| *** k0da has quit IRC | 00:07 | |
| *** kylek3h has joined #openstack-powervm | 00:29 | |
| *** thorst_ has joined #openstack-powervm | 00:38 | |
| *** thorst_ has quit IRC | 00:43 | |
| *** thorst_ has joined #openstack-powervm | 00:59 | |
| *** thorst_ has quit IRC | 01:03 | |
| *** thorst_ has joined #openstack-powervm | 01:04 | |
| *** thorst_ has quit IRC | 01:04 | |
| *** seroyer has joined #openstack-powervm | 01:29 | |
| *** svenkat has joined #openstack-powervm | 01:40 | |
| *** svenkat has quit IRC | 01:48 | |
| *** seroyer has quit IRC | 01:59 | |
| *** seroyer has joined #openstack-powervm | 02:05 | |
| *** thorst_ has joined #openstack-powervm | 02:05 | |
| *** thorst_ has quit IRC | 02:08 | |
| *** smatzek has joined #openstack-powervm | 02:30 | |
| *** smatzek has quit IRC | 02:49 | |
| *** thorst_ has joined #openstack-powervm | 03:35 | |
| *** thorst_ has quit IRC | 03:37 | |
| *** seroyer has joined #openstack-powervm | 03:47 | |
| *** seroyer has quit IRC | 03:48 | |
| *** tjakobs has joined #openstack-powervm | 04:39 | |
| *** tjakobs has quit IRC | 04:44 | |
| *** tlian has quit IRC | 05:10 | |
| *** thorst_ has joined #openstack-powervm | 05:37 | |
| *** thorst_ has quit IRC | 05:42 | |
| *** jwcroppe has quit IRC | 06:23 | |
| *** kairo has joined #openstack-powervm | 06:45 | |
| *** thorst_ has joined #openstack-powervm | 07:39 | |
| *** AlexeyAbashkin has joined #openstack-powervm | 07:39 | |
| *** kairo has quit IRC | 07:40 | |
| *** kairo has joined #openstack-powervm | 07:42 | |
| *** thorst_ has quit IRC | 07:43 | |
| *** kairo_ has joined #openstack-powervm | 07:54 | |
| *** k0da has joined #openstack-powervm | 07:57 | |
| *** kairo_ has quit IRC | 07:58 | |
| *** Sujeet_ has joined #openstack-powervm | 09:04 | |
| *** kairo has quit IRC | 09:05 | |
| *** kairo has joined #openstack-powervm | 09:05 | |
| *** openstackgerrit has quit IRC | 10:18 | |
| *** smatzek has joined #openstack-powervm | 11:26 | |
| *** Sujeet_ has quit IRC | 11:28 | |
| *** thorst_ has joined #openstack-powervm | 11:40 | |
| *** thorst_ has quit IRC | 11:45 | |
| *** seroyer has joined #openstack-powervm | 12:17 | |
| *** thorst_ has joined #openstack-powervm | 12:41 | |
| *** k0da has quit IRC | 12:48 | |
| *** seroyer has quit IRC | 12:59 | |
| *** seroyer has joined #openstack-powervm | 13:00 | |
| *** k0da has joined #openstack-powervm | 13:04 | |
| *** thorst_ has quit IRC | 13:05 | |
| *** thorst_ has joined #openstack-powervm | 13:09 | |
| *** k0da has quit IRC | 13:11 | |
| *** svenkat has joined #openstack-powervm | 13:16 | |
| *** AlexeyAbashkin has quit IRC | 13:37 | |
| *** AlexeyAbashkin has joined #openstack-powervm | 13:40 | |
| *** edmondsw has joined #openstack-powervm | 13:45 | |
| *** nbante has joined #openstack-powervm | 13:55 | |
| *** mdrabe has joined #openstack-powervm | 14:01 | |
| *** BorD__ has joined #openstack-powervm | 14:02 | |
| *** smatzek has quit IRC | 14:30 | |
| *** tlian has joined #openstack-powervm | 14:39 | |
| thorst_ | efried: sorry about all those RTC e-mails :-) | 14:44 |
|---|---|---|
| *** seroyer has quit IRC | 14:47 | |
| efried | thorst_: swhat the Delete key is for. | 14:49 |
| thorst_ | efried: good - that was the intention | 14:50 |
| *** apearson has joined #openstack-powervm | 14:53 | |
| *** seroyer has joined #openstack-powervm | 15:02 | |
| *** tblakes has joined #openstack-powervm | 15:11 | |
| *** kairo has quit IRC | 15:15 | |
| *** jwcroppe has joined #openstack-powervm | 15:17 | |
| *** smatzek has joined #openstack-powervm | 15:20 | |
| *** apearson has quit IRC | 15:37 | |
| *** nbante has quit IRC | 15:38 | |
| *** tjakobs has joined #openstack-powervm | 15:48 | |
| *** mdrabe has quit IRC | 15:50 | |
| *** tjakobs has quit IRC | 15:52 | |
| *** tjakobs has joined #openstack-powervm | 15:55 | |
| *** mdrabe has joined #openstack-powervm | 16:00 | |
| efried | adreznec: any progress on having 1.0.0.4 in pypi? | 16:07 |
| adreznec | Yeah, I'm fixing the job right now efried | 16:07 |
| adreznec | Had to get Julio to fix the build environment first | 16:07 |
| efried | adreznec: noyce. Was about to update the spawn change set, wanted to know whether I could bump the level in requirements. | 16:07 |
| adreznec | Once this fix merges in the future anyone will be able to push a new version by just going to the neo jenkins and triggering the job with the appropriate git tag | 16:08 |
| efried | lemme know when ready. | 16:08 |
| adreznec | FYI thorst_ ^ | 16:08 |
| tjakobs | efried: it sounds like you want a config option, to specify the cache volume group? Then pass the baseImage name & VG to rest instead of just name? | 16:22 |
| efried | tjakobs: I'd at least like to discuss the implications of the current design. thorst_, you around? | 16:22 |
| efried | tjakobs: while we wait for thorst_ - Was it your intention that the cache LV would live in the same VG as the VM's disk? | 16:34 |
| tjakobs | efried: yeah, I think that was his intention too (not sure if he is open to the conf idea or not). (see https://review.openstack.org/#/c/371946/) | 16:36 |
| efried | ohh, it's all coming together now. I guess I should read that blueprint. | 16:37 |
| efried | tjakobs: is there a nova change set for the nova/virt/imagecache.py work? | 16:39 |
| thorst_ | I'd prefer a single volume group...I don't see too much benefit in having two other than 'we could' | 16:39 |
| thorst_ | in a call, so slow to read/type | 16:39 |
| tjakobs | efried: just nova-powervm work that implements the nova imagecache https://review.openstack.org/#/c/408758 | 16:40 |
| efried | thorst_: okeydokey. I want to make sure we don't paint ourselves into a corner in case we want to make the cache VG separate at some point in the future, but I think the current implementation could easily be expanded to accomodate that if desired. | 16:42 |
| thorst_ | we can expand later if needed | 16:42 |
| thorst_ | can never pull back, can always expand | 16:42 |
| efried | I share your concerns about the naming convention. | 16:42 |
| thorst_ | well, so I was looking at nova-powervm earlier | 16:43 |
| thorst_ | and it looks like the way it was written adds IMAGE_ or something to the front? | 16:43 |
| thorst_ | BOOT_ | 16:44 |
| thorst_ | so, it kinda looks like there is a precedent for this... | 16:44 |
| thorst_ | I wanted to loop back and look more | 16:44 |
| *** AlexeyAbashkin has quit IRC | 16:49 | |
| efried | thorst_: got it. It looks like we're actually using single-character prefixes for VDisks because of the VDisk name limit (15c) | 16:49 |
| efried | So I think tjakobs has the right idea. | 16:50 |
| thorst_ | efried: it still concerns me. I kind of wonder if going to UUID's would be better though? | 16:50 |
| thorst_ | because UUIDs won't really overlap. | 16:50 |
| efried | We would be limited to the first 15c of the UUID. | 16:50 |
| efried | which is probably fine. | 16:50 |
| thorst_ | they *could*...but the probability would be much less than before | 16:50 |
| thorst_ | OK - that'd be OK | 16:50 |
| tjakobs | would we still want some identifier up front? or just the first 15 of uuid | 16:51 |
| adreznec | Yeah, the issue there was totally with the VIOS name limit | 16:51 |
| efried | I think consistency with the existing model is best. | 16:51 |
| thorst_ | so a g_<UUID> is fine with me | 16:51 |
| thorst_ | though I kinda want it as a constant rather than just inlined everywhere | 16:51 |
| efried | In fact, I would say we shouldn't even have a new prefix. | 16:51 |
| efried | We should just use the existing 'image' one. | 16:51 |
| tjakobs | will switch to i_, using constants | 16:52 |
| efried | Yeah, I was going to say that. We have methods in the disk driver base to generate the name - I assume you're using that on the creation end. | 16:52 |
| efried | Not sure if we have equivalent getter-type methods. But if not, we should make 'em. | 16:53 |
| efried | At the very least, use '%s_' % disk_type[0] rather than hardcoding the prefix. | 16:53 |
| efried | See nova_powervm.virt.powervm.disk.driver.DiskAdapter#_get_disk_name | 16:53 |
| tjakobs | should i make a new method that only uses uuid? | 16:54 |
| efried | Well, I need to talk through this a little more, if only for my understanding. | 16:54 |
| efried | A spawn request comes in with a glance image descriptor. You need to transform that somehow into a cache image name. Then you check whether that cache image exists in the localdisk VG. If not, you create it. Then you feed that cache image name to the VDisk.bld as its 'base'. So far correct? | 16:57 |
| tjakobs | correct | 16:57 |
| thorst_ | tjakobs: that image can't be attached to another VM, right? | 16:59 |
| thorst_ | it has to be the pure, checksum equivalent glance image. | 16:59 |
| efried | thorst_: that has to be true, yes. And matches the paradigm we have for IMAGE type disks elsewhere. | 16:59 |
| efried | ...And when you need to clean up, you get all the instances' boot disks, grab their 'base' property, and subtract that set from the set of all cache images. Anything that's left can be whacked. | 17:00 |
| tjakobs | yup | 17:00 |
| efried | Okay, so your method that generates the set of all cache image names - using the 'i_' prefix to find 'em is copacetic. I do want you to use the const to generate that prefix, though, and I see a potential circular import problem there. | 17:02 |
| efried | mebbe not, help me think it through. | 17:03 |
| efried | The consts live in driver. imagecache would have to import driver. localdisk imports imagecache and driver. Mm, that might be okay, right? Or we may want to consider moving those consts (disk/const.py?) | 17:05 |
| efried | Anyway, I think we're on the right track here. tjakobs - want me to dig into the review now, or wait for your next patch set? | 17:06 |
| tjakobs | could be move the consts and the _get_disk_name and _get_image_name into the const.py? | 17:07 |
| thorst_ | don't know that I want methods in const.py | 17:07 |
| efried | If it's gonna have methods, it should be called something other than const. Maybe base.py or util.py | 17:07 |
| tjakobs | so consts, and usage of the consts into a disk/util.py? | 17:08 |
| efried | I'm on board with that idea - but only if needed to resolve circular imports. Otherwise, I'm okay leaving the base stuff in driver.py. Unless thorst_ thinks otherwise? | 17:08 |
| thorst_ | I'm ok leaving stuff in driver.py, but find with that refactor if it helps circular dependencies (like efried said) | 17:09 |
| efried | coo | 17:09 |
| tjakobs | sounds good. efried go ahead and review it if you have time. the only things changing in the next patchset will be this naming stuff | 17:10 |
| efried | k, then many of my comments will be reiterating what we've discussed above. | 17:11 |
| efried | thorst_: is there a corresponding blueprint template in launchpad? | 17:17 |
| efried | I'm not really up to speed on how blueprint cross-references are supposed to work. | 17:19 |
| efried | aha, found it. | 17:19 |
| *** kairo has joined #openstack-powervm | 17:34 | |
| *** apearson has joined #openstack-powervm | 17:35 | |
| *** kairo has quit IRC | 17:41 | |
| *** jwcroppe has quit IRC | 17:56 | |
| *** jwcroppe has joined #openstack-powervm | 17:59 | |
| *** nil_ has joined #openstack-powervm | 18:09 | |
| *** nil_ has quit IRC | 18:16 | |
| *** nil_ has joined #openstack-powervm | 18:17 | |
| *** apearson has quit IRC | 18:35 | |
| *** apearson has joined #openstack-powervm | 18:36 | |
| *** newbie has joined #openstack-powervm | 18:40 | |
| *** newbie has quit IRC | 18:40 | |
| *** nbante has joined #openstack-powervm | 18:45 | |
| *** nil_ has quit IRC | 18:46 | |
| *** jwcroppe has quit IRC | 18:48 | |
| *** nbante has quit IRC | 18:53 | |
| *** jwcroppe has joined #openstack-powervm | 18:53 | |
| *** nbante has joined #openstack-powervm | 18:53 | |
| *** openstack has joined #openstack-powervm | 19:00 | |
| *** apearson has quit IRC | 19:03 | |
| *** apearson has joined #openstack-powervm | 19:24 | |
| *** apearson has quit IRC | 19:40 | |
| *** apearson has joined #openstack-powervm | 19:56 | |
| *** openstackstatus has quit IRC | 19:57 | |
| *** openstack has joined #openstack-powervm | 19:59 | |
| *** k0da has joined #openstack-powervm | 20:09 | |
| *** openstackgerrit has joined #openstack-powervm | 20:15 | |
| openstackgerrit | Taylor Jakobson proposed openstack/nova-powervm: WIP First pass at imagecache https://review.openstack.org/408758 | 20:15 |
| *** openstackstatus has quit IRC | 20:27 | |
| *** openstack has joined #openstack-powervm | 20:29 | |
| *** smatzek has quit IRC | 21:54 | |
| *** svenkat has quit IRC | 21:59 | |
| *** thorst_ has quit IRC | 22:02 | |
| *** thorst has joined #openstack-powervm | 22:03 | |
| *** thorst has quit IRC | 22:07 | |
| *** seroyer has quit IRC | 22:24 | |
| *** thorst has joined #openstack-powervm | 22:28 | |
| *** thorst has quit IRC | 22:31 | |
| *** smatzek has joined #openstack-powervm | 22:34 | |
| *** smatzek has quit IRC | 22:35 | |
| *** smatzek has joined #openstack-powervm | 22:35 | |
| *** smatzek has quit IRC | 23:01 | |
| *** tblakes has quit IRC | 23:05 | |
| *** tjakobs has quit IRC | 23:05 | |
| *** mdrabe has quit IRC | 23:10 | |
| *** edmondsw has quit IRC | 23:29 | |
| *** thorst has joined #openstack-powervm | 23:32 | |
| *** edmondsw has joined #openstack-powervm | 23:36 | |
| *** thorst has quit IRC | 23:37 | |
| *** edmondsw_ has joined #openstack-powervm | 23:38 | |
| *** edmonds__ has joined #openstack-powervm | 23:40 | |
| *** edmondsw has quit IRC | 23:41 | |
| *** edmondsw_ has quit IRC | 23:43 | |
| *** edmonds__ has quit IRC | 23:47 | |
| *** jwcroppe has quit IRC | 23:47 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!