| *** masber has quit IRC | 00:33 | |
| *** masber has joined #openstack-ha | 00:33 | |
| *** aasmith has joined #openstack-ha | 00:58 | |
| *** masahito has joined #openstack-ha | 01:11 | |
| *** masahito has quit IRC | 01:13 | |
| *** masahito has joined #openstack-ha | 01:14 | |
| *** kgaillot has quit IRC | 01:24 | |
| *** jmlowe has quit IRC | 01:49 | |
| *** masahito has quit IRC | 01:53 | |
| *** masahito has joined #openstack-ha | 01:53 | |
| *** masahito has quit IRC | 01:59 | |
| *** masahito has joined #openstack-ha | 02:00 | |
| *** markvoelker has quit IRC | 02:09 | |
| *** markvoelker has joined #openstack-ha | 02:09 | |
| *** markvoelker has quit IRC | 02:14 | |
| *** yangyapeng has joined #openstack-ha | 02:31 | |
| *** masahito has quit IRC | 02:43 | |
| *** masahito has joined #openstack-ha | 02:49 | |
| *** masahito has quit IRC | 03:04 | |
| *** masahito has joined #openstack-ha | 04:02 | |
| *** masahito_ has joined #openstack-ha | 04:03 | |
| *** masahito_ has quit IRC | 04:07 | |
| *** masahito has quit IRC | 04:07 | |
| *** markvoelker has joined #openstack-ha | 04:10 | |
| *** hoonetorg has joined #openstack-ha | 04:32 | |
| *** masahito has joined #openstack-ha | 04:35 | |
| *** masahito_ has joined #openstack-ha | 04:37 | |
| *** masahito has quit IRC | 04:40 | |
| *** masahito_ has quit IRC | 04:44 | |
| *** markvoelker has quit IRC | 04:45 | |
| *** masahito has joined #openstack-ha | 04:51 | |
| *** masahito_ has joined #openstack-ha | 04:52 | |
| *** masahito has quit IRC | 04:55 | |
| *** masahito_ has quit IRC | 05:05 | |
| *** masahito has joined #openstack-ha | 05:05 | |
| *** masahito has quit IRC | 05:32 | |
| *** radeks has joined #openstack-ha | 05:34 | |
| *** masahito has joined #openstack-ha | 05:39 | |
| *** masahito_ has joined #openstack-ha | 05:40 | |
| *** markvoelker has joined #openstack-ha | 05:42 | |
| *** masahito has quit IRC | 05:44 | |
| *** masahito_ has quit IRC | 05:45 | |
| *** masahito has joined #openstack-ha | 05:45 | |
| *** yangyapeng has quit IRC | 05:46 | |
| *** yangyapeng has joined #openstack-ha | 05:47 | |
| *** masahito has quit IRC | 05:47 | |
| *** masahito has joined #openstack-ha | 05:47 | |
| *** masahito has quit IRC | 05:49 | |
| *** masahito has joined #openstack-ha | 05:49 | |
| *** radeks has quit IRC | 05:51 | |
| *** masahito has quit IRC | 05:54 | |
| *** masahito has joined #openstack-ha | 05:55 | |
| *** masahito has quit IRC | 06:02 | |
| *** masahito has joined #openstack-ha | 06:02 | |
| *** masahito_ has joined #openstack-ha | 06:07 | |
| *** masahito_ has quit IRC | 06:08 | |
| *** masahito has quit IRC | 06:08 | |
| *** masahito has joined #openstack-ha | 06:08 | |
| *** masahito_ has joined #openstack-ha | 06:10 | |
| *** masahito has quit IRC | 06:12 | |
| *** markvoelker has quit IRC | 06:14 | |
| *** masahito_ has quit IRC | 06:23 | |
| *** masahito has joined #openstack-ha | 06:48 | |
| *** radeks has joined #openstack-ha | 07:02 | |
| *** markvoelker has joined #openstack-ha | 07:11 | |
| *** dgurtner has joined #openstack-ha | 07:24 | |
| *** tnarg has quit IRC | 07:36 | |
| *** masahito has quit IRC | 07:38 | |
| *** markvoelker has quit IRC | 07:45 | |
| *** rmart04 has joined #openstack-ha | 08:04 | |
| *** masahito has joined #openstack-ha | 08:10 | |
| *** masahito_ has joined #openstack-ha | 08:19 | |
| *** egonzalez has joined #openstack-ha | 08:21 | |
| *** masahito has quit IRC | 08:22 | |
| *** masahito_ has quit IRC | 08:25 | |
| *** masahito has joined #openstack-ha | 08:25 | |
| *** markvoelker has joined #openstack-ha | 08:42 | |
| *** MatthewLi has quit IRC | 08:48 | |
| *** MatthewLi has joined #openstack-ha | 08:49 | |
| *** MatthewLi has quit IRC | 08:49 | |
| *** masahito has quit IRC | 08:50 | |
| *** MatthewLi has joined #openstack-ha | 08:51 | |
| *** masahito has joined #openstack-ha | 09:01 | |
| *** masahito_ has joined #openstack-ha | 09:03 | |
| *** masahito has quit IRC | 09:05 | |
| *** masahito_ has quit IRC | 09:12 | |
| *** masahito has joined #openstack-ha | 09:12 | |
| *** masahito has quit IRC | 09:13 | |
| *** masahito has joined #openstack-ha | 09:13 | |
| *** masahito has quit IRC | 09:15 | |
| *** masahito has joined #openstack-ha | 09:16 | |
| *** markvoelker has quit IRC | 09:16 | |
| *** MatthewLi has quit IRC | 09:34 | |
| *** masahito has quit IRC | 09:39 | |
| *** markvoelker has joined #openstack-ha | 10:12 | |
| *** yangyapeng has quit IRC | 10:43 | |
| *** markvoelker has quit IRC | 10:45 | |
| *** markvoelker has joined #openstack-ha | 11:43 | |
| *** rossella__ has quit IRC | 11:50 | |
| *** rossella__ has joined #openstack-ha | 11:53 | |
| *** rossella__ has quit IRC | 11:59 | |
| *** rossella__ has joined #openstack-ha | 12:02 | |
| *** markvoelker has quit IRC | 12:16 | |
| *** markvoelker has joined #openstack-ha | 12:26 | |
| *** jmlowe has joined #openstack-ha | 12:39 | |
| *** jmlowe has quit IRC | 12:40 | |
| *** MatthewLi has joined #openstack-ha | 13:01 | |
| *** yangyapeng has joined #openstack-ha | 13:10 | |
| *** catintheroof has joined #openstack-ha | 13:25 | |
| *** catinthe_ has joined #openstack-ha | 13:26 | |
| *** jmlowe has joined #openstack-ha | 13:30 | |
| *** catintheroof has quit IRC | 13:30 | |
| *** Qiming has quit IRC | 13:46 | |
| *** Qiming has joined #openstack-ha | 13:50 | |
| *** rossella__ has quit IRC | 13:52 | |
| *** rossella__ has joined #openstack-ha | 13:54 | |
| *** radeks has quit IRC | 14:13 | |
| *** jmlowe has quit IRC | 14:15 | |
| *** kgaillot has joined #openstack-ha | 14:24 | |
| *** jmlowe has joined #openstack-ha | 14:28 | |
| *** zzhou_ has quit IRC | 14:33 | |
| *** rossella__ has quit IRC | 14:41 | |
| *** rossella__ has joined #openstack-ha | 14:43 | |
| *** rmart04 has quit IRC | 14:59 | |
| tbarron | i'm looking for a pcs fencing agent for openstack/nova where the target of the | 15:02 |
|---|---|---|
| tbarron | stonith operation is not an OpenStack compute node but rather a guest VM | 15:02 |
| tbarron | so as I (mis?) understand the matter, not fence_compute | 15:02 |
| tbarron | Anyone know of such a thing? | 15:02 |
| tbarron | basically, I want to interface with a fence_openstack_guest_vm device like | 15:03 |
| *** zzhou has joined #openstack-ha | 15:03 | |
| tbarron | interfacing with a fencing agent for ilmi or idrac but have it issue openstack nova | 15:03 |
| tbarron | commands to control the life-cycle of some "service" VMs for manila | 15:04 |
| tbarron | these service VMs provide NFS service to backend storage and currently | 15:04 |
| tbarron | have no HA and are therefore not suitable for enterprise production OpenStack | 15:05 |
| tbarron | distributions | 15:05 |
| *** hoonetorg has quit IRC | 15:05 | |
| tbarron | the idea is to run pairs of these SVMs active/hot-standby | 15:05 |
| tbarron | rather than just resurrecting a single SVM as boot up time will take significantly | 15:06 |
| tbarron | longer than failover to a hot-standby | 15:06 |
| *** MatthewLi has quit IRC | 15:06 | |
| aspiers | tbarron: it sounds like you are trying to reinvent part of masakari or senlin | 15:27 |
| aspiers | fencing is not required for managing individual VMs, because you can simply control them via nova | 15:29 |
| aspiers | tbarron: I'd be more than happy to discuss this problem with you further. It's firmly in the compute HA space which I've been working on the last couple of years | 15:30 |
| tbarron | aspiers: I've recently watched you-tubes of your summit presos | 15:31 |
| aspiers | tbarron: oh ok cool, I was about to send you links ;-) | 15:31 |
| aspiers | so one option is just to use masakari's vmmonitor | 15:31 |
| tbarron | aspiers: my thinking, correct or wrong, is that since these are "service" VMs, that is | 15:31 |
| tbarron | images running known, specialized software, the problem space is simpler than | 15:32 |
| tbarron | what masakari solves | 15:32 |
| aspiers | I'm not sure but I *think* I persuaded your colleague Andrew Beekhof that masakari is worth pursuing | 15:32 |
| tbarron | that is, i can run pcs/corosync in the service VMs themselves w/o any need of remote agents, etc. | 15:32 |
| aspiers | yes, masakari solves a lot more than this, but it's nicely modular so you don't have to bring in all the components if you don't want to | 15:32 |
| tbarron | aspiers: good tip on Beekhof, I will check with him for sure | 15:33 |
| aspiers | well you *could* place Pacemaker inside the VMs, but personally I'm not convinced that's the best approach | 15:33 |
| tbarron | b/c as you know we aren't using masakari currently | 15:33 |
| aspiers | right | 15:33 |
| aspiers | Pacemaker has a mostly in-band approach to management, with the fencing devices being the out-of-band exception | 15:34 |
| aspiers | but in this case you already have nova for out-of-band management | 15:34 |
| tbarron | i'm not sure it's the best approach either, but I want to understand how to do it | 15:34 |
| aspiers | well, you'd have to write your own fencing device | 15:34 |
| tbarron | without putting a nova client and keystone client inside the service VMs | 15:34 |
| aspiers | https://github.com/ClusterLabs/fence-agents | 15:34 |
| tbarron | aspiers: yeah, I'll study that | 15:35 |
| aspiers | if you don't put those clients inside, then you'd need some service on the outside for them to talk to | 15:35 |
| tbarron | aspiers: personally I think the masakari stuff looks really cool and there is clearly a need | 15:35 |
| aspiers | which sounds even uglier | 15:35 |
| tbarron | for general purpose "pet" VM management | 15:35 |
| aspiers | right | 15:35 |
| aspiers | but actually this particular use case (i.e. service VMs) is a very interesting one which I hadn't thought of before | 15:36 |
| aspiers | and it definitely ties into the architectural thoughts I have been having recently | 15:36 |
| tbarron | well, the "ugly" thing I was thinking about was an external fencing agent | 15:36 |
| aspiers | ah no, the fencing agent needs to live inside the Pacemaker cluster | 15:36 |
| tbarron | external virtual power cycler I should have said | 15:36 |
| aspiers | like I said, if you want the heavy lifting to be done outside, then the fencing agent would become a proxy for the outside mechanism | 15:37 |
| aspiers | but when Pacemaker wants to fence something, initially it has to invoke something inside the cluster | 15:37 |
| tbarron | that one could access via ssh key instead of keystone credentials, after somebody with the | 15:37 |
| tbarron | right keystone credentials configured it for *this* nova VM | 15:37 |
| aspiers | and to me introducing a proxy sounds like a fundamentally flawed architecture, because if the connection between the inside and outside is lost, you're totally hosed | 15:37 |
| * tbarron will reflect on that point | 15:38 | |
| aspiers | tbarron: if it's easier, we could talk on Red Hat's fancy video conferencing thingy (I forget the name) or Google Hangouts or whatever | 15:39 |
| tbarron | aspiers: that would be great (bluejeans) or hangouts - I will email you and check availability | 15:39 |
| aspiers | presumably SUSE will have just as strong a need for HA service VMs as you guys, so it's not an entirely selfless offer ;-) | 15:40 |
| aspiers | yeah bluejeans, that's the one | 15:40 |
| tbarron | aspiers: yeah, say hi to toabctl for me :D | 15:40 |
| aspiers | haha OK :) | 15:40 |
| aspiers | of course ... the manila connection | 15:40 |
| aspiers | not sure if you saw my most recent talk from Israel? this one is the most accurate in terms of architecture https://youtu.be/uMCMDF9VkYk | 15:41 |
| aspiers | in the previous ones I think I slightly messed up the diagrams | 15:41 |
| aspiers | https://aspiers.github.io/openstack-day-israel-2017-compute-ha/#/nova-host-alerter | 15:42 |
| aspiers | tbarron: also, I don't know too much about Senlin, but this use case does potentially sound like a good match for it https://wiki.openstack.org/wiki/Senlin | 15:44 |
| aspiers | tbarron: Qiming could probably comment on whether I'm right | 15:44 |
| aspiers | in fact for this particular use case, Senlin might be a better choice than masakari's vmmonitor | 15:45 |
| tbarron | aspiers: I hadn't seen the Israel talk, will view it in a a little bit, thanks! | 15:49 |
| aspiers | :) | 15:49 |
| tbarron | Qiming: aspiers I don't know much about Senlin, will do some homework | 15:50 |
| tbarron | aspiers: we had dinner together in Atlanta but this issue wasn't really on my radar at the time ... | 15:52 |
| aspiers | ah ok | 15:53 |
| aspiers | tbarron: I'm intending to propose a new Working Group for covering these kinds of architectural topics | 15:54 |
| tbarron | aspiers: makes sense | 15:57 |
| aspiers | actually we have a talk submitted for Sydney trying to tie all these projects together to some extent | 15:57 |
| tbarron | i'm on a manila topic proposed but don't know whether I'll be going to Sydney at this point. Might rely on you tube again :D | 15:59 |
| aspiers | yeah I don't know whether I'll make it either | 15:59 |
| *** egonzalez has quit IRC | 16:16 | |
| *** dougs1 has quit IRC | 16:49 | |
| -openstackstatus- NOTICE: Restarting Gerrit for our weekly memory leak cleanup. | 17:05 | |
| -openstackstatus- NOTICE: docs.o.o is currently broken, we're investigating | 18:38 | |
| *** dgurtner has quit IRC | 18:51 | |
| *** catinthe_ has quit IRC | 19:58 | |
| *** jmlowe has quit IRC | 20:32 | |
| *** v12aml has quit IRC | 21:30 | |
| *** markvoelker has quit IRC | 23:39 | |
| *** catintheroof has joined #openstack-ha | 23:58 | |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!