04:02:11 #startmeeting masakari 04:02:12 Meeting started Tue Jan 17 04:02:11 2017 UTC and is due to finish in 60 minutes. The chair is samP_. Information about MeetBot at http://wiki.debian.org/MeetBot. 04:02:13 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 04:02:16 The meeting name has been set to 'masakari' 04:02:30 ok, hi everyone 04:02:33 o/ 04:02:37 o/ 04:02:49 Hi 04:03:07 since we don"t have critical bugs.. lets jump in to discussion points 04:03:40 OK. 04:03:54 last week it was action item for me to writedown about ironc HA 04:04:15 so, I put wrote some words about it. 04:04:31 #link https://etherpad.openstack.org/p/baremetal-ha 04:04:40 thanks. 04:05:53 To implement this we need ironic instance boot from cinder vols. 04:06:13 #link https://blueprints.launchpad.net/nova/+spec/ironic-boot-from-volume 04:06:49 ah.. forgot to change the topic 04:07:36 I will start understanding about Ironic so that we can implement this use case 04:08:12 samP_: do you know any shcedule about implementing the feature, ironic-boot-from-volume in nova project? 04:08:56 The implementation is in blocked state: https://blueprints.launchpad.net/nova/+spec/ironic-boot-from-volume 04:09:11 It seems that blueprint for nova is currently blocked, and we still need some changes, right? 04:09:43 takashi: they are holding it till spec from ironic side to merge. and it will done after refactoring in ironic 04:09:55 samP_: ok 04:10:22 I just found nova spec propsed for Pike: https://review.openstack.org/#/c/311696/ 04:11:12 takashi: there was some discussion in ironic team to remove this block and start code implementation. 04:11:49 samP_: it is great if we can get some chance to discuss with ironic team in PTG, covering their plan about bfv 04:11:50 however, there will be anther discussion in PTG 04:12:03 samP_: ok 04:12:18 takashi: sure, I will try to get it 04:14:45 I will add more details for ironic HA and container HA on that etherpad. please put you comment, questions and ideas on it too.. 04:14:55 sam 04:15:06 takashi: yes 04:16:03 samP_: can we not add this ha support for ironic instance in masakari if it's booted from image? 04:17:22 tpatil: we can, but if it boot from a local disk ,then you can not have the same instance after it failover to a new one 04:17:56 tpatil: "can we ""not"" add " 04:18:32 samP_: But if you are using shared storage to store local disks, it's possible, correct? 04:19:43 tpatil: it would be possible, but I am not sure you can do it with current ironic/nova APIs 04:20:20 samP_: Ok, I will check current Ironic/Nova APIs to figure that out 04:21:28 tpatil: IMO it can be possible that ironic does not support shared storage for 'image store' 04:21:41 tpatil: there will be ways around. its all about cutoff the old instance and attaching the same disk to new instance 04:22:52 samP_: right 04:22:53 takashi: correct 04:23:04 samP_: Will check this point 04:23:10 sure.. 04:23:52 then shall we move to next topic, if no more questions or comments about this 04:24:42 samP_: yes 04:24:49 Sure. 04:25:09 what would be the next topic? 04:25:44 Can I talk about AOB? 04:25:57 rkmrHonjo: sure 04:26:04 #topic AOB 04:26:05 thanks. 04:26:18 tpatil: You and abhishek said that you were working about https://review.openstack.org/#/c/397064/ last week. 04:26:25 Any update on this? 04:26:40 Abhishek is working on this issue 04:28:32 rkmrHonjo: He is working on your patch to see how to handle all pending requests before gracefully exiting the child process 04:29:13 tpati: OK, thanks. 04:29:28 rkmrHonjo: It's requires lot of refactoring, so it's taking some time 04:29:43 tpatil: In gerrit you mentioned "service should use ServiceLauncher instead of ProcessLaunche", is this because of the normal openstack way or any other reason? 04:29:50 tpatil: One question. Is current monitor supports gracefull shutdown? 04:30:01 s/Is/Does/ s/supports/support/ 04:30:07 many mistakes, sorry... 04:30:09 takashi: No 04:30:37 samP_: Considering future plan, it's ok to use ProcessLauncher 04:31:09 tpatil: ok. Does it make sense to split the patch to two, like adding signal handlers and graceful shutdown support? 04:31:09 tpatil: ok, thanks 04:31:53 takashi: I don't think there is a need to handle signal handler in child process 04:31:57 samP: In my understanding, it's the normal openstack way. But takahara wrote the reason of using ProcessLauncher on gerrit. 04:32:32 tpatil: If it requires big effort for gracefull shutdown, it can be good if we can land signal handler support first, IMO. 04:32:50 rkmrHonjo: thanks. I read it 04:33:30 I don't mean that we should land current patch, if we have better solution for signal handling, 04:33:53 but I'm afraid to leave current signal handling problem for long time, on the other hand. 04:34:17 takashi: Let's take that decision in the next week depending on the progress of refactoring the existing patch 04:34:26 tpatil: makes sense 04:34:30 takashi: I agree with your opinion. graceful shutdown is other problem. 04:35:35 tpatil: make sense, lets discuss next action on next meeting 04:35:54 samP_: Ok 04:37:52 Ok then, please share the refac status on the existing patch and we can discuss on next meeting on how to proceed with this 04:38:32 samP_: Sure 04:39:20 OK then, any other topics? 04:39:38 no. 04:41:47 Nothing from my side 04:42:50 Me neither 04:43:02 no 04:43:12 Ok then, thank you all.. 04:43:29 let end this... thank you again 04:43:37 bye.. 04:43:38 thanks to all 04:43:40 Bye 04:43:50 have a nice day! 04:44:04 takashi: you too 04:44:05 thanks. 04:44:08 #endmeeting