| *** woodard has quit IRC | 00:01 | |
| eil397 | evrardjp: interesting, thanks for sharing. | 00:15 |
|---|---|---|
| *** greg_a has quit IRC | 00:19 | |
| *** sacharya has quit IRC | 00:22 | |
| *** markvoelker has quit IRC | 00:26 | |
| *** bangfrog has quit IRC | 00:29 | |
| *** cemmason has quit IRC | 00:40 | |
| *** mss has quit IRC | 00:59 | |
| openstackgerrit | Kevin Carter proposed openstack/openstack-ansible: [WIP] Rewrote the Ansible SSH plugin https://review.openstack.org/249500 | 01:05 |
| *** eil397 has quit IRC | 01:06 | |
| *** mss has joined #openstack-ansible | 01:07 | |
| openstackgerrit | Kevin Carter proposed openstack/openstack-ansible: [WIP] Rewrote the Ansible SSH plugin https://review.openstack.org/249500 | 01:09 |
| *** eil397 has joined #openstack-ansible | 01:11 | |
| *** mss has quit IRC | 01:11 | |
| *** woodard_ has quit IRC | 01:20 | |
| *** markvoelker has joined #openstack-ansible | 01:27 | |
| *** darrenc is now known as darrenc_afk | 01:27 | |
| *** BjoernT has quit IRC | 01:29 | |
| cloudnull | have a good night all | 01:29 |
| eil397 | cloudnull: good night | 01:31 |
| *** markvoelker has quit IRC | 01:31 | |
| *** mss has joined #openstack-ansible | 01:35 | |
| *** mss has quit IRC | 01:39 | |
| *** karimb has quit IRC | 01:40 | |
| *** woodard has joined #openstack-ansible | 01:42 | |
| *** skamithi has joined #openstack-ansible | 01:45 | |
| *** skamithi has left #openstack-ansible | 01:45 | |
| *** eil397 has quit IRC | 01:56 | |
| *** woodard has quit IRC | 01:59 | |
| *** darrenc_afk is now known as darrenc | 02:05 | |
| *** woodard has joined #openstack-ansible | 02:14 | |
| *** woodard has quit IRC | 02:19 | |
| *** jwagner is now known as jwagner_away | 02:25 | |
| *** markvoelker has joined #openstack-ansible | 02:28 | |
| *** markvoelker has quit IRC | 02:33 | |
| *** skamithi has joined #openstack-ansible | 02:43 | |
| *** skamithi has left #openstack-ansible | 02:43 | |
| *** jmccrory has quit IRC | 03:02 | |
| *** jmccrory has joined #openstack-ansible | 03:04 | |
| *** eil397 has joined #openstack-ansible | 03:04 | |
| *** eil397 has quit IRC | 03:10 | |
| *** eil397 has joined #openstack-ansible | 03:11 | |
| *** woodard has joined #openstack-ansible | 03:15 | |
| *** jduhamel has joined #openstack-ansible | 03:19 | |
| *** woodard has quit IRC | 03:19 | |
| *** jduhamel has quit IRC | 03:21 | |
| *** jduhamel has joined #openstack-ansible | 03:22 | |
| *** kjelly has joined #openstack-ansible | 03:24 | |
| *** mss has joined #openstack-ansible | 03:28 | |
| *** markvoelker has joined #openstack-ansible | 03:29 | |
| *** woodard has joined #openstack-ansible | 03:29 | |
| *** woodard has quit IRC | 03:30 | |
| *** darrenc is now known as darrenc_afk | 03:33 | |
| *** markvoelker has quit IRC | 03:34 | |
| *** markvoelker has joined #openstack-ansible | 03:35 | |
| *** sacharya has joined #openstack-ansible | 03:36 | |
| *** darrenc_afk is now known as darrenc | 03:59 | |
| *** mss has quit IRC | 04:01 | |
| *** sacharya has quit IRC | 04:27 | |
| *** sdake has quit IRC | 04:42 | |
| *** sacharya has joined #openstack-ansible | 04:51 | |
| *** jduhamel has quit IRC | 05:12 | |
| openstackgerrit | Matthew Oliver proposed openstack/openstack-ansible: Add statsd configuration for swift https://review.openstack.org/246182 | 05:13 |
| openstackgerrit | Matthew Oliver proposed openstack/openstack-ansible: Add container-reconciler and object-expirer to os-swift https://review.openstack.org/247290 | 05:13 |
| *** baker has joined #openstack-ansible | 05:14 | |
| *** baker_ has joined #openstack-ansible | 05:16 | |
| *** baker has quit IRC | 05:18 | |
| *** shausy has joined #openstack-ansible | 05:21 | |
| *** galstrom_zzz is now known as galstrom | 05:25 | |
| mattoliverau | so I've got my OSA-AIO working and testing change https://review.openstack.org/247290 , however I'm a little stumped, and if statement doesn't evaluate to true when in some printf debugging should, see http://paste.openstack.org/show/479957/ Any thoughts? | 05:26 |
| *** galstrom is now known as galstrom_zzz | 05:41 | |
| *** baker_ has quit IRC | 05:49 | |
| *** baker has joined #openstack-ansible | 05:51 | |
| *** baker has quit IRC | 06:00 | |
| *** phiche has joined #openstack-ansible | 06:39 | |
| *** sdake has joined #openstack-ansible | 06:40 | |
| *** sacharya has quit IRC | 06:43 | |
| *** sdake has quit IRC | 06:43 | |
| *** javeriak has joined #openstack-ansible | 06:58 | |
| *** javeriak has quit IRC | 07:05 | |
| *** phiche has quit IRC | 07:09 | |
| *** phiche has joined #openstack-ansible | 07:16 | |
| *** sdake has joined #openstack-ansible | 07:28 | |
| *** javeriak has joined #openstack-ansible | 07:31 | |
| *** toddnni has quit IRC | 07:38 | |
| *** sdake has quit IRC | 07:38 | |
| *** javeriak has quit IRC | 07:42 | |
| *** javeriak has joined #openstack-ansible | 07:47 | |
| openstackgerrit | Jean-Philippe Evrard proposed openstack/openstack-ansible: Improve message of the teardown.sh https://review.openstack.org/249325 | 07:50 |
| evrardjp | mattoliverau: what are you trying to achieve? | 07:54 |
| evrardjp | define the amount of process a host can/should run? | 07:54 |
| evrardjp | processes* | 07:54 |
| *** javeriak has quit IRC | 07:54 | |
| *** javeriak has joined #openstack-ansible | 07:54 | |
| *** mpavone has joined #openstack-ansible | 07:54 | |
| *** mpavone has quit IRC | 07:55 | |
| evrardjp | I have the feeling you're trying to have a process variable that is 0 for all the swift_hosts and 1 for the others | 07:55 |
| evrardjp | if it's that, there is another way to do it | 07:56 |
| evrardjp | if it's something to be defined live, then there are maybe other ways to do it than inside a template | 07:57 |
| *** javeriak_ has joined #openstack-ansible | 07:57 | |
| *** javeriak has quit IRC | 07:59 | |
| evrardjp | each task is applied per host (I don't know how to explain that) | 08:02 |
| *** sdake has joined #openstack-ansible | 08:09 | |
| *** sacharya has joined #openstack-ansible | 08:14 | |
| *** jamielennox has joined #openstack-ansible | 08:19 | |
| *** sacharya has quit IRC | 08:20 | |
| *** eil397 has quit IRC | 08:24 | |
| *** javeriak_ has quit IRC | 08:25 | |
| *** javeriak has joined #openstack-ansible | 08:27 | |
| *** javeriak has quit IRC | 08:32 | |
| openstackgerrit | Jean-Philippe Evrard proposed openstack/openstack-ansible-apt_package_pinning: [WIP] Tentative to use sudo for role testing https://review.openstack.org/248774 | 08:37 |
| evrardjp | and good morning everyone | 08:39 |
| *** sdake has quit IRC | 08:41 | |
| *** subscope has joined #openstack-ansible | 08:44 | |
| mattt | morning evrardjp | 08:47 |
| *** karimb has joined #openstack-ansible | 08:52 | |
| mattoliverau | evrardjp: to set up the swift-expirer you need to define the number of expirer daemons running and the id of that configuration. So it counts the number of hosts in the group, and sets the number of that particular configuration for that server with a unique index (were it sits in the group[]). | 09:15 |
| evrardjp | I see | 09:16 |
| evrardjp | hello mattt :) | 09:16 |
| evrardjp | the number of expirer daemons is easy, it's groups['whatever'] | length IIRC | 09:17 |
| mattoliverau | The thought was that by the end of the loop processes is the number of hosts in the group and process is a unique indec on each host. the latter found when ansible_hostname == host.. when it does but doesn't seem to be evaulating to true (the if isn't triggered or the loop for that matter) so the if could be a red herring. | 09:17 |
| evrardjp | yeah I understood | 09:18 |
| evrardjp | I'm thinking to add {hostvars[ ]} on your thingy | 09:18 |
| evrardjp | but I didn't got the chance to test | 09:18 |
| mattoliverau | I'm sure I'm probably missing something obvious.. but I'm strugging to find it. So I came to the you guys ;) | 09:19 |
| mattoliverau | evrardjp: you guys know much more then simple ole me. And thoughts and pointers would be useful. I wont play with it again until the morrow (its evening here). | 09:20 |
| mattoliverau | evrardjp: thanks for taking an interest :) | 09:20 |
| evrardjp | cloudnull: would be a better help than me to be honest | 09:20 |
| evrardjp | I'll do a grep if I can find something in a template to avoid the iteritems() | 09:21 |
| evrardjp | I think this has been done in the past here | 09:22 |
| odyssey4me | o/ | 09:22 |
| mattoliverau | evrardjp: thanks, I can pole him tomorrow if worse comes to worse. So no stress, but if you find the reason that would be awesome, oh and yes, if there is a better way to do then great :) | 09:23 |
| evrardjp | do you mond if I send you something and you try it? | 09:24 |
| *** gparaskevas has joined #openstack-ansible | 09:25 | |
| mattoliverau | ahh yeah, but I wont be around for too long | 09:27 |
| odyssey4me | o/ | 09:29 |
| odyssey4me | mattoliverau you're stuck on something? | 09:29 |
| mattoliverau | odyssey4me: o/ | 09:30 |
| odyssey4me | evrardjp I see that https://review.openstack.org/248774 succeeded! | 09:30 |
| evrardjp | yup | 09:30 |
| evrardjp | I need to improve the stuff though | 09:30 |
| evrardjp | but FYI, the inventory was correct :p | 09:31 |
| odyssey4me | evrardjp well, it's a start - thanks for persisting! | 09:31 |
| mattoliverau | odyssey4me: http://paste.openstack.org/show/479957/ <-- in regards to the swift-expirer config, evrardjp is taking a look. | 09:31 |
| odyssey4me | mattoliverau hmm, that's odd | 09:32 |
| odyssey4me | mattoliverau perhaps http://jinja.pocoo.org/docs/dev/templates/#equalto is a better option? | 09:34 |
| mattoliverau | odyssey4me: yeah, I'm sure I'm missing something obvious, the 'if' could be a red herring, cause the processes count isn't incrementing off 0, so might be the for loop failing, though if does print the blah.. so it must be dying post blah and pre increment. | 09:35 |
| mattoliverau | tho the rest of the template is written. | 09:36 |
| odyssey4me | mattoliverau alternatively there may be a type issue - perhaps try var1 | string() == var2 | string() | 09:36 |
| mattoliverau | I actually tried that too :( | 09:36 |
| mattoliverau | processes remained at 0 and process stayed at -1 | 09:37 |
| mattoliverau | unless its simply the set's failing in the loop | 09:37 |
| mattoliverau | I assume you can use set while looping. | 09:37 |
| odyssey4me | mattoliverau I have yet to see that not work - let me pull this review into my AIO ans test it out. | 09:43 |
| odyssey4me | *and | 09:43 |
| mattoliverau | odyssey4me: thanks muchly | 09:44 |
| *** subscope has quit IRC | 09:46 | |
| *** kjelly has quit IRC | 09:50 | |
| odyssey4me | mattoliverau so the issue is that process is never set, right? | 09:52 |
| mattoliverau | yeah, but processes also isn't incremented. So it's like the loop is failing | 09:53 |
| *** bapalm has quit IRC | 09:53 | |
| evrardjp | mattoliverau: I have something working here | 09:54 |
| *** sshen_ has joined #openstack-ansible | 09:54 | |
| *** sshen has quit IRC | 09:55 | |
| *** bapalm has joined #openstack-ansible | 09:55 | |
| evrardjp | simple as that: https://gist.github.com/evrardjp/446df7031463e5d4504c | 09:55 |
| evrardjp | first one will have processid 0, second 1... | 09:55 |
| evrardjp | it's what you were looking for, right? | 09:55 |
| evrardjp | for the total, as it's the same for every host, you should simply use | length | 09:56 |
| odyssey4me | boo, no tagging in the swift roles to scope tasks run - we'll have to fix this | 09:56 |
| odyssey4me | hang on a minute - you're wanting each host in the inventory to be a sequential number, and you want a total | 09:58 |
| odyssey4me | perhaps this could be easier by using the inventory key? | 09:58 |
| evrardjp | yup that's what I thought first | 09:58 |
| odyssey4me | ie swift_hosts[0], [1], [2] | 09:58 |
| odyssey4me | there's likely a way to grab that number | 09:58 |
| evrardjp | like some loop.index somewhere? | 09:59 |
| evrardjp | that's why I think cloudnull would be best for it, inventory stuff | 09:59 |
| odyssey4me | yeah, extract the index number | 10:00 |
| evrardjp | hostvar[host].index? | 10:00 |
| evrardjp | >< | 10:00 |
| evrardjp | I don't know | 10:00 |
| evrardjp | that's why I took the easiest path: redo the loop :p | 10:00 |
| evrardjp | but it's definitely uglier | 10:00 |
| evrardjp | this seem to work too | 10:06 |
| evrardjp | in a template | 10:06 |
| evrardjp | {{ groups['all'].index(inventory_hostname) }} | 10:06 |
| odyssey4me | testing that idea now | 10:08 |
| mattoliverau | Yeah so the processes has to be the number of object-expirer daemons, so yeah the length would work. The process has to be a number below processes (zero based) so for processes of 3 there should be 3 daemons to split the load with process 0,1,2 out there. | 10:08 |
| odyssey4me | evrardjp that seems to work! mattoliverau I'll pastebin the mod now | 10:09 |
| evrardjp | for the processid, you have one solution :) | 10:09 |
| mattoliverau | ugly is fine if it works.. I'm just dumb founded that the loop didn't. Though if you can just pull and index out and also do a length then thats better cause less looping :) | 10:09 |
| odyssey4me | well, actually -very simply: processes = {{ groups['swift_hosts'] | length }} | 10:10 |
| evrardjp | both work :p | 10:10 |
| odyssey4me | process = {{ groups['swift_hosts'].index(inventory_hostname) }} | 10:10 |
| odyssey4me | no loop required | 10:10 |
| evrardjp | odyssey4me: I think there should be jinja magic for processes: we need to remove 1 ;) | 10:10 |
| evrardjp | but the idea is there :p | 10:10 |
| evrardjp | may I delete the gist? | 10:11 |
| odyssey4me | well, the result in this case is processes = 1, process = 0 | 10:11 |
| evrardjp | odyssey4me: for the length it will find 2 instead of max | 10:12 |
| evrardjp | yeah in fact you could use max I think | 10:12 |
| evrardjp | oh no I misunderstood | 10:12 |
| evrardjp | processes = 3 processs = 0,1,2 | 10:13 |
| evrardjp | so it's perfect | 10:13 |
| odyssey4me | with these mods, I'm sure that andymccr will validate it on the multi-node cluster he has up once mattoliverau has finished the changes | 10:13 |
| evrardjp | anyway, plenty of solutions | 10:13 |
| evrardjp | oh is there someone using sublime text here? I'm tired of editing this python code with vim | 10:14 |
| andymccr | yeh i'll test those prs out hopefully today! but otherwise tomorrow | 10:14 |
| mattoliverau | awesome that's even better!! thanks both of you! | 10:14 |
| odyssey4me | mattoliverau thanks to evrardjp with his ninja fingers :) | 10:15 |
| evrardjp | jinja you mean | 10:15 |
| evrardjp | :p | 10:15 |
| odyssey4me | and yeah evrardjp you can delete the gist :) | 10:15 |
| *** subscope has joined #openstack-ansible | 10:16 | |
| mattoliverau | tahnks odyssey4me and evrardjp you both rock! | 10:17 |
| odyssey4me | thank the coffee I just had :p | 10:17 |
| odyssey4me | evrardjp so it appears to me that https://review.openstack.org/249348 is working to resolve the ssh issues - your thoughts? | 10:20 |
| evrardjp | my thoughts? :D | 10:21 |
| evrardjp | I think we should avoid delivering ansible.cfg in all cases, and let the deployer decide... To give him safe defaults, we could use the bootstrap ansible to write the ansible.cfg for him, and putting in it the safe defaults (no pipelining, no ssh_args) | 10:22 |
| *** mss has joined #openstack-ansible | 10:23 | |
| evrardjp | and mention him what would be the best way to improve speed (enabling pipeling/new connection plugins/...) | 10:23 |
| odyssey4me | evrardjp yeah, that'll be much easier once we've split out all the roles as the convention will be to be using the /etc/ansible/ folder for things | 10:23 |
| evrardjp | my personal opinion: I'll stick with paramiko | 10:23 |
| evrardjp | I'm just tired of trying all these options | 10:24 |
| odyssey4me | evrardjp so yeah, I agree that perhaps the bootstrap-ansible can setup a base file if one doesn't already exist | 10:24 |
| evrardjp | paramiko just works | 10:24 |
| evrardjp | odyssey4me: I agree | 10:24 |
| evrardjp | and putting it in /etc/ansible/ IS a good way to do things | 10:24 |
| evrardjp | you can still overwrite on a folder basis, but let the deployer override is the best way | 10:25 |
| evrardjp | and overriding in-tree isn't so good | 10:25 |
| evrardjp | you have my opinion | 10:25 |
| evrardjp | :) | 10:25 |
| odyssey4me | evrardjp agreed, but for now we have to work with previous life choices and we can't change it up without a deprecation cycle | 10:28 |
| odyssey4me | evrardjp I have something I'd like your input on. mattt & hughsaunders if you're around I'd like your input too. | 10:29 |
| evrardjp | I didn't say I don't agree :) | 10:30 |
| openstackgerrit | Matthew Oliver proposed openstack/openstack-ansible: Add statsd configuration for swift https://review.openstack.org/246182 | 10:30 |
| openstackgerrit | Matthew Oliver proposed openstack/openstack-ansible: Add container-reconciler and object-expirer to os-swift https://review.openstack.org/247290 | 10:30 |
| *** javeriak has joined #openstack-ansible | 10:30 | |
| odyssey4me | bgmccollum has discovered that on some instances in RAX cloud (and this may be an issue in other environments too), libvirt doesn't do the cpu mapping in the right way - resulting in the 'no host found' error in tempest which we're also seeing frequently in the gate checks | 10:31 |
| *** mss has quit IRC | 10:31 | |
| mattoliverau | thanks guys ^^ theres the patch chain (expirer, reconiler and statsd), note I haven't actaully tested statsd properly yet but _should_ work | 10:31 |
| odyssey4me | he has determined a solution, which is to add https://github.com/cloud-training/rpc-heat-ansible/blob/master/playbooks/files/usr/share/libvirt/cpu_map.xml#L337-L381 to the libvirt cpu map | 10:31 |
| mattoliverau | And on that note I'm going to go spend time with da fam. thanks again odyssey4me and evrardjp | 10:32 |
| odyssey4me | thanks mattoliverau for the reviews! I'm sure that andymccr will validate soon! | 10:32 |
| odyssey4me | so my thinking is that we have a few choices to solve this | 10:32 |
| evrardjp | mattoliverau: thanks for the patch | 10:32 |
| odyssey4me | 1 - in the AIO bootstrap, add the extra cpu map | 10:32 |
| odyssey4me | 2 - (untested) set nova.conf in the AIO to use a specific cpu type | 10:33 |
| odyssey4me | I expect that 1 will perform better, but 2 is far less fiddly. | 10:33 |
| mattt | i always prefer experimentation via gate AIO first | 10:33 |
| evrardjp | odyssey4me: do you have more about the bug description and the map? | 10:33 |
| evrardjp | more description about the bug* | 10:34 |
| mattt | odyssey4me: can we do a query to find out how many failures we've had in last 30 days due to that, update gate AIO, and then revisit that query in another 30 days? | 10:34 |
| odyssey4me | evrardjp that's all I got - bgmccollum knows more... he sprang it on me late last night, just before I left | 10:34 |
| evrardjp | I don't recall having seen an issue of that sort | 10:34 |
| mattt | yeah i'm not familiar with this one either | 10:34 |
| odyssey4me | mattt yeah, I think we can - let me give that a whirl | 10:34 |
| odyssey4me | see the last CI results in https://review.openstack.org/249348 and you'll see it | 10:35 |
| mattt | another question is, does this change affect other deploys in any way? | 10:35 |
| evrardjp | this is sensitive | 10:35 |
| odyssey4me | specifically http://logs.openstack.org/48/249348/1/check/gate-openstack-ansible-dsvm-commit/62a86e1/console.html#_2015-11-25_03_06_36_574 | 10:35 |
| mattt | yeah that is a pesky error | 10:35 |
| odyssey4me | mattt so I'm not entirely sure whether this has any production effects - and I expect that the updated CPU map may be something valuable to add upstream... I'll chat to bgmccollum later today to see if we can get that done | 10:36 |
| mattt | ok cool! | 10:36 |
| evrardjp | this is worth documenting | 10:37 |
| evrardjp | I think AIO bootstrap is not a bad idea | 10:37 |
| odyssey4me | evrardjp you mean that code is not documentation? ;) | 10:38 |
| evrardjp | I mean some people won't even read code | 10:38 |
| odyssey4me | actually I think that bgmccollum should send out an ML post about it - this very likely affects infra more broadly | 10:38 |
| evrardjp | not sure | 10:38 |
| evrardjp | but it could yeah | 10:39 |
| hughsaunders | odyssey4me: how did you determine from that log, that its a libvirt cpu type issue? | 10:39 |
| evrardjp | when you deploy nova, you're supposed to know what's behind it, and configure appropriately | 10:39 |
| evrardjp | behind = under | 10:40 |
| odyssey4me | hughsaunders I did not. bgmccollum did. | 10:40 |
| hughsaunders | odyssey4me: whoever did, I'm intrigued by the process.. | 10:40 |
| odyssey4me | hughsaunders me too :) but considering that we've often got those errors, if this is a solution then I'm game to try it out | 10:41 |
| evrardjp | I guess if no resource available when it should, and no issue with scheduler would mean an issue with libvirt itself? | 10:41 |
| evrardjp | here it's not that I guess | 10:44 |
| evrardjp | on your link | 10:44 |
| evrardjp | when you check this: | 10:44 |
| evrardjp | http://logs.openstack.org/48/249348/1/check/gate-openstack-ansible-dsvm-commit/62a86e1/logs/aio1_nova_scheduler_container-fa33bd5c/nova-scheduler.log | 10:44 |
| odyssey4me | ah, a disk space issue | 10:45 |
| odyssey4me | good catch evrardjp | 10:45 |
| evrardjp | I really think the error messages could be improved, it's really 1 line of the conductor and a few lines of the scheduler | 10:46 |
| evrardjp | >< | 10:46 |
| *** subscope has quit IRC | 10:47 | |
| *** javeriak has quit IRC | 10:47 | |
| odyssey4me | right, so this is one of those annoying situations where we have different behaviour between cloud providers in infra | 10:47 |
| odyssey4me | but nice catch - let me work on a fix for that | 10:47 |
| evrardjp | thx | 10:47 |
| odyssey4me | mattt hughsaunders in order to resolve other common gate issues, please review https://review.openstack.org/249348 and https://review.openstack.org/248887 | 10:47 |
| hughsaunders | pretty sure those ssh args were added to resolve ssh reliability issues... and now they are being removed for the same reason -_- | 10:49 |
| hughsaunders | https://github.com/openstack/openstack-ansible/commit/75ab3be99f17e3dff3d28a7c1a0d921f4feaa84f | 10:49 |
| odyssey4me | hughsaunders yep, they were an attempt to resolve the ssh connectivity reliability issues... which turned out to be related to forks | 10:50 |
| odyssey4me | forks were the actual issue - it had nothing to do with the ssh args | 10:50 |
| hughsaunders | forks was one of the issues | 10:50 |
| odyssey4me | we allowed the ssh args because we were, at the time, at a loss of options and were willing to try anything | 10:50 |
| odyssey4me | see https://github.com/ansible/ansible/pull/9246 | 10:51 |
| *** javeriak has joined #openstack-ansible | 10:52 | |
| odyssey4me | hughsaunders also bear in mind that it's only the keepalive bits onward that are being disabled, as the defaults include controlmaster and controlpersist: https://github.com/ansible/ansible/blob/v1.9.4-1/lib/ansible/runner/connection_plugins/ssh.py#L65-L72 | 10:58 |
| hughsaunders | odyssey4me: yeah, I approved it after reading thebug | 10:58 |
| odyssey4me | thanks | 10:58 |
| *** hughsaunders sets mode: -o hughsaunders | 10:59 | |
| odyssey4me | what does confuse me is why this appeared recently again after working well for so long | 10:59 |
| mattt | hughsaunders: https://review.openstack.org/#/c/249348/ didn't pass gate check | 11:05 |
| odyssey4me | mattt yeah, but that's due to a disk space issue on an hp instance | 11:06 |
| odyssey4me | I'm working on resolving that now. | 11:06 |
| mattt | odyssey4me: yeah, i just mean he'll have to recheck it before it actually merges | 11:06 |
| odyssey4me | it's already rechecking :) | 11:07 |
| mattt | ah derp | 11:07 |
| odyssey4me | blast, this is a pain | 11:09 |
| odyssey4me | I'm tempted to try and make the AIO work the same way every time - ie use file system backed lxc and loopback mounts for cinder-volume and swap | 11:09 |
| odyssey4me | hmm, that said it seems that I can use some of the lxc vg space ... let me give that a go | 11:11 |
| evrardjp | you have a large enough lxc vg on all the CI nodes? | 11:12 |
| odyssey4me | evrardjp the second disk on the hp nodes is large enough, yes | 11:12 |
| odyssey4me | http://logs.openstack.org/48/249348/1/check/gate-openstack-ansible-dsvm-commit/62a86e1/logs/instance-info/host_info_1448420796.log | 11:12 |
| odyssey4me | I can carve it up a little differently into three partitions - one for cinder-volume (same as now), one for lxc (same as now), one for /opt/ (where the loopbacks are) and one for /var/lib/nova/instances/ | 11:14 |
| *** bangfrog has joined #openstack-ansible | 11:31 | |
| *** javeriak_ has joined #openstack-ansible | 11:33 | |
| *** javeriak has quit IRC | 11:37 | |
| evrardjp | 200/150/the rest for lxc/cinder/nova ? | 11:39 |
| evrardjp | it's a nice space we have :D | 11:40 |
| evrardjp | 3TB for swift | 11:40 |
| evrardjp | if it's real it's cool | 11:40 |
| odyssey4me | haha, the swift disks are loopbacks implemented with sparse files | 11:41 |
| evrardjp | I know | 11:41 |
| evrardjp | it'd be cool | 11:41 |
| evrardjp | sorry for my english :p | 11:41 |
| *** jamielennox is now known as jamielennox|away | 11:42 | |
| odyssey4me | so lxc right now only needs 150, but perhaps 180 is a better point to set it | 11:42 |
| odyssey4me | cinder probably only needs 20 for its testing | 11:42 |
| odyssey4me | nova should also only need 20 or so | 11:42 |
| evrardjp | nova could need more, depending the amount of instances run by the tempest tests, right ? | 11:45 |
| evrardjp | btw how do you configure that? You have a config file you can use for the partitionning of our CI instances? | 11:45 |
| evrardjp | or you have to re-write the partitioning live ? | 11:46 |
| evrardjp | it doesn't change much, just out of curiousity | 11:46 |
| odyssey4me | evrardjp we can't change what we get, so we figure out what we have and partition it here: https://github.com/openstack/openstack-ansible/blob/master/scripts/scripts-library.sh#L77-L123 | 11:51 |
| evrardjp | yep I thought so :p | 11:51 |
| evrardjp | it could be nice to adapt 'though | 11:51 |
| odyssey4me | I think this could be a lot simpler if we use the loopback method consistently | 11:51 |
| odyssey4me | then we can simply look for enough space and partition | 11:52 |
| evrardjp | true | 11:52 |
| odyssey4me | yeah, I think this will be a better way - let me get on that | 11:53 |
| odyssey4me | first, coffee | 11:53 |
| evrardjp | survival first! | 11:54 |
| *** alkari has joined #openstack-ansible | 11:58 | |
| *** jamielennox|away is now known as jamielennox | 12:01 | |
| *** subscope has joined #openstack-ansible | 12:02 | |
| *** jaypipes has joined #openstack-ansible | 12:05 | |
| openstackgerrit | Merged openstack/openstack-ansible: Update juno for new dev work - 20 Nov 2015 https://review.openstack.org/248017 | 12:05 |
| *** jmccrory has quit IRC | 12:12 | |
| openstackgerrit | Merged openstack/openstack-ansible: Add "httpchk /versions" for glance-api haproxy. https://review.openstack.org/248688 | 12:13 |
| *** adac has joined #openstack-ansible | 12:16 | |
| adac | My notebook got the upgrade. I'm now running the openstack ansible installation AIO build | 12:16 |
| adac | :) :) | 12:16 |
| *** sacharya has joined #openstack-ansible | 12:17 | |
| adac | I was wondering now, when starrting this line, everything gets setup and probably started: | 12:17 |
| adac | curl https://raw.githubusercontent.com/openstack/openstack-ansible/master/scripts/run-aio-build.sh | sudo bash | 12:17 |
| adac | but how would I restart the components afterwards? | 12:17 |
| odyssey4me | adac after rebooting, or after an interruption? which components? | 12:19 |
| adac | odyssey4me, well lets say after a reboot yes | 12:19 |
| odyssey4me | adac well, everything should restart on its own - but your MariaDB cluster will be borked because all cluster hosts were down. | 12:19 |
| *** jmccrory has joined #openstack-ansible | 12:20 | |
| odyssey4me | adac as per http://docs.openstack.org/developer/openstack-ansible/install-guide/ops-galera-recoverycomplete.html execute: ansible galera_container -m shell -a "cat /var/lib/mysql/grastate.dat" | 12:21 |
| *** sacharya has quit IRC | 12:21 | |
| odyssey4me | check if the seqno is -1 on all nodes | 12:21 |
| openstackgerrit | Merged openstack/openstack-ansible: Update Kilo SHA's - 24.11.2015 https://review.openstack.org/245968 | 12:21 |
| openstackgerrit | Merged openstack/openstack-ansible: Update some steps in quickstart-aio.rst https://review.openstack.org/248567 | 12:21 |
| adac | odyssey4me, ah thanks a lot | 12:22 |
| adac | odyssey4me, why is mariadb installed a s a cluster in first place? I mean it is a standalone installation, right? | 12:23 |
| adac | AIO | 12:24 |
| odyssey4me | adac as the AIO is used for dev/testing we try to set it up as close to a multinode deployment as possible - just using a single host | 12:24 |
| odyssey4me | this way you get to test clustered infrastructure too | 12:24 |
| adac | odyssey4me, ok I see yes! Can I use AIO installation also for multinodes? Assuming just having two AIO on two different hosts in the same datacenter? | 12:25 |
| adac | odyssey4me, are you also planning to use docker at some point instead of lxc? | 12:26 |
| odyssey4me | adac you could adapt it for testing multinode if you want to, but it's better in that case to do a multinode build from scratch - the AIO is certainly not for production use | 12:26 |
| odyssey4me | the AIO does all sorts of nasty things to the host | 12:26 |
| odyssey4me | adac docker as a hypervisor is not on the cards at this point | 12:27 |
| adac | odyssey4me, which installation would be production ready if I may ask? | 12:27 |
| adac | odyssey4me, ok I see. I'm just more familiar with docker then with lxc right now, but this can be changed I guess :) | 12:27 |
| *** javeriak has joined #openstack-ansible | 12:29 | |
| *** javeriak_ has quit IRC | 12:29 | |
| odyssey4me | adac we use lxc containers as machine containers - so they're accessible via ssh and behave similarly to vm's (but are much more lightweight) | 12:31 |
| adac | Ok I undestand yes. Just heard Orcale uses docker containers now for its openstack installation | 12:32 |
| adac | *Oracle | 12:32 |
| openstackgerrit | Jean-Philippe Evrard proposed openstack/openstack-ansible-apt_package_pinning: Use sudo for role testing https://review.openstack.org/248774 | 12:35 |
| adac | odyssey4me, is there an "all in one" installation that is production ready? Or do I have to install then simply all openstack-ansible components manually? | 12:40 |
| mattt | AIO and production-ready shouldn't go in same sentence :P | 12:40 |
| adac | arrrrgggg :) | 12:41 |
| evrardjp | odyssey4me: are you ok if I create the use sudo for role testing for every IRR ? | 12:41 |
| evrardjp | or you'd rather wait for a merge? | 12:41 |
| adac | mattt, how do you install then a production ready system if I may ask? | 12:42 |
| odyssey4me | evrardjp you're welcome to patch all the role test playbooks as needed :) | 12:42 |
| evrardjp | Ok starting script | 12:43 |
| odyssey4me | adac for a production ready or multi node install you'd need to work through http://docs.openstack.org/developer/openstack-ansible/install-guide/index.html as your install guide | 12:43 |
| adac | odyssey4me, I'm on that page. It looks a bit huge to me... | 12:44 |
| odyssey4me | adac it's smaller than it looks - we're in the process of improving the docs, but bear in mind that this is for production | 12:45 |
| adac | odyssey4me, ok I will just go and read trough | 12:46 |
| evrardjp | I have something tricky here: I'd like to invert the dependency in git: atm I have one commit on top of another, I'd like to invert the order... is there a simple way, or I should just reset to the appropriate commit and pick where appropriate | 12:47 |
| adac | odyssey4me, it states "The recommended layout contains a minimum of five hosts". Initially i wanted to have a complete installation of everything on one single host. Is this still possible (but only not recommended) with openstack-ansible? | 12:48 |
| odyssey4me | adac you can use any number of hosts - anything is possible - the AIO is a clear showcase of that | 12:50 |
| odyssey4me | that recommendation is for production | 12:50 |
| odyssey4me | for production - at least 3 controller hosts, as you need 3 to maintain cluster quorum | 12:50 |
| evrardjp | nvm my comment, I did the old way | 12:50 |
| odyssey4me | then as many compute and storage hosts as you deem fit | 12:51 |
| *** jaypipes_ has joined #openstack-ansible | 12:51 | |
| *** jaypipes has quit IRC | 12:51 | |
| adac | odyssey4me, ok I see thanks! | 12:52 |
| *** jaypipes_ is now known as jaypipes | 12:53 | |
| openstackgerrit | Jean-Philippe Evrard proposed openstack/openstack-ansible-lxc_container_create: Use sudo for role testing https://review.openstack.org/249749 | 12:55 |
| openstackgerrit | Jean-Philippe Evrard proposed openstack/openstack-ansible-lxc_hosts: Use sudo for role testing https://review.openstack.org/249750 | 12:55 |
| openstackgerrit | Jean-Philippe Evrard proposed openstack/openstack-ansible-openstack_hosts: Use sudo for role testing https://review.openstack.org/249751 | 12:55 |
| openstackgerrit | Jean-Philippe Evrard proposed openstack/openstack-ansible-pip_install: Use sudo for role testing https://review.openstack.org/249752 | 12:55 |
| *** bangfrog has quit IRC | 12:55 | |
| openstackgerrit | Jean-Philippe Evrard proposed openstack/openstack-ansible-py_from_git: Use sudo for role testing https://review.openstack.org/249753 | 12:56 |
| openstackgerrit | Jean-Philippe Evrard proposed openstack/openstack-ansible-rsyslog_client: Use sudo for role testing https://review.openstack.org/249754 | 12:56 |
| mattt | lotes of gate failures :( | 12:57 |
| mattt | *lots | 12:57 |
| odyssey4me | mattt yeah, mostly hpcloud I'm guessing due to the disk space issue | 12:57 |
| openstackgerrit | Jean-Philippe Evrard proposed openstack/openstack-ansible-rsyslog_client: Use sudo for role testing https://review.openstack.org/249754 | 12:59 |
| adac | odyssey4me, one more question: the recommended layout, is this all located in one single datacenter or is it distributed across many datatacenters? I guess somehow nodes should be in different locations | 13:00 |
| mattt | odyssey4me: seems it, so did their instance configuration change recently? | 13:00 |
| openstackgerrit | Jean-Philippe Evrard proposed openstack/openstack-ansible-py_from_git: Use sudo for role testing https://review.openstack.org/249753 | 13:00 |
| *** javeriak has quit IRC | 13:00 | |
| odyssey4me | adac all in one DC usually - if you need multiple failure zones for the control plane you could split them up depending on your connectivity, or you could introduce regions which is a more advanced topic | 13:01 |
| openstackgerrit | Jean-Philippe Evrard proposed openstack/openstack-ansible-pip_install: Use sudo for role testing https://review.openstack.org/249752 | 13:01 |
| *** phiche has quit IRC | 13:01 | |
| openstackgerrit | Jean-Philippe Evrard proposed openstack/openstack-ansible-openstack_hosts: Use sudo for role testing https://review.openstack.org/249751 | 13:01 |
| odyssey4me | mattt I don't think so - I think we've just organically grown our usage that much | 13:01 |
| openstackgerrit | Jean-Philippe Evrard proposed openstack/openstack-ansible-lxc_hosts: Use sudo for role testing https://review.openstack.org/249750 | 13:01 |
| openstackgerrit | Jean-Philippe Evrard proposed openstack/openstack-ansible-lxc_container_create: Use sudo for role testing https://review.openstack.org/249749 | 13:02 |
| *** javeriak has joined #openstack-ansible | 13:02 | |
| *** phiche has joined #openstack-ansible | 13:03 | |
| *** jaypipes has quit IRC | 13:04 | |
| evrardjp | cloudnull: I've overwritten your inventory for openstack-hosts: https://review.openstack.org/#/c/249751/2/tests/inventory should I add groups and different hostnames (mapping to the same host) for the testing? | 13:04 |
| openstackgerrit | Jean-Philippe Evrard proposed openstack/openstack-ansible-rsyslog_client: Use sudo for role testing https://review.openstack.org/249754 | 13:05 |
| adac | odyssey4me, So region 1 would be for example the complete setup (as in the install guide) in Datacenter A and region2 would be the one that is the complete setup in Datacenter B and then those two are somehow connected via openstack mechanisms. Is my understanding of this more or less correct? | 13:06 |
| evrardjp | I don't know what's the distance between your DC, but if it's on two different geographical regions, you should split in regions | 13:08 |
| odyssey4me | adac yes, that is often a way to do it - the two regions are entirely seperate installations | 13:08 |
| evrardjp | if they are close, maybe you should stick to availability zones | 13:09 |
| odyssey4me | if you want to you can have an LDAP backend for Keystone in the two regions so that the identities are shared | 13:09 |
| evrardjp | true that | 13:09 |
| odyssey4me | you can also have a seperate database for keystone which is synchronised across regions | 13:09 |
| evrardjp | isn't LDAP + fernet far more easier odyssey4me? | 13:10 |
| evrardjp | sync databases that are far from each other can be tricky | 13:10 |
| odyssey4me | but yes, what evrardjp has said is correct - if you have multiple failure zones close to each other (eg: multiple rooms in a DC or multiple DC's near each other with great connectivity), then you can split the control plane across the failure zones and simply use OpenStack's availability zones for services | 13:10 |
| odyssey4me | evrardjp I would agree with using LDAP + fernet tokens as being a good strategy. | 13:14 |
| adac | odyssey4me, evrardjp So that means just for an example with the case of two DC close to each other: I have only one full installation where 2 control planes in Datacenter A and one in Datacenter B | 13:14 |
| evrardjp | whaat do you call "close"? in terms of latency? | 13:15 |
| evrardjp | -a | 13:16 |
| adac | evrardjp, I would say so, yes | 13:16 |
| evrardjp | do you know the latency between the 2? To know what you can do with dbs... | 13:16 |
| evrardjp | (and what's the speed/BW too) | 13:17 |
| evrardjp | for example, in our case we have only one installation, but all the controllers are dispatched in different datacenters, which makes the system quite resilient | 13:17 |
| adac | evrardjp, not yet no. But what is considered a good latency/speed for to be able to use the availabilty zone featue? | 13:18 |
| evrardjp | it's a difficult question adac :) | 13:18 |
| adac | evrardjp, I see :) | 13:18 |
| evrardjp | it depends on where you want your traffic to flow too... | 13:19 |
| odyssey4me | adac as with most complex architectures, it's often best to do some sort of testing to figure out what works | 13:19 |
| adac | I think the best way is two regions that is much easier alltogether I would say, isn't it? | 13:19 |
| odyssey4me | it has so many variables | 13:19 |
| evrardjp | for example, I guess you want your storage as close as possible to the nodes using it... it's difficult to generalize | 13:19 |
| odyssey4me | adac so the thing with two regions is that they are entirely separate for your users, so interconnectivity between instances needs to go through L3 networking or some sort of provider networks | 13:20 |
| adac | evrardjp, yes and that would arise even more questions: shouldn't the storage be "clustered" too in different datacenters | 13:20 |
| adac | odyssey4me, it would be the same as digitalocean has its setup, right? | 13:21 |
| evrardjp | odyssey4me: good remark :) | 13:21 |
| odyssey4me | adac I don't really know, but based on a chat I had at the Tokyo summit with one of the DO guys I expect that is so, yes. | 13:22 |
| evrardjp | odyssey4me: DO is using OpenStack? | 13:22 |
| odyssey4me | evrardjp I think so. It's entirely possible that I have my wires crossed. | 13:22 |
| evrardjp | interesting :) | 13:23 |
| adac | odyssey4me, from my experience in using DO and what you guys now tell me abot the differenent mechanisms, DO is having regions with " interconnectivity between instances needs to go through L3 networking or some sort of provider networks" | 13:23 |
| adac | since all you can do with DO is create machines in different regions | 13:23 |
| odyssey4me | adac some bed time reading for you :p http://docs.openstack.org/arch-design/content/ | 13:23 |
| evrardjp | http://digitalocean.uservoice.com/forums/136585-digitalocean/suggestions/3572231-implement-openstack-as-backend | 13:23 |
| adac | odyssey4me, you mean some MORE bed time reading :D | 13:24 |
| odyssey4me | :) | 13:24 |
| *** javeriak has quit IRC | 13:24 | |
| adac | oh and evrardjp just added some more MORE | 13:24 |
| adac | :) | 13:24 |
| evrardjp | that was for odyssey4me ;) | 13:24 |
| evrardjp | but you could | 13:25 |
| adac | hehe good! | 13:25 |
| evrardjp | just to explain what DO does | 13:25 |
| odyssey4me | evrardjp hmm, it must have been another provider | 13:25 |
| odyssey4me | evrardjp ah, it was DreamHost - not DigitalOcean | 13:28 |
| evrardjp | oh | 13:28 |
| evrardjp | yeah DreamHost are using OpenStack | 13:28 |
| mattt | "While OpenStack is great any large open source project also has its own set of issues and as a number of core contributors to OpenStack have openly discussed there are some issues with how the OpenStack project is being managed by Rackspace and the influence that they have over it." | 13:32 |
| mattt | that is a pretty odd statement for 2014 :P | 13:32 |
| evrardjp | yeah 2014 >< | 13:33 |
| evrardjp | it's weird to be stuck in 2009 | 13:33 |
| evrardjp | replacing Rackspace by HP would have been fun | 13:34 |
| evrardjp | anyway, that's trolling, sorry | 13:34 |
| *** bangfrog has joined #openstack-ansible | 13:37 | |
| openstackgerrit | Merged openstack/openstack-ansible: Update Liberty SHA's - 23.11.2015 https://review.openstack.org/245964 | 13:53 |
| *** jaypipes has joined #openstack-ansible | 13:57 | |
| *** shausy has quit IRC | 14:03 | |
| *** openstackgerrit has quit IRC | 14:06 | |
| *** openstackgerrit has joined #openstack-ansible | 14:07 | |
| *** mss has joined #openstack-ansible | 14:09 | |
| *** mss has quit IRC | 14:14 | |
| *** Bjoern_ has joined #openstack-ansible | 14:31 | |
| *** rromans_afk is now known as rromans | 14:36 | |
| *** javeriak has joined #openstack-ansible | 14:42 | |
| *** mgoddard_ has joined #openstack-ansible | 14:46 | |
| *** mgoddard has quit IRC | 14:50 | |
| *** vCloudernBeer has joined #openstack-ansible | 14:54 | |
| vCloudernBeer | anyone using vagrant(virtualbox as provider) to run the AIO? | 14:55 |
| vCloudernBeer | I tried this once and it took me 20 hours to run and eventually failed to connect to host #fail :( | 14:55 |
| mattt | vCloudernBeer: ouch! | 14:57 |
| vCloudernBeer | ping cloudnull | 14:57 |
| mattt | vCloudernBeer: no, i'd love for us to have a Vagrantfile for an AIO tho | 14:57 |
| mattt | (which works consistently :P) | 14:57 |
| vCloudernBeer | I wanted to build an env so I can start playing around and eventually contribute to the project | 14:58 |
| *** jwagner_away is now known as jwagner | 14:58 | |
| *** alkari has quit IRC | 14:58 | |
| vCloudernBeer | If use KVM to host the AIO how much RAM should I give as minimium? | 14:59 |
| vCloudernBeer | I was on twitter and <cloudnull> told me to bring this vagrant issue up and see if anyone has the same problem. | 14:59 |
| vCloudernBeer | her on IRC | 15:00 |
| vCloudernBeer | mean here on IRC | 15:00 |
| mattt | vCloudernBeer: 8 GB minimum | 15:02 |
| vCloudernBeer | <matt> thanks. My 2 Linux machines that can run KVM has only 8GB. Need to find some other ways for me. | 15:03 |
| odyssey4me | I trashed my AIO at just the wrong time. :( | 15:06 |
| odyssey4me | vCloudernBeer yeah, I've heard recently that using vagrant is super slow for some reason. I have yet to try it myself, but it seems untenable. | 15:06 |
| odyssey4me | mattt do you have an AIO up to verify something for me? | 15:07 |
| *** spotz_zzz is now known as spotz | 15:07 | |
| odyssey4me | as I recall we store all data into /openstack right? for instance /openstack/logs | 15:07 |
| vCloudernBeer | My main objective is to work on creating roles for Astara with a test env. | 15:08 |
| odyssey4me | vCloudernBeer 8GB is fine - we do all gate testing with 8GB instances in openstack-ci | 15:09 |
| mattt | odyssey4me: i don't, no .... | 15:09 |
| vCloudernBeer | It could be my host machine that is running the vagrant box has the performance issue? | 15:09 |
| odyssey4me | vCloudernBeer do you have access to cloud instances somewhere? | 15:09 |
| odyssey4me | vCloudernBeer it could be | 15:09 |
| mattt | odyssey4me: i have an 11.x deploy running on hardware though if i can check anything for you | 15:09 |
| odyssey4me | mattt you probably remember this from memory - mine is smashed from doing bash bits all day | 15:10 |
| odyssey4me | do we create /openstack and shove all our data bits in there? | 15:10 |
| mattt | odyssey4me: yeah | 15:10 |
| odyssey4me | ok cool - I'm going to make use of the same directory for other purposes in the AIO | 15:10 |
| mattt | odyssey4me: believe the galera files etc. are stored under there | 15:11 |
| vCloudernBeer | <odyssey4me> I got free access to Ravello Systems but I have not sign up. Maybe it is time now | 15:11 |
| openstackgerrit | Merged openstack/openstack-ansible: Fixed haproxy backend config issue https://review.openstack.org/245883 | 15:14 |
| *** baker has joined #openstack-ansible | 15:15 | |
| adac | woah how long does it take to install AIO :) I'm installing since 1,5 hours now I think | 15:17 |
| odyssey4me | adac it depends on many factors - the speed of your internet link, the speed of your disk and ho much RAM you have | 15:17 |
| odyssey4me | it can be as quick as 45 mins, but longer depending on your circumstances | 15:18 |
| odyssey4me | oh, and number of CPU's available is also a factor | 15:18 |
| odyssey4me | many of us test in public clouds where we use 8CPU, 8GB RAM instances and it take around an hour from scratch | 15:19 |
| vCloudernBeer | <adac> the step-by-step guide has a rough number from 30 min to 120 min depending on hardware | 15:19 |
| adac | Ok I see thank you! I think I'm on a good way then using 8GB of Ram and 4 cores. The internet is not the fastest | 15:20 |
| vCloudernBeer | I found this https://docs.vagrantup.com/v2/virtualbox/configuration.html let me give the trusty64 box more memory and try again tonight. | 15:20 |
| adac | On which hosting provider do you run your openstack-ansible installatione? I'm planning to run it on Hetzner datacenters (in Germany) but some things I still have to find out (i.e: if it is possible to have two external network interfaces for a root server ) | 15:25 |
| evrardjp | vCloudernBeer: I just tried yesterday the vagrant thingy. it worked | 15:26 |
| vCloudernBeer | <evrardjp> thanks. Do you have to tune the ram and cpu of your vagrant box? I am using virutalbox as the provider. | 15:28 |
| odyssey4me | adac for the AIO you only need one interface | 15:29 |
| adac | odyssey4me, OK I see yes. The AIO I only try out locally. On the htzner server then I will follwo your advice and isntall it with the 5 nodes | 15:29 |
| *** devl_ has joined #openstack-ansible | 15:30 | |
| adac | sorry for the typos | 15:30 |
| evrardjp | vCloudernBeer: yeah, I had to edit cpu and ram | 15:30 |
| evrardjp | but it's really not the best way to do it | 15:30 |
| vCloudernBeer | <evrardjp> I am new and is trying to setup a good dev env for openstack ansible so I can start contributing. :) | 15:32 |
| *** mgoddard has joined #openstack-ansible | 15:32 | |
| evrardjp | it's great! | 15:32 |
| evrardjp | don't forget to strip down your requirements (removing swift/ceilometer/...) | 15:33 |
| *** mgoddard_ has quit IRC | 15:33 | |
| evrardjp | the requirements are listed on the documentation (quickstart page) | 15:33 |
| vCloudernBeer | <evrardjp> oh yeah, I did not set swift and ceilometer to "no". Do you test on the cloud too? the best machine I have is a 8GB SSD laptop with Linux | 15:34 |
| evrardjp | 8GB ? | 15:35 |
| evrardjp | 8GB RAM or 8GB SSD Storage? | 15:35 |
| evrardjp | >< | 15:35 |
| vCloudernBeer | <evrardjp> 8GB of RAM, ssd size is 256 | 15:36 |
| evrardjp | 8GB RAM is a little short, but if you want to try, it's not impossible | 15:36 |
| evrardjp | I guessed vCloudernBeer ;) | 15:36 |
| adac | I see that with openstack-ansible also a log server is installed. Is this needed for openstack itself? Mean does this logstash server only collect logs for the openstack? | 15:36 |
| evrardjp | by default, there is no logstash | 15:36 |
| evrardjp | only a rsyslog | 15:36 |
| evrardjp | it's handy | 15:37 |
| evrardjp | it's not needed | 15:37 |
| adac | Yeah log servers are handy indeed | 15:37 |
| evrardjp | you could disable it, but I seriously doubt that you'd want it | 15:37 |
| cloudnull | morning | 15:37 |
| evrardjp | morning cloudnull | 15:37 |
| adac | nono, I'll use it evrardjp :-) | 15:37 |
| vCloudernBeer | good morning cloudnull | 15:37 |
| adac | godd morning! | 15:37 |
| adac | *good | 15:37 |
| cloudnull | o/ hows it ? | 15:38 |
| evrardjp | we need to change the doc to explain there is no logstash by default... | 15:38 |
| evrardjp | I'll do it | 15:38 |
| odyssey4me | adac it's not a logstash server - it's a centralised rsyslog - and no it's not compulsory if you don't want it | 15:38 |
| odyssey4me | ah, thanks evrardjp | 15:38 |
| cloudnull | vCloudernBeer: how goes the war on the VM front ? | 15:38 |
| adac | odyssey4me, evrardjp ah now I see yes. Logstash was mentioned in the hostlayout overview http://docs.openstack.org/developer/openstack-ansible/install-guide/overview-hostlayout.html | 15:39 |
| adac | just an example probably | 15:39 |
| odyssey4me | adac ah, that's old info which needs an overhaul | 15:39 |
| evrardjp | yeah, exactly what I should remove :p | 15:39 |
| adac | :) | 15:40 |
| vCloudernBeer | <cloudnull> Will try 2 things: tune the ram and cpu of my vagrant box. Investigate ravello systems for hosting the VM to run the openstack ansible. | 15:40 |
| adac | evrardjp, btw what is a "repository target host" I think it also not in this infographic | 15:41 |
| vCloudernBeer | <cloudnull> practice makes perfect and this goes my Thanksgiving holidays | 15:41 |
| evrardjp | I don't have the svg to redraw something | 15:41 |
| evrardjp | I'll just override text right now | 15:41 |
| cloudnull | vCloudernBeer: you get it to go on VMWare workstation ? | 15:41 |
| vCloudernBeer | <cloudnull> not yet. I tried but got connection issue to power up the Ubuntu server VM on Workstation. | 15:42 |
| cloudnull | adac evrardjp: talking about https://github.com/openstack/openstack-ansible/blob/master/doc/source/developer-docs/quickstart-aio.rst#reference-diagram-for-the-aio-build ? | 15:43 |
| evrardjp | no https://github.com/openstack/openstack-ansible/blob/9db681c49461d4939a02610058b54eb95467157b/doc/source/install-guide/install-infrastructure.rst | 15:43 |
| cloudnull | ah, RE: Sam-I-Am ^^ | 15:44 |
| spotz | morning | 15:44 |
| vCloudernBeer | thanks everyone. I am signing off and will try again with setting up an AIO on vagrant with tuned params tonight. Happy Thanksgiving. | 15:46 |
| spotz | Happy thanksgiving vcloudernbeer | 15:47 |
| evrardjp | cloudnull: I'll make a commit, I hope Sam-I-Am will jump on it :) | 15:48 |
| cloudnull | he's aware that he can fix docs or deal with the mess we create :p | 15:49 |
| openstackgerrit | Jean-Philippe Evrard proposed openstack/openstack-ansible: Removing the references of the elk stack https://review.openstack.org/249865 | 15:51 |
| evrardjp | cloudnull: I guess it could be nice if we get the sudo changes to merge | 15:52 |
| *** iceyao has quit IRC | 15:52 | |
| evrardjp | this way I can add my .git* files too | 15:52 |
| evrardjp | (the other pending reviews on the IRR) | 15:52 |
| evrardjp | and then bring interesting features (like the mtu stuff :p) | 15:53 |
| cloudnull | ++ | 15:55 |
| cloudnull | we need to do that | 15:55 |
| cloudnull | it seems that we need to investigate some of the test failures at this point. | 15:55 |
| cloudnull | one passed | 15:55 |
| cloudnull | https://review.openstack.org/#/c/249752/2 | 15:55 |
| spotz | hey I've been trying to keep up! | 15:56 |
| *** iceyao has joined #openstack-ansible | 15:56 | |
| cloudnull | ++ spotz :) | 15:56 |
| spotz | I do need to get back to doing stuff vs just reviewing though:( | 15:57 |
| evrardjp | cloudnull: another one passed before | 15:58 |
| evrardjp | the apt one | 15:58 |
| cloudnull | odyssey4me: can you pop into openstack-oslo inc0 is proding to see if theres are ways to make oslo + syslog happier. idk if theyll reply but i know youve worked on that in the past so it might be good to join that convo should it ever happen | 15:58 |
| cloudnull | yes https://review.openstack.org/#/c/248774/ | 15:59 |
| cloudnull | nvmd it seems your already in that channel | 15:59 |
| cloudnull | logan-: RE: I remember you had some issues with nova+syslog too. right? | 16:00 |
| *** phiche has quit IRC | 16:01 | |
| evrardjp | cloudnull: http://logs.openstack.org/54/249754/3/check/gate-openstack-ansible-rsyslog_client-dsvm-functional-trusty/299f677/console.html#_2015-11-25_13_10_27_607 | 16:01 |
| kmARC | hey guys. I have succesfully installed openstack at last. (Hooray) However, have two questions. | 16:02 |
| evrardjp | are you sure about the assert thingy you added in your tests? | 16:03 |
| kmARC | 1) The documentation mentions kibana and logstash installation. Am I missing something here, or it's not installed at all? | 16:03 |
| cloudnull | evrardjp: it worked locally however im trying it again now | 16:04 |
| kmARC | 2) The automatically created openrc file doesn't seem to work | 16:04 |
| evrardjp | ok | 16:04 |
| evrardjp | I'll check the other during that time | 16:04 |
| kmARC | ... however Horizon works perfectly | 16:04 |
| cloudnull | kmARC: the kibana stuff was removed from the stack in Kilo | 16:04 |
| cloudnull | its now in https://github.com/rcbops/rpc-openstack | 16:05 |
| kmARC | oh, too bad. Why? I need the fancy graphs :) | 16:05 |
| cloudnull | it worked-ish, but was deamed not an openstack concern | 16:05 |
| kmARC | okay. hopefully I can create my own ansible script based on these. Thanks! | 16:06 |
| cloudnull | so in rpc-openstack we created roles to facilitate kibana / logstash and we fixed the "it worked-ish" bits too | 16:06 |
| kmARC | any ideas about the 2) question? | 16:06 |
| cloudnull | evrardjp: so cloning https://github.com/openstack/openstack-ansible-rsyslog_client and running tox -e ansible-functional worked locally | 16:07 |
| cloudnull | and I kicked a clean vm to test | 16:07 |
| evrardjp | interesting | 16:08 |
| *** Mudpuppy has joined #openstack-ansible | 16:08 | |
| cloudnull | well. it worked with the base role executing as root | 16:08 |
| evrardjp | yeah sure | 16:08 |
| evrardjp | but I wonder how it's possible to install but not doing the assert | 16:08 |
| cloudnull | so now im creating a user + sudo and testing your patch to mimic the gate | 16:08 |
| evrardjp | it should have failed before | 16:08 |
| *** daneyon_ has quit IRC | 16:08 | |
| *** daneyon has joined #openstack-ansible | 16:09 | |
| evrardjp | if you have a vagrant machine that's by default on user/sudo | 16:09 |
| cloudnull | i dont, i beat on rax public cloud like my distributed vagrant env :p | 16:09 |
| evrardjp | which is pure awesome. Lucky basterd | 16:10 |
| evrardjp | :p | 16:10 |
| openstackgerrit | Miguel Alex Cantu proposed openstack/openstack-ansible: [WIP]Added ceilometer-polling upstart jobs https://review.openstack.org/249881 | 16:11 |
| adac | Maybe what also would be helpful to have some minimum hardwar requirements for the single hosts in the howto | 16:12 |
| odyssey4me | adac how long is a piece of string? ;) | 16:12 |
| adac | hehe | 16:12 |
| odyssey4me | if it can all run on a single 8vCPU, 8GB RAM host... albeit only for testing... then the fact is it can run on just about anything - it just depends on what you want to use it for | 16:13 |
| evrardjp | btw cloudnull, shouldn't we use -vvv in the ansible-playbook calls started under tox? | 16:13 |
| cloudnull | evrardjp: so i see whats up | 16:13 |
| cloudnull | https://review.openstack.org/#/c/249754/3/tests/inventory,cm | 16:13 |
| adac | odyssey4me, yes that is true | 16:13 |
| cloudnull | thats the locallized inventory needed for the role to execute | 16:13 |
| evrardjp | oh yeah, I mentionned that in a PM | 16:14 |
| pabelanger | morning! Did you guys / gals manage to get your gate jobs working correctly? | 16:14 |
| evrardjp | sorry I scratched everything by doing a mass copy | 16:14 |
| cloudnull | evrardjp: to me ? maybe my bouncer dropped it ? | 16:14 |
| evrardjp | cloudnull: yup indeed | 16:14 |
| cloudnull | or i fogot that i replied :) all is certainly possible given appropriate amounts of beer | 16:14 |
| spotz | pabelanger I think they did but not 100% | 16:15 |
| cloudnull | pabelanger: they're on track | 16:16 |
| pabelanger | nice | 16:17 |
| cloudnull | we have some localized tuning we need to get done but I think thats outside of the test jobs | 16:17 |
| *** gparaskevas has quit IRC | 16:17 | |
| evrardjp | it was today | 16:17 |
| evrardjp | "evrardjp: cloudnull: I've overwritten your inventory for openstack-hosts: https://review.openstack.org/#/c/249751/2/tests/inventory should I add groups and different hostnames (mapping to the same host) for the testing?" | 16:17 |
| evrardjp | so it's just your bouncer | 16:17 |
| *** sacharya has joined #openstack-ansible | 16:18 | |
| evrardjp | so cloudnull: what's the purpose of these lines? | 16:19 |
| evrardjp | want multiple hostnames only? | 16:19 |
| evrardjp | and the same groups you mentionned? | 16:19 |
| openstackgerrit | Jesse Pretorius proposed openstack/openstack-ansible: Change AIO bootstrap disk preparation https://review.openstack.org/249887 | 16:20 |
| evrardjp | I can edit that for you, fix my mistake | 16:21 |
| evrardjp | during that time, could you check on this? https://review.openstack.org/#/c/249753/2 | 16:21 |
| odyssey4me | evrardjp cloudnull mattt any volunteers to review this scary rewrite? https://review.openstack.org/249887 | 16:22 |
| *** sacharya has quit IRC | 16:22 | |
| odyssey4me | this review will hopefully take care of the 'no host found' issues in the gate, which happen to be due to a lack of disk space on the AIO | 16:23 |
| evrardjp | can't promise anything but I'll look at it odyssey4me :p | 16:23 |
| cloudnull | evrardjp: http://cdn.pasteraw.com/339f3s5u4r5w67unw8ch9z7y14x9f3i made the tests work for me | 16:24 |
| odyssey4me | hopefully it makes it more understanndable now | 16:25 |
| evrardjp | cloudnull: my question was: is this fancy ip really needed? Do you really want to test the ssh connection to localhost ? | 16:25 |
| evrardjp | or can I switch back to a local connection? | 16:26 |
| cloudnull | on test1 and 2 ? | 16:26 |
| cloudnull | localhost is using local | 16:26 |
| cloudnull | the other two entries are there to exercise the client options of two rsyslog server targets | 16:27 |
| cloudnull | odyssey4me: so i have a question ... | 16:28 |
| odyssey4me | cloudnull yep | 16:28 |
| cloudnull | can we get what you've done in those scripty bits into https://review.openstack.org/#/c/244708/ ? | 16:29 |
| *** devl_ has quit IRC | 16:29 | |
| *** baker has quit IRC | 16:29 | |
| cloudnull | IMO the bootstrap AIO as a playbook would be much better overall | 16:29 |
| evrardjp | let's say ok, but I don't think it's needed to give an IP (no need for ssh connection). Same for https://review.openstack.org/#/c/249751/2/tests/inventory I think we don't need the [all] section at all | 16:29 |
| odyssey4me | cloudnull yep, for now I want to see it work - then I'll add it to mhayden's bootstrap rewrite | 16:29 |
| *** subscope has quit IRC | 16:29 | |
| *** greg_a has joined #openstack-ansible | 16:30 | |
| *** greg_a has quit IRC | 16:30 | |
| odyssey4me | cloudnull I did that already with https://review.openstack.org/248887 | 16:30 |
| cloudnull | the role worked that mhayden put together but it will never work right if we dont put cycles into it and if we're looking to improve the gate process it would be nice to see that go in as a first step. | 16:30 |
| cloudnull | just my 2 cents | 16:31 |
| *** jwagner is now known as jwagner_away | 16:32 | |
| *** jwagner_away is now known as jwagner | 16:32 | |
| openstackgerrit | Jean-Philippe Evrard proposed openstack/openstack-ansible-rsyslog_client: Use sudo for role testing https://review.openstack.org/249754 | 16:34 |
| alextricity | I hate when I'm working on a bug fix..and then I realize the AIO's mariadb crashed | 16:34 |
| odyssey4me | cloudnull yep, except that https://review.openstack.org/248887 needs some changes that are in https://review.openstack.org/248887 to work | 16:34 |
| alextricity | I should just have 3-4 AIOs on stand by ready to go | 16:35 |
| odyssey4me | but yeah, I will work on that patch tomorrow - we just need the ssh working properly for it to work | 16:35 |
| openstackgerrit | Jean-Philippe Evrard proposed openstack/openstack-ansible-openstack_hosts: Use sudo for role testing https://review.openstack.org/249751 | 16:35 |
| evrardjp | call me pessimistic, I don't think it will pass tests :p | 16:36 |
| *** devl_ has joined #openstack-ansible | 16:40 | |
| *** devl_ has quit IRC | 16:42 | |
| *** devl_ has joined #openstack-ansible | 16:42 | |
| *** devl_ has quit IRC | 16:44 | |
| cloudnull | evrardjp: that update failed in my env | 16:44 |
| cloudnull | but with the inventory using http://cdn.pasteraw.com/339f3s5u4r5w67unw8ch9z7y14x9f3i it worked | 16:44 |
| *** devl_ has joined #openstack-ansible | 16:44 | |
| *** devl_ has quit IRC | 16:46 | |
| *** mgoddard_ has joined #openstack-ansible | 16:46 | |
| openstackgerrit | Jesse Pretorius proposed openstack/openstack-ansible: Change AIO bootstrap disk preparation https://review.openstack.org/249887 | 16:47 |
| *** javeriak has quit IRC | 16:48 | |
| *** javeriak has joined #openstack-ansible | 16:48 | |
| *** eil397 has joined #openstack-ansible | 16:49 | |
| *** mgoddard has quit IRC | 16:50 | |
| evrardjp | cloudnull: ok I'll check that | 16:51 |
| *** galstrom_zzz is now known as galstrom | 16:52 | |
| *** javeriak_ has joined #openstack-ansible | 16:53 | |
| openstackgerrit | Jesse Pretorius proposed openstack/openstack-ansible: Change AIO bootstrap disk preparation https://review.openstack.org/249887 | 16:53 |
| *** javeriak has quit IRC | 16:53 | |
| *** mgoddard_ has quit IRC | 16:56 | |
| *** mgoddard has joined #openstack-ansible | 16:57 | |
| openstackgerrit | Jesse Pretorius proposed openstack/openstack-ansible: Change AIO bootstrap disk preparation https://review.openstack.org/249887 | 17:00 |
| *** markvoelker has quit IRC | 17:03 | |
| adac | with the playbooks, is there also some firewall installed or some firewall rules enabled guys? | 17:06 |
| *** javeriak has joined #openstack-ansible | 17:06 | |
| *** javeriak_ has quit IRC | 17:06 | |
| cloudnull | adac: no | 17:07 |
| adac | cloudnull, thanks! | 17:07 |
| *** galstrom is now known as galstrom_zzz | 17:07 | |
| cloudnull | itll configure iptables to work within the env but nothing has been done to secure it | 17:07 |
| cloudnull | in terms of the firewall | 17:08 |
| cloudnull | you can however use https://github.com/openstack/openstack-ansible-security | 17:08 |
| spotz | odyssey4me You ready for eyes on that? | 17:08 |
| cloudnull | which will secure the env using the STIG work mhayden had done | 17:08 |
| openstackgerrit | Merged openstack/openstack-ansible: Remove ssh args to ensure pipelining is working https://review.openstack.org/249348 | 17:08 |
| adac | cloudnull, could messing around with the rules (mean like add additional ones) possibly disrupt the setup? | 17:08 |
| odyssey4me | spotz where do my eyes need to be? | 17:08 |
| spotz | hehe I wanted to know if you made the changes to your patch and wanted me to look?:) | 17:09 |
| odyssey4me | evrardjp I see that https://review.openstack.org/249752 is working - do we have the other repositories working yet? | 17:09 |
| odyssey4me | spotz it's not working yet - I just discovered a bug which crosses multiple repositories :/ | 17:09 |
| evrardjp | odyssey4me: nope, I'm busy with something else, sorry | 17:10 |
| spotz | okie, I'm gonna go work on a heat doc bug. poke if needed | 17:10 |
| *** jwagner is now known as jwagner_away | 17:10 | |
| cloudnull | adac: its possible . | 17:10 |
| cloudnull | depends on what you're messing with | 17:10 |
| adac | cloudnull, hehe sounds logically yes | 17:11 |
| cloudnull | the lxc forward and mangle rules I'd try to leave alone | 17:11 |
| cloudnull | anything thats adding a checksum should remain as well | 17:11 |
| adac | cloudnull, What is open to the public at all I mean. It is probably the webinterface right and probably also the API | 17:11 |
| cloudnull | and on compute nodes there will be a mess of neutron rules that should be mostly left alone | 17:11 |
| adac | mean which ports | 17:12 |
| *** eil397 has quit IRC | 17:12 | |
| cloudnull | just about everything is open any/any -- you can disable just about all public ingress. at a glance you can see the ports needed and where they may live using the haproxy config | 17:14 |
| cloudnull | https://github.com/openstack/openstack-ansible/blob/master/playbooks/vars/configs/haproxy_config.yml | 17:14 |
| cloudnull | not that your using haproxy | 17:14 |
| cloudnull | but those are the openstack ports that will need to be paid attention too | 17:14 |
| adac | cloudnull, thanks! | 17:14 |
| openstackgerrit | Merged openstack/openstack-ansible-pip_install: Use sudo for role testing https://review.openstack.org/249752 | 17:18 |
| odyssey4me | evrardjp ^ :) | 17:19 |
| *** sdake has joined #openstack-ansible | 17:21 | |
| adac | Some of you guys are rackspace guys? Does rackspace also offer dedicated servers in Europe? | 17:21 |
| openstackgerrit | Jesse Pretorius proposed openstack/openstack-ansible-pip_install: New git dotfiles for independant repository https://review.openstack.org/248637 | 17:22 |
| *** kmARC has quit IRC | 17:22 | |
| cloudnull | adac: we do | 17:22 |
| odyssey4me | cloudnull review please: https://review.openstack.org/249754 | 17:22 |
| * cloudnull reviewing | 17:22 | |
| cloudnull | boom ! | 17:23 |
| odyssey4me | cloudnull I have discovered that we're referring to '/var/lib/lxc' all over the place instead of '{{ lxc_container_directory }}' :( | 17:24 |
| *** b3rnard0 is now known as b3rnard0_away | 17:25 | |
| *** vCloudernBeer has quit IRC | 17:25 | |
| cloudnull | odyssey4me: http://cdn.pasteraw.com/q7hik1zp6fa57avyi7sdolue1n7u1b2 :'( -- not surprising though | 17:26 |
| adac | cloudnull, oh wow I just saw it on the webpage. Not that cheap I have to say. | 17:26 |
| cloudnull | nope. | 17:26 |
| openstackgerrit | Merged openstack/openstack-ansible-rsyslog_client: Use sudo for role testing https://review.openstack.org/249754 | 17:26 |
| adac | But one day when I have more money I will switch maybe! | 17:27 |
| cloudnull | but their a managed offering | 17:27 |
| adac | cloudnull, what exactly does that mean? | 17:27 |
| cloudnull | we have people on hand 24/7 to manage the boxes. | 17:27 |
| *** eil397 has joined #openstack-ansible | 17:28 | |
| adac | cloudnull, Ok I see yes! I'm currently with Hetzner, it is just cheaper. But maybe later I will change. Give something back | 17:28 |
| adac | (well its a lot cheaper I have to say) | 17:28 |
| adac | :) | 17:28 |
| openstackgerrit | Jesse Pretorius proposed openstack/openstack-ansible-rsyslog_client: New git dotfiles for independant repository https://review.openstack.org/248639 | 17:29 |
| openstackgerrit | Jesse Pretorius proposed openstack/openstack-ansible: Remove ssh args to ensure pipelining is working https://review.openstack.org/249942 | 17:35 |
| cloudnull | adac: i can imagine . | 17:36 |
| cloudnull | :) | 17:36 |
| *** tiagogomes_ has quit IRC | 17:38 | |
| openstackgerrit | Miguel Alex Cantu proposed openstack/openstack-ansible: Added ceilometer-polling upstart jobs https://review.openstack.org/249881 | 17:40 |
| openstackgerrit | Miguel Alex Cantu proposed openstack/openstack-ansible: Added ceilometer-polling upstart jobs https://review.openstack.org/249946 | 17:42 |
| alextricity | How do I do the thing to add branches to a specific bug? | 17:45 |
| alextricity | Maybe that requires special permissions | 17:46 |
| *** Mudpuppy has quit IRC | 17:47 | |
| cloudnull | change the topic | 17:47 |
| openstackgerrit | Jesse Pretorius proposed openstack/openstack-ansible: Change AIO bootstrap disk preparation https://review.openstack.org/249887 | 17:50 |
| *** tlian has joined #openstack-ansible | 17:51 | |
| odyssey4me | cloudnull review please https://review.openstack.org/248637 | 17:54 |
| openstackgerrit | Merged openstack/openstack-ansible-apt_package_pinning: Use sudo for role testing https://review.openstack.org/248774 | 17:55 |
| odyssey4me | cloudnull and this one https://review.openstack.org/248639 | 17:55 |
| openstackgerrit | Merged openstack/openstack-ansible-apt_package_pinning: Use sudo for role testing https://review.openstack.org/248774 | 17:55 |
| openstackgerrit | Jesse Pretorius proposed openstack/openstack-ansible-apt_package_pinning: New git dotfiles for independant repository https://review.openstack.org/248632 | 17:56 |
| openstackgerrit | Jesse Pretorius proposed openstack/openstack-ansible-apt_package_pinning: Update run_tests to be more complete https://review.openstack.org/248707 | 17:56 |
| cloudnull | done | 17:57 |
| openstackgerrit | Jesse Pretorius proposed openstack/openstack-ansible: Updated the py_pkgs lookup plugin for multi source https://review.openstack.org/241483 | 17:58 |
| openstackgerrit | Merged openstack/openstack-ansible-pip_install: New git dotfiles for independant repository https://review.openstack.org/248637 | 17:58 |
| openstackgerrit | Merged openstack/openstack-ansible-rsyslog_client: New git dotfiles for independant repository https://review.openstack.org/248639 | 17:59 |
| cloudnull | we need to get the rest of the sudo updates completed. evrardjp can i help out on that front ? | 18:00 |
| *** sacharya has joined #openstack-ansible | 18:00 | |
| evrardjp | yup sure | 18:00 |
| evrardjp | sorry I'm a little overwhelmed :p | 18:00 |
| openstackgerrit | Merged openstack/openstack-ansible: Update requirements to suit the existence of lxc vg https://review.openstack.org/247490 | 18:00 |
| openstackgerrit | Merged openstack/openstack-ansible: Removing the references of the elk stack https://review.openstack.org/249865 | 18:01 |
| cloudnull | no worries at all | 18:02 |
| cloudnull | where can i do things ? | 18:02 |
| evrardjp | I'll just check my current status | 18:03 |
| *** sacharya has quit IRC | 18:05 | |
| openstackgerrit | Kevin Carter proposed openstack/openstack-ansible-lxc_container_create: Use sudo for role testing https://review.openstack.org/249749 | 18:05 |
| openstackgerrit | Merged openstack/openstack-ansible-apt_package_pinning: New git dotfiles for independant repository https://review.openstack.org/248632 | 18:06 |
| *** karimb has quit IRC | 18:07 | |
| evrardjp | ok cloudnull I'm up now | 18:09 |
| evrardjp | maybe start with this one | 18:09 |
| evrardjp | https://review.openstack.org/#/c/249753/ | 18:09 |
| evrardjp | so yes, I don't have any in-tree edit, so feel free | 18:11 |
| *** adac has quit IRC | 18:12 | |
| *** jduhamel has joined #openstack-ansible | 18:12 | |
| openstackgerrit | Jesse Pretorius proposed openstack/openstack-ansible: Change AIO bootstrap disk preparation https://review.openstack.org/249887 | 18:13 |
| evrardjp | I'll fix the openstack-hosts | 18:14 |
| openstackgerrit | Jean-Philippe Evrard proposed openstack/openstack-ansible-openstack_hosts: Use sudo for role testing https://review.openstack.org/249751 | 18:15 |
| evrardjp | for the lxc-hosts you'll need a key generated I guess | 18:16 |
| evrardjp | I'll let you this one | 18:16 |
| *** toddnni has joined #openstack-ansible | 18:17 | |
| openstackgerrit | Kevin Carter proposed openstack/openstack-ansible-py_from_git: Use sudo for role testing https://review.openstack.org/249753 | 18:19 |
| cloudnull | ill bang on lxc-hosts | 18:20 |
| openstackgerrit | Jesse Pretorius proposed openstack/openstack-ansible: Change AIO bootstrap disk preparation https://review.openstack.org/249887 | 18:21 |
| *** dmsimard is now known as dmsimard|food | 18:22 | |
| evrardjp | oh you changed py-from-git | 18:23 |
| evrardjp | cool | 18:23 |
| openstackgerrit | Kevin Carter proposed openstack/openstack-ansible-lxc_hosts: Use sudo for role testing https://review.openstack.org/249750 | 18:24 |
| evrardjp | not sure if tmp is the right place though | 18:24 |
| cloudnull | it should work for the test case | 18:24 |
| cloudnull | idk though we'll see | 18:25 |
| cloudnull | the failure in that package was that the test install was installing the py lxc bindings which needed the lxc.h lib | 18:25 |
| cloudnull | so https://review.openstack.org/#/c/249749/ will go when https://review.openstack.org/#/c/249750/ is in | 18:27 |
| *** jwagner_away is now known as jwagner | 18:27 | |
| evrardjp | ok good to know | 18:28 |
| evrardjp | it makes sense | 18:28 |
| evrardjp | so, cloudnull what do you think of adding -vvv to the playbook runs? If it's for gate logs, it's maybe better to have more of them? | 18:31 |
| cloudnull | makes sense to me | 18:31 |
| cloudnull | odyssey4me: ? | 18:31 |
| cloudnull | anyone around that can bang these through https://review.openstack.org/#/c/249751/ https://review.openstack.org/#/c/249753/ ? | 18:35 |
| cloudnull | andymccr d34dh0r53 odyssey4me hughsaunders mattt stevelle ? | 18:36 |
| *** jwagner is now known as jwagner_away | 18:36 | |
| odyssey4me | cloudnull you can do https://review.openstack.org/249753 | 18:37 |
| odyssey4me | evrardjp maybe -v, but I think that -vvv is a bit much as it contains all the connection data | 18:38 |
| evrardjp | true | 18:38 |
| *** galstrom_zzz is now known as galstrom | 18:39 | |
| *** jwagner_away is now known as jwagner | 18:39 | |
| odyssey4me | it's so cool to have such quick feedback! | 18:39 |
| *** b3rnard0_away is now known as b3rnard0 | 18:40 | |
| spotz | no feedback for you odyssey4me! | 18:41 |
| openstackgerrit | Merged openstack/openstack-ansible-openstack_hosts: Use sudo for role testing https://review.openstack.org/249751 | 18:43 |
| openstackgerrit | Merged openstack/openstack-ansible-py_from_git: Use sudo for role testing https://review.openstack.org/249753 | 18:43 |
| odyssey4me | cloudnull so https://review.openstack.org/249887 should work now - I've just simply partitioned two spaces instead of one big one - the trouble is that https://review.openstack.org/239525 needs to be functionally the same... so I'll work on incorporating the updated changes, but we'll need the first to merge to get the gate moving | 18:44 |
| openstackgerrit | Jesse Pretorius proposed openstack/openstack-ansible-py_from_git: New git dotfiles for independant repository https://review.openstack.org/248638 | 18:44 |
| openstackgerrit | Jesse Pretorius proposed openstack/openstack-ansible-openstack_hosts: New git dotfiles for independant repository https://review.openstack.org/248636 | 18:44 |
| odyssey4me | cloudnull need https://review.openstack.org/248636 and https://review.openstack.org/248638 reviewed now | 18:45 |
| *** devl_ has joined #openstack-ansible | 18:47 | |
| d34dh0r53 | cloudnull: looking now | 18:48 |
| openstackgerrit | Kevin Carter proposed openstack/openstack-ansible-lxc_hosts: Use sudo for role testing https://review.openstack.org/249750 | 18:49 |
| cloudnull | odyssey4me: done btw | 18:50 |
| odyssey4me | cloudnull FYI it seems there's an issue in https://review.openstack.org/241483 | 18:51 |
| *** galstrom is now known as galstrom_zzz | 18:51 | |
| cloudnull | thats exciting , maybe a regression i introduced in the last pr... | 18:52 |
| * cloudnull looking now | 18:52 | |
| openstackgerrit | Merged openstack/openstack-ansible-py_from_git: New git dotfiles for independant repository https://review.openstack.org/248638 | 18:52 |
| alextricity | How much ram are you guys giving the AIO? My 8GB VMs are lasting 10min before mariadb crashes | 18:52 |
| eil397 | alextricity: as I remember 16GB is described as requirement | 18:53 |
| alextricity | eil397: Thanks :) I'll beef them up | 18:53 |
| openstackgerrit | Merged openstack/openstack-ansible-openstack_hosts: New git dotfiles for independant repository https://review.openstack.org/248636 | 18:56 |
| odyssey4me | alextricity The 8GB RAM VM's struggle a bit, especially if you're actively doing stuff on them. 16GB is recommended. I have just thought that it may be good to put 'minimal' and 'recommended' requirements into the docs. | 18:56 |
| odyssey4me | evrardjp cloudnull any thoughts on doing https://review.openstack.org/248707 for all the role repositories? | 18:58 |
| *** adac has joined #openstack-ansible | 19:01 | |
| *** Mudpuppy has joined #openstack-ansible | 19:03 | |
| *** alextricity has quit IRC | 19:05 | |
| *** jwagner is now known as jwagner_away | 19:12 | |
| *** Mudpuppy has quit IRC | 19:13 | |
| *** dmsimard|food is now known as dmsimard | 19:13 | |
| openstackgerrit | Jesse Pretorius proposed openstack/openstack-ansible: Improve message of the teardown.sh https://review.openstack.org/249325 | 19:13 |
| openstackgerrit | Jesse Pretorius proposed openstack/openstack-ansible-apt_package_pinning: Update run_tests to be more complete https://review.openstack.org/248707 | 19:16 |
| spotz | ervardjp I put some inline comments on https://review.openstack.org/249325 let me know if they didn't come through | 19:26 |
| odyssey4me | spotz feel free to do a vote of -1 if you think a patch could do with improvement :) | 19:28 |
| odyssey4me | it helps to give visibility to your comments :) | 19:28 |
| spotz | odyssey4me - I wasn't sure they would go through. First time I've used the inline comments:) | 19:29 |
| spotz | But if you saw them I can vote with a -1 | 19:29 |
| odyssey4me | spotz well, congrats then! inline comments are great as they're seen in context :) | 19:29 |
| spotz | ha you already did the vote:) | 19:30 |
| odyssey4me | spotz yes, but it's your vote that goes with the comments - my vote is simply an agreement | 19:30 |
| spotz | ok I'll go add a vote though | 19:31 |
| spotz | While not the best grammar person myself if you guys ever want me to read through and tweek let me know | 19:32 |
| openstackgerrit | Jesse Pretorius proposed openstack/openstack-ansible: Implement new Liberty default for Neutron rpc_workers https://review.openstack.org/249990 | 19:38 |
| *** tlian2 has joined #openstack-ansible | 19:39 | |
| *** tlian has quit IRC | 19:41 | |
| spotz | Have a Happy Thanksgiving guys, I'm out but will check for re-reviews | 19:45 |
| *** spotz is now known as spotz_zzz | 19:46 | |
| eil397 | Happy a Happy Thanksgiving | 19:47 |
| openstackgerrit | Kevin Carter proposed openstack/openstack-ansible: Updated the py_pkgs lookup plugin for multi source https://review.openstack.org/241483 | 19:47 |
| cloudnull | odyssey4me: ^ | 19:47 |
| openstackgerrit | Jesse Pretorius proposed openstack/openstack-ansible: Implement tunable heat api and engine worker overrides https://review.openstack.org/249995 | 19:48 |
| openstackgerrit | Jesse Pretorius proposed openstack/openstack-ansible: Implement tunable Horizon/Apache processes/threads overrides https://review.openstack.org/249997 | 19:56 |
| *** javeriak has quit IRC | 20:08 | |
| openstackgerrit | Jesse Pretorius proposed openstack/openstack-ansible: Implement tunable Ceilometer workers https://review.openstack.org/250001 | 20:10 |
| openstackgerrit | Jesse Pretorius proposed openstack/openstack-ansible: Implement tunable Ceilometer workers https://review.openstack.org/250001 | 20:17 |
| openstackgerrit | Jesse Pretorius proposed openstack/openstack-ansible: Implement tunable Ceilometer workers https://review.openstack.org/250001 | 20:22 |
| openstackgerrit | Jesse Pretorius proposed openstack/openstack-ansible: Implement tunable Horizon/Apache processes/threads overrides https://review.openstack.org/249997 | 20:22 |
| openstackgerrit | Jesse Pretorius proposed openstack/openstack-ansible: Implement tunable heat api and engine worker overrides https://review.openstack.org/249995 | 20:23 |
| openstackgerrit | Jesse Pretorius proposed openstack/openstack-ansible: Implement new Liberty default for Neutron rpc_workers https://review.openstack.org/249990 | 20:23 |
| openstackgerrit | Jesse Pretorius proposed openstack/openstack-ansible: Change AIO bootstrap disk preparation https://review.openstack.org/249887 | 20:24 |
| *** adac has quit IRC | 20:31 | |
| openstackgerrit | Jesse Pretorius proposed openstack/openstack-ansible: Implement tunable Ceilometer workers https://review.openstack.org/250001 | 20:33 |
| openstackgerrit | Jesse Pretorius proposed openstack/openstack-ansible: Added ceilometer-polling upstart jobs https://review.openstack.org/249881 | 20:39 |
| openstackgerrit | Kevin Carter proposed openstack/openstack-ansible-lxc_hosts: Use sudo for role testing https://review.openstack.org/249750 | 20:39 |
| *** sdake has quit IRC | 20:40 | |
| *** sacharya has joined #openstack-ansible | 20:41 | |
| *** adac has joined #openstack-ansible | 20:46 | |
| * cloudnull going to eat | 20:48 | |
| *** karimb has joined #openstack-ansible | 20:54 | |
| openstackgerrit | Merged openstack/openstack-ansible-lxc_hosts: Use sudo for role testing https://review.openstack.org/249750 | 20:55 |
| openstackgerrit | Jesse Pretorius proposed openstack/openstack-ansible-lxc_hosts: New git dotfiles for independant repository https://review.openstack.org/248635 | 20:56 |
| *** grumpycatt has joined #openstack-ansible | 20:57 | |
| openstackgerrit | Kevin Carter proposed openstack/openstack-ansible-lxc_hosts: Removed undesired .DS_Store file https://review.openstack.org/248643 | 20:58 |
| eil397 | cloudnull Bon Appétit | 21:03 |
| openstackgerrit | Merged openstack/openstack-ansible-lxc_hosts: New git dotfiles for independant repository https://review.openstack.org/248635 | 21:06 |
| openstackgerrit | Merged openstack/openstack-ansible-lxc_hosts: Removed undesired .DS_Store file https://review.openstack.org/248643 | 21:07 |
| *** tlian has joined #openstack-ansible | 21:08 | |
| *** tlian2 has quit IRC | 21:12 | |
| bgmccollum | odyssey4me: would be helpful to grab the libvirt logs in the gate...that is where you will see errors about unidentifiable CPU | 21:13 |
| *** tlian has quit IRC | 21:15 | |
| *** jwitk0 has quit IRC | 21:17 | |
| *** mgoddard_ has joined #openstack-ansible | 21:25 | |
| *** mgoddard__ has joined #openstack-ansible | 21:26 | |
| *** mgoddard has quit IRC | 21:26 | |
| *** mgoddard_ has quit IRC | 21:30 | |
| openstackgerrit | Jesse Pretorius proposed openstack/openstack-ansible: Remove flake8 test from gate-check-commit https://review.openstack.org/247535 | 21:31 |
| openstackgerrit | Jesse Pretorius proposed openstack/openstack-ansible: Add "httpchk /versions" for glance-api haproxy. https://review.openstack.org/248689 | 21:31 |
| *** mgoddard__ has quit IRC | 21:33 | |
| *** phiche has joined #openstack-ansible | 21:35 | |
| *** openstackgerrit has quit IRC | 21:36 | |
| *** openstackgerrit has joined #openstack-ansible | 21:37 | |
| *** mgoddard has joined #openstack-ansible | 21:42 | |
| *** mgoddard_ has joined #openstack-ansible | 21:43 | |
| *** sacharya has quit IRC | 21:44 | |
| *** mgoddard has quit IRC | 21:47 | |
| openstackgerrit | Jesse Pretorius proposed openstack/openstack-ansible: Allow Horizon Custom Branding https://review.openstack.org/248819 | 21:52 |
| *** adac has quit IRC | 22:01 | |
| openstackgerrit | Kevin Carter proposed openstack/openstack-ansible-lxc_container_create: Use sudo for role testing https://review.openstack.org/249749 | 22:08 |
| *** jaypipes has quit IRC | 22:30 | |
| *** phiche has quit IRC | 22:45 | |
| openstackgerrit | Kevin Carter proposed openstack/openstack-ansible-lxc_container_create: Use sudo for role testing https://review.openstack.org/249749 | 22:46 |
| *** phiche has joined #openstack-ansible | 22:49 | |
| *** shanec has joined #openstack-ansible | 22:50 | |
| cloudnull | odyssey4me: still around ? | 22:52 |
| cloudnull | https://review.openstack.org/#/c/249749/ | 22:52 |
| cloudnull | can you bang that through | 22:52 |
| shanec | hey all, anyone from rpc-openstack here that I can ping regarding maas? | 22:53 |
| cloudnull | mattt: you available? you mind pushing this through if possible https://review.openstack.org/#/c/249749/ ? | 22:55 |
| cloudnull | shanec: im w/ rax however idk much about maas. | 22:55 |
| cloudnull | whats going on ? | 22:55 |
| cloudnull | maybe i can help regardless | 22:55 |
| mattt | cloudnull: looking now | 22:56 |
| cloudnull | tyvm | 22:56 |
| *** phiche has quit IRC | 22:57 | |
| mattt | cloudnull: what is the point of that change ? | 23:00 |
| cloudnull | its the last change in a chain to update independent roles to run with sudo privledges | 23:01 |
| cloudnull | the infra tox job executes as a user | 23:01 |
| cloudnull | not as root | 23:01 |
| mattt | k | 23:01 |
| mattt | just realised i also don't have gerrit watching all those new repos :-/ | 23:01 |
| cloudnull | mattt: im using https://review.openstack.org/#/q/project:%255Eopenstack/openstack-ansible.*+status:open,n,z | 23:02 |
| mattt | cool i'll use that, thanks cloudnull | 23:04 |
| mattt | (also +2d the other one) | 23:04 |
| cloudnull | tyvm | 23:04 |
| openstackgerrit | Merged openstack/openstack-ansible-lxc_container_create: Use sudo for role testing https://review.openstack.org/249749 | 23:05 |
| *** sdake has joined #openstack-ansible | 23:10 | |
| mattoliverau | you guys should set up a gerrit dashboard, that makes life easier too. (if you haven't already). | 23:31 |
| *** bangfrog has quit IRC | 23:46 | |
| openstackgerrit | Merged openstack/openstack-ansible: Remove flake8 test from gate-check-commit https://review.openstack.org/247535 | 23:50 |
| *** devl_ has quit IRC | 23:55 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!