| *** cknight has joined #openstack-ha | 00:00 | |
| *** smoriya has quit IRC | 00:50 | |
| *** smoriya_afk has joined #openstack-ha | 00:52 | |
| *** smoriya_afk is now known as smoriya | 00:53 | |
| *** cknight has quit IRC | 01:18 | |
| *** masahito has joined #openstack-ha | 01:21 | |
| *** masahito has quit IRC | 01:26 | |
| *** masahito has joined #openstack-ha | 01:41 | |
| *** masahito has quit IRC | 01:54 | |
| *** smoriya has quit IRC | 02:00 | |
| *** masahito has joined #openstack-ha | 02:03 | |
| *** masahito has quit IRC | 02:23 | |
| *** masahito has joined #openstack-ha | 02:24 | |
| *** masahito has quit IRC | 02:44 | |
| *** smoriya_afk has joined #openstack-ha | 03:00 | |
| *** smoriya_afk is now known as smoriya | 03:00 | |
| *** ianbrown has joined #openstack-ha | 04:00 | |
| *** masahito has joined #openstack-ha | 04:07 | |
| *** ianbrown has quit IRC | 04:15 | |
| *** ianbrown_ has joined #openstack-ha | 04:15 | |
| *** masahito has quit IRC | 04:25 | |
| *** ianbrown_ is now known as ianbrown | 04:28 | |
| *** cknight has joined #openstack-ha | 04:38 | |
| *** ianbrown has quit IRC | 04:42 | |
| *** ianbrown has joined #openstack-ha | 04:42 | |
| *** ianbrown has quit IRC | 05:29 | |
| *** ianbrown has joined #openstack-ha | 05:29 | |
| *** cknight has quit IRC | 05:30 | |
| *** masahito has joined #openstack-ha | 05:51 | |
| *** masahito has quit IRC | 06:08 | |
| *** masahito has joined #openstack-ha | 06:14 | |
| *** ianbrown has quit IRC | 06:26 | |
| *** ianbrown has joined #openstack-ha | 06:26 | |
| *** ianbrown has quit IRC | 06:30 | |
| *** ianbrown has joined #openstack-ha | 06:30 | |
| *** ianbrown has quit IRC | 06:31 | |
| *** ianbrown has joined #openstack-ha | 06:32 | |
| *** masahito has quit IRC | 06:38 | |
| *** openstackgerrit_ has joined #openstack-ha | 06:52 | |
| *** mjura has joined #openstack-ha | 06:54 | |
| *** masahito has joined #openstack-ha | 07:02 | |
| *** masahito has quit IRC | 07:29 | |
| *** masahito has joined #openstack-ha | 07:31 | |
| *** vuntz has quit IRC | 07:39 | |
| *** vuntz has joined #openstack-ha | 07:40 | |
| *** vuntz has quit IRC | 07:40 | |
| *** vuntz has joined #openstack-ha | 07:40 | |
| *** mobaer has joined #openstack-ha | 07:43 | |
| *** ianbrown_ has joined #openstack-ha | 07:55 | |
| *** ianbrown has quit IRC | 07:58 | |
| *** jpena has joined #openstack-ha | 08:06 | |
| *** masahito has quit IRC | 08:09 | |
| *** masahito has joined #openstack-ha | 08:14 | |
| *** dgurtner has joined #openstack-ha | 08:18 | |
| *** dgurtner has joined #openstack-ha | 08:18 | |
| *** ianbrown_ has quit IRC | 08:19 | |
| *** ianbrown_ has joined #openstack-ha | 08:23 | |
| *** ianbrown_ has quit IRC | 08:51 | |
| aspiers | morning all | 08:53 |
|---|---|---|
| _gryf | hi | 08:54 |
| ddeja | hello | 08:55 |
| masahito | hi | 08:55 |
| *** ianbrown has joined #openstack-ha | 08:56 | |
| aspiers | everyone -> #openstack-meeting :-) | 08:59 |
| aspiers | #startmeeting ha | 09:00 |
| openstack | Meeting started Mon Dec 7 09:00:12 2015 UTC and is due to finish in 60 minutes. The chair is aspiers. Information about MeetBot at http://wiki.debian.org/MeetBot. | 09:00 |
| openstack | Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. | 09:00 |
| openstack | The meeting name has been set to 'ha' | 09:00 |
| *** kazuIchikawa has joined #openstack-ha | 09:00 | |
| aspiers | Welcome everyone :-) | 09:00 |
| aspiers | Let's start with some quick status updates | 09:00 |
| masahito | aspiers: here is wrong channel I think | 09:00 |
| ddeja | wrong channel? | 09:00 |
| aspiers | haha I'm an idiot | 09:01 |
| aspiers | #endmeeting | 09:01 |
| openstack | Meeting ended Mon Dec 7 09:01:15 2015 UTC. Information about MeetBot at http://wiki.debian.org/MeetBot . (v 0.1.4) | 09:01 |
| openstack | Minutes: http://eavesdrop.openstack.org/meetings/ha/2015/ha.2015-12-07-09.00.html | 09:01 |
| openstack | Minutes (text): http://eavesdrop.openstack.org/meetings/ha/2015/ha.2015-12-07-09.00.txt | 09:01 |
| openstack | Log: http://eavesdrop.openstack.org/meetings/ha/2015/ha.2015-12-07-09.00.log.html | 09:01 |
| *** vuntz has quit IRC | 09:11 | |
| *** ianbrown has quit IRC | 09:40 | |
| *** vuntz has joined #openstack-ha | 09:41 | |
| *** vuntz has quit IRC | 09:41 | |
| *** vuntz has joined #openstack-ha | 09:41 | |
| *** ianbrown has joined #openstack-ha | 09:45 | |
| beekhof | aspiers: what it does is trivial yet manages to be unreliable | 10:00 |
| beekhof | every time i think i've worked around the last systemd issue, i find another one | 10:00 |
| beekhof | 2 since friday | 10:00 |
| beekhof | not even exagerating | 10:00 |
| aspiers | beekhof: well it was invented by a RH guy, so go and shout at him ;-) | 10:00 |
| beekhof | i have | 10:01 |
| *** dguitarbite has quit IRC | 10:01 | |
| beekhof | aspiers: actually his partner in crime is ex-suse labs | 10:01 |
| aspiers | so we can conclude that all brokenness always stems from RH and ex-SUSE, right? ;-P | 10:02 |
| aspiers | SCNR :) | 10:02 |
| beekhof | i think he was still suse at the time | 10:02 |
| beekhof | now he is ex :) | 10:02 |
| aspiers | haha :) | 10:03 |
| beekhof | nice guy, shame he fell in with a bad crowd ;-) | 10:03 |
| aspiers | lol | 10:03 |
| beekhof | anyway, i get that the values in the unit files are interesting | 10:03 |
| beekhof | but i reject that anything systemd does with them has value | 10:03 |
| aspiers | I worry that it's not just the values though | 10:03 |
| beekhof | there may be a way forward though | 10:04 |
| aspiers | well the problem is that, unless you're working on the rpm packaging project and trying to ensure distros converge, you can't safely say that | 10:04 |
| beekhof | that makes everyone happy | 10:04 |
| beekhof | i eluded to it earlier | 10:04 |
| aspiers | yeah? | 10:04 |
| beekhof | there may be a way for the lrmd to be a | 10:04 |
| * beekhof looks up the word | 10:05 | |
| *** dguitarbite has joined #openstack-ha | 10:05 | |
| beekhof | systemd scope | 10:05 |
| beekhof | https://wiki.freedesktop.org/www/Software/systemd/dbus/ (section Scope | 10:05 |
| beekhof | Unit Objects -> Controller) | 10:05 |
| beekhof | Scopes are units which manages group of processes. | 10:05 |
| beekhof | essentially pacemaker would take the responsibility of managing those unit files | 10:06 |
| beekhof | i believe | 10:06 |
| aspiers | oooh | 10:06 |
| aspiers | I think that achieves the goals of what I was suggesting, except with a much nicer native implementation | 10:07 |
| aspiers | but wait | 10:07 |
| aspiers | how is that different to just using a systemd:* primitive? | 10:07 |
| * beekhof proposes we hold off on an OpenstackAnything RA until we can figure out if this scope thing will work out | 10:07 | |
| beekhof | much more control | 10:07 |
| beekhof | i believe | 10:07 |
| beekhof | so we'd not be fighting systemd | 10:08 |
| aspiers | and if the lrmd talks dbus directly, where does the monitor op get implemented? | 10:08 |
| beekhof | good question | 10:08 |
| beekhof | -ENOIDEAYET | 10:08 |
| aspiers | why would this avoid fighting systemd? it's just using a different interface to it? | 10:08 |
| beekhof | it may be a dead end | 10:09 |
| aspiers | albeit a nicer one than systemctl | 10:09 |
| beekhof | no | 10:09 |
| beekhof | thats the thing, systemctl would redirect to us | 10:09 |
| aspiers | I probably don't understand your suggestion yet | 10:09 |
| beekhof | "oh, thats for pacemaker" | 10:09 |
| beekhof | *punt* | 10:09 |
| beekhof | that sort of thing | 10:09 |
| beekhof | if it works as advertised | 10:10 |
| beekhof | IF | 10:10 |
| aspiers | OK, so how would the params get shared between HA and non-HA scenarios? | 10:10 |
| beekhof | i believe the unit files would still be involved somehow | 10:10 |
| * beekhof has to head out | 10:11 | |
| aspiers | and I still don't understand how an interface to systemd helps us make systemd delegate stuff elsewhere | 10:11 |
| aspiers | OK | 10:11 |
| beekhof | but lets continue again soon | 10:11 |
| aspiers | maybe you can mull over it and write something up at some point? | 10:11 |
| aspiers | sure, very happy to discuss any time! | 10:11 |
| beekhof | it may present new opportunities | 10:11 |
| aspiers | indeed | 10:11 |
| ddeja | guys, sorry for interupting - but could you shortly descibe what's the problem you are trying to resolve? | 10:11 |
| ddeja | I'm getting lost ;) | 10:11 |
| beekhof | i'm hopeful it wont be worse | 10:11 |
| aspiers | ddeja: I can summarise whilst beekhof heads out | 10:11 |
| aspiers | beekhof: I may well be revisiting the host/nova name mapping thing this week or next | 10:12 |
| aspiers | so we can continue that too | 10:12 |
| beekhof | k | 10:12 |
| beekhof | i'll be looking at it tomorrow | 10:12 |
| aspiers | cool | 10:12 |
| aspiers | thanks, talk soon then! | 10:12 |
| beekhof | cya | 10:12 |
| aspiers | ddeja: so the main problems are | 10:12 |
| *** ianbrown has quit IRC | 10:13 | |
| aspiers | 1. duplication of data/logic for starting/stopping services (between systemd service files and OCF RAs) | 10:13 |
| *** ianbrown has joined #openstack-ha | 10:13 | |
| aspiers | 2. achieving something which is distro-agnostic | 10:13 |
| aspiers | 3. ensuring that OCF RAs provide good application-level monitoring of services | 10:14 |
| aspiers | my view is that distro-specific stuff should be dealt with by the distro (typically via packaging) | 10:14 |
| aspiers | even though there is a separate project to converge packaging, that won't happen overnight | 10:14 |
| ddeja | OK, thanks | 10:14 |
| aspiers | so my idea to solve all of this was to change the RAs so that when they need to start/stop services, they simply call service(8) which is distro-agnostic | 10:15 |
| aspiers | but beekhof hates that idea :-/ | 10:15 |
| ddeja | Me and _gryf have similar idea in May I think | 10:16 |
| aspiers | of course I'm keen to hear other ideas, but I thinking we are still missing implementation details for any alternative approaches | 10:16 |
| aspiers | my idea would make the RAs *much* thinner | 10:16 |
| aspiers | effectively all they would do is add the monitor action | 10:17 |
| aspiers | and delegate start/stop/status to service(8) | 10:17 |
| ddeja | That should work | 10:17 |
| aspiers | so, easier to maintain, simpler to understand, work with any distro, and consistent behaviour with non-HA environments | 10:17 |
| ddeja | As I remember, beekhof was complaining about not very good monitoring by systemd, but if we provide our own monitor... | 10:18 |
| ddeja | I'll go that way | 10:18 |
| aspiers | but beekhof is pushing hard against it for reasons I don't fully understand yet, but he's an expert so I need to work to understand his objections | 10:18 |
| aspiers | I think beekhof is mainly complaining about relying on systemd at all | 10:18 |
| aspiers | he said systemctl stop is not reliable | 10:19 |
| ddeja | Sadly, that's true | 10:19 |
| aspiers | which could of course be a big problem | 10:19 |
| ddeja | I have an issue with it once ;) | 10:19 |
| aspiers | but IMHO the right approach is to fix systemd | 10:19 |
| aspiers | if we can't rely on systemd then we have a lot of other problems too | 10:19 |
| ddeja | +1 | 10:20 |
| ddeja | OK, I just wanted to understand what are you guys working on (to be able to help), so thanks for explenations :) | 10:21 |
| *** openstackgerrit_ has quit IRC | 10:23 | |
| aspiers | sure :) | 10:31 |
| aspiers | http://ourobengr.com/2015/12/pets-vs-cattle/ :-) | 11:03 |
| *** bogdando has quit IRC | 11:35 | |
| *** masahito has quit IRC | 11:37 | |
| *** bogdando has joined #openstack-ha | 11:49 | |
| *** jpena is now known as jpena|lunch | 12:29 | |
| *** smoriya_ has quit IRC | 13:01 | |
| *** kazuIchikawa has quit IRC | 13:02 | |
| *** jpena|lunch is now known as jpena | 13:40 | |
| *** cknight has joined #openstack-ha | 13:55 | |
| *** kazuIchikawa has joined #openstack-ha | 13:59 | |
| *** kazuIchikawa has quit IRC | 14:03 | |
| *** cknight has quit IRC | 14:22 | |
| *** mobaer has quit IRC | 14:33 | |
| *** mobaer has joined #openstack-ha | 14:38 | |
| *** kgaillot has joined #openstack-ha | 14:42 | |
| *** mobaer has quit IRC | 14:42 | |
| *** raginbajin has quit IRC | 14:44 | |
| *** raginbajin has joined #openstack-ha | 14:46 | |
| *** cknight has joined #openstack-ha | 14:54 | |
| *** smoriya has quit IRC | 14:54 | |
| *** smoriya_afk has joined #openstack-ha | 14:56 | |
| *** smoriya_afk is now known as smoriya | 14:56 | |
| *** mjura has quit IRC | 15:21 | |
| *** sigmavirus24_awa is now known as sigmavirus24 | 15:22 | |
| *** smoriya_ has joined #openstack-ha | 15:40 | |
| *** mobaer has joined #openstack-ha | 16:45 | |
| *** FL1SK has quit IRC | 16:46 | |
| *** jpena has quit IRC | 16:57 | |
| *** mobaer has quit IRC | 17:31 | |
| *** dgurtner has quit IRC | 17:44 | |
| *** smoriya_ has quit IRC | 18:12 | |
| *** FL1SK has joined #openstack-ha | 18:57 | |
| *** sigmavirus24 is now known as sigmavirus24_awa | 19:04 | |
| *** sigmavirus24_awa is now known as sigmavirus24 | 19:05 | |
| *** mobaer has joined #openstack-ha | 20:05 | |
| *** cknight has quit IRC | 20:15 | |
| *** cknight has joined #openstack-ha | 21:01 | |
| *** cknight has quit IRC | 21:21 | |
| *** royoung1 has joined #openstack-ha | 21:37 | |
| *** ianbrown has quit IRC | 21:39 | |
| *** ianbrown has joined #openstack-ha | 21:39 | |
| *** royoung has quit IRC | 21:40 | |
| *** royoung1 has quit IRC | 21:57 | |
| *** royoung has joined #openstack-ha | 22:00 | |
| *** vuntz has quit IRC | 22:07 | |
| *** vuntz has joined #openstack-ha | 22:08 | |
| *** vuntz has quit IRC | 22:08 | |
| *** vuntz has joined #openstack-ha | 22:08 | |
| *** cknight has joined #openstack-ha | 22:13 | |
| *** royoung has quit IRC | 22:35 | |
| *** ianbrown has quit IRC | 23:12 | |
| *** ianbrown has joined #openstack-ha | 23:12 | |
| *** kazuIchikawa has joined #openstack-ha | 23:31 | |
| *** mobaer has quit IRC | 23:33 | |
| *** kazuIchikawa has quit IRC | 23:36 | |
| *** ianbrown has quit IRC | 23:41 | |
| *** ianbrown has joined #openstack-ha | 23:42 | |
| *** cknight has quit IRC | 23:55 | |
| *** kgaillot has quit IRC | 23:56 | |
| *** sigmavirus24 is now known as sigmavirus24_awa | 23:57 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!