| *** boveir has joined #openstack-powervm | 00:08 | |
| *** boveir has quit IRC | 00:10 | |
| *** manous_ has joined #openstack-powervm | 01:27 | |
| *** manous_ has quit IRC | 01:37 | |
| *** AlexeyAbashkin has joined #openstack-powervm | 06:37 | |
| *** AlexeyAbashkin has quit IRC | 06:58 | |
| *** AlexeyAbashkin has joined #openstack-powervm | 07:41 | |
| *** AlexeyAbashkin has quit IRC | 07:46 | |
| *** k0da has joined #openstack-powervm | 10:54 | |
| *** manous has joined #openstack-powervm | 12:16 | |
| *** apearson has joined #openstack-powervm | 12:35 | |
| openstackgerrit | prashkre proposed openstack/nova-powervm master: Return iSCSI Initiatos for all VIOSes https://review.openstack.org/557800 | 12:45 |
|---|---|---|
| *** edmondsw_ has joined #openstack-powervm | 12:46 | |
| *** edmonds__ has joined #openstack-powervm | 12:47 | |
| *** edmondsw has quit IRC | 12:50 | |
| *** edmondsw_ has quit IRC | 12:51 | |
| *** edmondsw has joined #openstack-powervm | 12:54 | |
| *** efried is now known as fried_rice | 12:55 | |
| *** edmondsw_ has joined #openstack-powervm | 12:56 | |
| *** edmonds__ has quit IRC | 12:56 | |
| *** edmondsw has quit IRC | 12:59 | |
| *** prashkre has joined #openstack-powervm | 13:00 | |
| openstackgerrit | prashkre proposed openstack/nova-powervm master: WIP: Return iSCSI Initiatos for all VIOSes https://review.openstack.org/557800 | 13:11 |
| fried_rice | prashkre: Are you WIPing that because you're still testing? | 13:13 |
| prashkre | fried_rice: Auto UT's are failing, looking at them | 13:14 |
| fried_rice | prashkre: Are you running the tests locally? | 13:14 |
| prashkre | fried_rice: yes. I am running locally. | 13:15 |
| fried_rice | k | 13:15 |
| edmondsw_ | fried_rice I added my own comments on those iscsi patches | 13:35 |
| *** tjakobs has joined #openstack-powervm | 13:38 | |
| *** edmondsw_ is now known as edmondsw | 13:39 | |
| *** prashkre has quit IRC | 13:44 | |
| *** prashkre has joined #openstack-powervm | 13:53 | |
| *** esberglu has joined #openstack-powervm | 14:00 | |
| *** prashkre has quit IRC | 14:05 | |
| *** prashkre has joined #openstack-powervm | 14:09 | |
| *** prashkre has quit IRC | 14:21 | |
| *** prashkre has joined #openstack-powervm | 14:22 | |
| *** AlexeyAbashkin has joined #openstack-powervm | 14:45 | |
| *** AlexeyAbashkin has quit IRC | 14:49 | |
| *** openstackgerrit has quit IRC | 14:50 | |
| *** prashkre has quit IRC | 15:07 | |
| *** k0da has quit IRC | 15:18 | |
| *** prashkre has joined #openstack-powervm | 15:35 | |
| catmando | hey all | 15:52 |
| catmando | anyone about? | 15:52 |
| fried_rice | catmando: Sort of, for a little while, then out for a couple hours. What's up? | 16:08 |
| catmando | is there any channel where svc people hang out? | 16:17 |
| catmando | (i am assuming you work for ibm) | 16:17 |
| fried_rice | catmando: If there was, it would be here or possibly #openstack-cinder. Let me ask... | 16:37 |
| fried_rice | catmando: (yes, IBM) | 16:38 |
| fried_rice | btw, by "would be here" I meant "but nobody SVC hangs around here, that I know about" | 16:38 |
| fried_rice | catmando: Okay, Gerald confirms that, if they're anywhere, it's #openstack-cinder | 16:39 |
| fried_rice | ...but they'll be in Chinese time zones for the most part. | 16:39 |
| *** gman-tx has joined #openstack-powervm | 16:40 | |
| fried_rice | catmando: gman-tx; gman-tx, likewise. | 16:41 |
| fried_rice | He might be able to help you with whatever your question is. | 16:41 |
| fried_rice | And with that, I'm out for a couple hours. | 16:41 |
| *** fried_rice is now known as fried_rolls | 16:41 | |
| catmando | Hey gman-tx | 16:43 |
| gman-tx | Hey catmando | 16:43 |
| catmando | Quick question. Do you know much about SVCs | 16:43 |
| gman-tx | from a cinder sandpoint yes ... but cinder doesn't fully take advantage of everything the SVC can do | 16:44 |
| catmando | We've had a total cluster failure, all nodes down with error 564. It may not be something you've ever seen before | 16:45 |
| gman-tx | whats the question? | 16:45 |
| catmando | Just a shot in the dark | 16:45 |
| catmando | We did open a ticket, but I am curious if anyone has seen this before | 16:45 |
| gman-tx | yuck a hardware failure | 16:45 |
| gman-tx | nope haven't see that one before | 16:47 |
| gman-tx | Too many machine code crashes have occurred? | 16:48 |
| gman-tx | what your firmware level? | 16:50 |
| *** esberglu has quit IRC | 16:52 | |
| manous | hi all | 17:03 |
| manous | is it possible to install openstack powervm on ibm x86 server ? | 17:03 |
| *** k0da has joined #openstack-powervm | 17:41 | |
| *** apearson has quit IRC | 17:52 | |
| *** esberglu has joined #openstack-powervm | 17:56 | |
| *** manous has quit IRC | 17:57 | |
| *** apearson has joined #openstack-powervm | 18:38 | |
| prashkre | fried_rice: Hi. Sorry, didn't understood your comment at https://review.openstack.org/#/c/557800/23/nova_powervm/virt/powervm/volume/iscsi.py | 19:03 |
| prashkre | It should be legit to pass [] (which will effectively make this method just return whatever's already in the _ISCSI_INITIATORS cache without the expensive get_active_vioses lookup). | 19:03 |
| prashkre | fried_rice: you mean to say "return _ISCSI_INITIATORS if vios_ids is None" ? | 19:07 |
| *** apearson has quit IRC | 19:25 | |
| *** fried_rolls is now known as fried_rice | 19:26 | |
| fried_rice | prashkre: No, I mean if you pass [] for vios_ids, we should not go try and get active VIOSes again. | 19:29 |
| fried_rice | prashkre: Though I still think this is a little weird, and maybe we should rethink it. | 19:30 |
| fried_rice | prashkre: Because we need to make sure we find the same initiator every time from get_volume_connector. | 19:32 |
| *** apearson has joined #openstack-powervm | 19:37 | |
| *** gman-tx has quit IRC | 19:41 | |
| fried_rice | prashkre: I think I have a solution. I left comments in the relevant places. | 19:48 |
| prashkre | fried_rice: sure. | 19:53 |
| *** gman-tx has joined #openstack-powervm | 19:57 | |
| prashkre | fried_rice: Sorry, I am new to storage, please don't laugh at me if I raise poor questions. I remember initiator IQN on VIOS is used to connect to initialize connection. | 20:07 |
| fried_rice | prashkre: I promise you, I know less about storage and ISCSI than you do. | 20:07 |
| prashkre | fried_rice: Since we are caching these initiators, what if the VIOS1 goes down whose IQN we are always picking. | 20:08 |
| fried_rice | I'm leaning heavily on folks like gman-tx to indicate that this patch is doing the right thing. I can just look at it from the perspective of code correctness and style, really. | 20:08 |
| fried_rice | prashkre: I agree that's an issue, and I think that's actually just tough luck. Because if we used VIOS1 to connect, and then VIOS1 goes down, I don't think it does us any good to try to disconnect via VIOS2. | 20:09 |
| fried_rice | gman-tx: please confirm that that's what you meant by this comment: https://review.openstack.org/#/c/557800/23/nova_powervm/virt/powervm/driver.py@1109 | 20:09 |
| fried_rice | Perhaps gman-tx is suggesting we actually use the instance ID in some way to figure out which VIOS we used for attach, so that we can reliably find the same one for detach. | 20:11 |
| fried_rice | If we set it up as I suggested, we can be sure we'll always use the same for both. | 20:11 |
| fried_rice | But it does leave a hole as you suggest | 20:11 |
| prashkre | ok. | 20:11 |
| fried_rice | which is that, if we've been happily using VIOS1 for everything, and VIOS1 goes down, it would be nice if the next *attach* could use VIOS2 instead. | 20:12 |
| gman-tx | you can't disconnect from vios2 if you connected to vios1 ... | 20:12 |
| fried_rice | even if attempting a *detach* from same would fail. | 20:12 |
| fried_rice | I'm not sure how we resolve this without a) refreshing the list of active VIOSes every time, b) using that state data to make sure we're retrieving a live initiator, and c) somehow saving which VIOS we used for which instance so that we can tell whether we *must* return the same one (even if it's down?) or if we can switch to a different one. | 20:13 |
| gman-tx | somewhere we would have to remember which vios we connected to | 20:14 |
| fried_rice | gman-tx: Is the "failure" any worse if we try to detach from VIOS2 than if we try to disconnect from VIOS1 when it's down? | 20:14 |
| fried_rice | I guess that would be confusing in the logs, no matter how we tried to paper over it. | 20:14 |
| fried_rice | I sure wouldn't love the idea of trying to persist instance IDs. There be tygers. | 20:15 |
| fried_rice | Using a hash only helps us if we can be assured we have the full list when we start. | 20:16 |
| fried_rice | I gotta run, guys. Should be back... later. | 20:16 |
| gman-tx | hash is a bit better but doesn't always get us the org value | 20:16 |
| prashkre | gman-tx: do you mean, remember that instance_1 is connected through vios1, instance_2 through vios2 like that ? | 20:17 |
| gman-tx | if you disconnect from vios 1 while it is down you leave the hdisk mapped to the VM ... a stale mapping | 20:17 |
| gman-tx | yea | 20:18 |
| prashkre | gman-tx: thanks for clarification. signing off for today. will catch you later. | 20:20 |
| gman-tx | yea need to step away my self | 20:21 |
| *** esberglu has quit IRC | 21:15 | |
| *** tjakobs has quit IRC | 21:30 | |
| *** esberglu has joined #openstack-powervm | 22:20 | |
| *** esberglu has quit IRC | 22:21 | |
| *** apearson has quit IRC | 22:26 | |
| *** prashkre has quit IRC | 22:32 | |
| *** k0da has quit IRC | 23:05 | |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!