the galera_wsrep_provider_options to include gmcast.segment, etc in the host vars | 16:53 |
| Sam-I-Am | i still need to make some changes to osad for neutron | 16:54 |
| Sam-I-Am | they finally solidified the mystery around config files | 16:54 |
| richoid | odyssey4me thanks - as an end user it's difficult to know those possibilities | 16:55 |
| odyssey4me | richoid yep - unfortunately we never quite got down to documenting this stuff because the focal shift of the work switched to federation | 16:56 |
| richoid | I guess I should dig into each role configs | 16:56 |
| richoid | I wouldn't have known to do that either, not a galera expert :) | 16:57 |
| odyssey4me | richoid nor me, but in doing the research that's what I found out - along with some other settings we never did get to test | 17:00 |
| odyssey4me | it's highly likely that the multi-region compute blueprint will get attention again - we just need people to spend time on it | 17:00 |
| odyssey4me | time enough to lab it, and document it - that's pretty much it | 17:01 |
| *** javeriak has joined #openstack-ansible | 17:07 | |
| *** cfarquhar has joined #openstack-ansible | 17:08 | |
| *** cfarquhar has quit IRC | 17:08 | |
| *** cfarquhar has joined #openstack-ansible | 17:08 | |
| richoid | understood | 17:18 |
| openstackgerrit | Javeria Khan proposed stackforge/os-ansible-deployment-specs: Add PLUMgrid Neutron Plugin to neutron playbooks https://review.openstack.org/221788 | 17:21 |
| *** skamithi15 has quit IRC | 17:22 | |
| *** skamithi13 has joined #openstack-ansible | 17:22 | |
| openstackgerrit | Javeria Khan proposed stackforge/os-ansible-deployment-specs: Add modularize-neutron-liberty specification https://review.openstack.org/221777 | 17:22 |
| *** javeriak has quit IRC | 17:28 | |
| *** javeriak has joined #openstack-ansible | 17:31 | |
| *** devlaps has joined #openstack-ansible | 17:32 | |
| *** abitha has joined #openstack-ansible | 17:37 | |
| *** sdake_ has joined #openstack-ansible | 17:37 | |
| *** sdake has quit IRC | 17:41 | |
| *** sdake_ is now known as sdake | 17:45 | |
| openstackgerrit | Kevin Carter proposed stackforge/os-ansible-deployment: Updated MariaDB to the new release version (10.0) https://review.openstack.org/178259 | 17:49 |
| cloudnull | Cores please review the following commits as they're working on completing several of the Liberty approved specs https://review.openstack.org/#/q/starredby:cloudnull+status:open,n,z | 17:53 |
| cloudnull | additionally these items need love too https://review.openstack.org/#/q/starredby:"Jesse+Pretorius"+status:open,n,z | 17:53 |
| *** shoutm has quit IRC | 17:55 | |
| *** agireud has joined #openstack-ansible | 18:08 | |
| openstackgerrit | Kevin Carter proposed stackforge/os-ansible-deployment: Update Ansible to v1.9.3-1 https://review.openstack.org/221899 | 18:14 |
| *** alextricity has quit IRC | 18:23 | |
| *** javeriak_ has joined #openstack-ansible | 18:27 | |
| *** javeriak has quit IRC | 18:30 | |
| *** javeriak_ has quit IRC | 18:32 | |
| *** agireud has quit IRC | 18:36 | |
| *** agireud has joined #openstack-ansible | 18:42 | |
| *** javeriak has joined #openstack-ansible | 18:42 | |
| *** phalmos has joined #openstack-ansible | 18:45 | |
| openstackgerrit | Kevin Carter proposed stackforge/os-ansible-deployment: Updated MariaDB to the new release version (10.0) https://review.openstack.org/178259 | 18:47 |
| *** cloudtrainme has quit IRC | 18:49 | |
| *** arbrandes has quit IRC | 18:50 | |
| *** chachaji has joined #openstack-ansible | 18:50 | |
| openstackgerrit | Kevin Carter proposed stackforge/os-ansible-deployment: Compartmentalizing RabbitMQ https://review.openstack.org/202822 | 18:54 |
| * cloudnull lunching | 18:58 | |
| openstackgerrit | Merged stackforge/os-ansible-deployment: Add variable for cirros url https://review.openstack.org/221669 | 19:01 |
| openstackgerrit | Merged stackforge/os-ansible-deployment: Disable python buffering for gate checks https://review.openstack.org/220083 | 19:03 |
| *** arbrandes has joined #openstack-ansible | 19:07 | |
| *** javeriak has quit IRC | 19:08 | |
| *** javeriak has joined #openstack-ansible | 19:16 | |
| *** javeriak_ has joined #openstack-ansible | 19:33 | |
| openstackgerrit | Major Hayden proposed stackforge/os-ansible-deployment: Enabling log compression by default https://review.openstack.org/221924 | 19:34 |
| *** lkoranda has quit IRC | 19:35 | |
| *** javeriak has quit IRC | 19:36 | |
| *** javeria__ has joined #openstack-ansible | 19:36 | |
| *** javeriak_ has quit IRC | 19:40 | |
| *** lkoranda has joined #openstack-ansible | 19:40 | |
| *** phalmos has quit IRC | 19:40 | |
| *** lkoranda_ has joined #openstack-ansible | 19:41 | |
| *** phalmos has joined #openstack-ansible | 19:42 | |
| *** devlaps has quit IRC | 19:43 | |
| *** lkoranda has quit IRC | 19:45 | |
| *** lkoranda_ has quit IRC | 19:46 | |
| *** phalmos has quit IRC | 19:49 | |
| *** lkoranda has joined #openstack-ansible | 19:50 | |
| *** lkoranda has quit IRC | 19:55 | |
| *** lkoranda has joined #openstack-ansible | 19:59 | |
| *** pabelanger has joined #openstack-ansible | 20:00 | |
| *** arbrandes has quit IRC | 20:09 | |
| *** cloudtrainme has joined #openstack-ansible | 20:10 | |
| *** javeria__ has quit IRC | 20:13 | |
| *** alextricity_h has joined #openstack-ansible | 20:21 | |
| openstackgerrit | Merged stackforge/os-ansible-deployment: Fix the heat stack user create https://review.openstack.org/218184 | 20:24 |
| *** arbrandes has joined #openstack-ansible | 20:25 | |
| *** javeriak has joined #openstack-ansible | 20:30 | |
| *** kukacz has quit IRC | 20:43 | |
| *** woodard has quit IRC | 20:48 | |
| *** javeriak has quit IRC | 20:52 | |
| openstackgerrit | Kevin Carter proposed stackforge/os-ansible-deployment: Set affinity to 1 if running on hpcloud https://review.openstack.org/221957 | 20:54 |
| openstackgerrit | Kevin Carter proposed stackforge/os-ansible-deployment: Fix keystone shibboleth template to detect ssl usage correctly https://review.openstack.org/221347 | 20:57 |
| *** darrenc_ has joined #openstack-ansible | 21:02 | |
| *** darrenc has quit IRC | 21:03 | |
| *** agireud has quit IRC | 21:05 | |
| *** sdake has quit IRC | 21:08 | |
| openstackgerrit | Kevin Carter proposed stackforge/os-ansible-deployment: Archive Keystone to Keystone Federation rst content https://review.openstack.org/211747 | 21:10 |
| *** skamithi13 has quit IRC | 21:16 | |
| *** agireud has joined #openstack-ansible | 21:27 | |
| *** CheKoLyN has quit IRC | 21:36 | |
| openstackgerrit | Miguel Alejandro Cantu proposed stackforge/os-ansible-deployment: Fixes jinja typo in keystone-httpd.conf.j2 https://review.openstack.org/221974 | 21:45 |
| openstackgerrit | Miguel Alejandro Cantu proposed stackforge/os-ansible-deployment: Fixes jinja typo in keystone-httpd.conf.j2 https://review.openstack.org/221974 | 21:47 |
| openstackgerrit | Miguel Alejandro Cantu proposed stackforge/os-ansible-deployment: Fixes jinja typo in keystone-httpd.conf.j2 https://review.openstack.org/221977 | 21:49 |
| *** agireud has quit IRC | 21:50 | |
| *** jwagner is now known as jwagner_away | 21:55 | |
| *** sdake has joined #openstack-ansible | 22:05 | |
| *** galstrom_zzz is now known as galstrom | 22:11 | |
| *** shoutm has joined #openstack-ansible | 22:12 | |
| *** shoutm has quit IRC | 22:12 | |
| *** galstrom is now known as galstrom_zzz | 22:13 | |
| openstackgerrit | Kevin Carter proposed stackforge/os-ansible-deployment: set unique galera server-id https://review.openstack.org/209404 | 22:20 |
| cloudnull | EOD can rattling for reviews - https://review.openstack.org/#/q/starredby:cloudnull+status:open,n,z - Core team/Anyone?! please?! thank you. :) | 22:20 |
| cloudnull | Sam-I-Am: yes this keeps failing . looks like we may still need those vars in liberty https://review.openstack.org/#/c/221869 -cc odyssey4me hughsaunders | 22:21 |
| *** chachaji has quit IRC | 22:33 | |
| cloudnull | Sam-I-Am: https://review.openstack.org/#/c/211747/ this is doc failure if you have a sec can you make it go ? | 22:41 |
| *** darrenc_ is now known as darrenc | 22:51 | |
| *** markvoelker has quit IRC | 22:56 | |
| *** Mudpuppy has quit IRC | 23:11 | |
| Sam-I-Am | moo | 23:14 |
| *** Mudpuppy has joined #openstack-ansible | 23:20 | |
| *** darrenc is now known as darrenc_afk | 23:20 | |
| openstackgerrit | Kevin Carter proposed stackforge/os-ansible-deployment: Fixes jinja typo in keystone-httpd.conf.j2 https://review.openstack.org/221974 | 23:27 |
| *** cloudtrainme has quit IRC | 23:28 | |
| *** Mudpuppy_ has joined #openstack-ansible | 23:31 | |
| *** Mudpuppy has quit IRC | 23:34 | |
| openstackgerrit | Kevin Carter proposed stackforge/os-ansible-deployment: Adds the ability to provide user certificates to HAProxy https://review.openstack.org/221667 | 23:35 |
| *** Mudpuppy_ has quit IRC | 23:42 | |
| *** shoutm has joined #openstack-ansible | 23:44 | |
| *** markvoelker has joined #openstack-ansible | 23:57 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!