| *** emagana has joined #openstack-cinder | 00:30 | |
| *** emagana has quit IRC | 00:36 | |
| *** dimsum__ has quit IRC | 00:37 | |
| *** dims has joined #openstack-cinder | 00:44 | |
| *** jwcroppe has quit IRC | 00:45 | |
| *** vivek_nandavanam has joined #openstack-cinder | 00:48 | |
| *** lixiaoy1 has joined #openstack-cinder | 01:11 | |
| *** markvoelker has quit IRC | 01:18 | |
| *** heyun has joined #openstack-cinder | 01:24 | |
| *** emagana has joined #openstack-cinder | 01:25 | |
| *** tobe has joined #openstack-cinder | 01:30 | |
| *** emagana has quit IRC | 01:30 | |
| *** mriedem has quit IRC | 01:30 | |
| *** dylan has joined #openstack-cinder | 01:34 | |
| *** dims has quit IRC | 01:37 | |
| *** dims has joined #openstack-cinder | 01:38 | |
| *** dims has quit IRC | 01:43 | |
| *** Lee1092 has joined #openstack-cinder | 01:45 | |
| openstackgerrit | wanghao proposed openstack/cinder: Support for force-delete backups https://review.openstack.org/166127 | 01:54 |
|---|---|---|
| *** garthb has joined #openstack-cinder | 01:58 | |
| openstackgerrit | wanghao proposed openstack/cinder: Support for force-delete backups https://review.openstack.org/166127 | 01:59 |
| *** xiaohui has quit IRC | 02:02 | |
| asselin_ | nikeshm, great job! I see fc is working even though this message is in the console.log. Probably need to update the fc script to be more robust.... | 02:06 |
| asselin_ | 12:55:45 2015-05-31 19:55:45.363 | Failed to install FC Drivers. Aborting. | 02:06 |
| asselin_ | http://115.119.152.41/65/177665/5/check/dsvm-tempest-full-fc/4925660/console.html | 02:06 |
| *** nikeshm has quit IRC | 02:11 | |
| *** emagana has joined #openstack-cinder | 02:19 | |
| *** akshai has joined #openstack-cinder | 02:22 | |
| *** emagana has quit IRC | 02:23 | |
| *** akshai has quit IRC | 02:30 | |
| *** asselin_ has quit IRC | 02:33 | |
| *** markvoelker has joined #openstack-cinder | 02:34 | |
| *** markvoelker has quit IRC | 02:39 | |
| openstackgerrit | Rick Chen proposed openstack/cinder: Prophetstor driver needs to return snapshot objects for create_cgsnapshot and delete_cgsnapshot. https://review.openstack.org/185495 | 02:39 |
| openstackgerrit | Rick Chen proposed openstack/cinder: Complete switch to snapshot objects https://review.openstack.org/163910 | 02:39 |
| openstackgerrit | Rick Chen proposed openstack/cinder: Switch get_all_snapshots to use objects https://review.openstack.org/161077 | 02:39 |
| *** Longgeek has joined #openstack-cinder | 02:42 | |
| *** Longgeek has quit IRC | 02:45 | |
| openstackgerrit | XinXiaohui proposed openstack/cinder-specs: capacity-headroom https://review.openstack.org/170380 | 02:47 |
| *** Longgeek has joined #openstack-cinder | 02:53 | |
| *** nikeshm has joined #openstack-cinder | 03:11 | |
| nikeshm | asselin: thanks | 03:11 |
| nikeshm | asselin: how could i improve the speed of test, | 03:12 |
| nikeshm | its taking 1 hr 40 minutes | 03:12 |
| nikeshm | i have good network and hardware | 03:12 |
| nikeshm | using kvm in os provider | 03:13 |
| *** emagana has joined #openstack-cinder | 03:13 | |
| nikeshm | but devtstack itself taking half an hr in nodepool vms | 03:13 |
| *** Yogi1 has joined #openstack-cinder | 03:14 | |
| *** emagana has quit IRC | 03:17 | |
| *** jwcroppe has joined #openstack-cinder | 03:25 | |
| *** asselin_ has joined #openstack-cinder | 03:27 | |
| *** jwcroppe has quit IRC | 03:42 | |
| *** haomaiw__ has joined #openstack-cinder | 03:47 | |
| *** Yogi1 has quit IRC | 03:49 | |
| *** haomaiwang has quit IRC | 03:49 | |
| *** sgotliv has joined #openstack-cinder | 03:51 | |
| *** haomaiw__ has quit IRC | 03:59 | |
| *** haomaiwang has joined #openstack-cinder | 04:02 | |
| *** yamada-h has joined #openstack-cinder | 04:03 | |
| *** emagana has joined #openstack-cinder | 04:07 | |
| *** emagana has quit IRC | 04:11 | |
| *** garthb has quit IRC | 04:13 | |
| *** markvoelker has joined #openstack-cinder | 04:23 | |
| *** markvoelker has quit IRC | 04:28 | |
| *** sks has joined #openstack-cinder | 04:35 | |
| *** vivek_nandavanam has quit IRC | 04:43 | |
| *** rushiagr_away is now known as rushiagr | 04:52 | |
| *** vilobhmm has quit IRC | 04:59 | |
| *** tobe has quit IRC | 05:01 | |
| *** emagana has joined #openstack-cinder | 05:01 | |
| *** emagana has quit IRC | 05:06 | |
| *** tobe has joined #openstack-cinder | 05:07 | |
| *** vilobhmm has joined #openstack-cinder | 05:07 | |
| *** zhenguo has joined #openstack-cinder | 05:39 | |
| *** tobe has quit IRC | 05:44 | |
| *** Miouge has joined #openstack-cinder | 05:46 | |
| *** emagana has joined #openstack-cinder | 05:55 | |
| *** ankit_ag has joined #openstack-cinder | 05:57 | |
| openstackgerrit | Vincent Hou proposed openstack/cinder-specs: Volume migration improvement for L https://review.openstack.org/186327 | 06:00 |
| *** emagana has quit IRC | 06:00 | |
| openstackgerrit | Vincent Hou proposed openstack/cinder-specs: Volume migration improvement for L https://review.openstack.org/186327 | 06:02 |
| *** tobe has joined #openstack-cinder | 06:03 | |
| *** yamada-h has quit IRC | 06:04 | |
| *** yamada-h has joined #openstack-cinder | 06:06 | |
| *** yamada-h has quit IRC | 06:10 | |
| *** markvoelker has joined #openstack-cinder | 06:12 | |
| *** nlevinki has joined #openstack-cinder | 06:13 | |
| *** markvoelker has quit IRC | 06:16 | |
| *** nkrinner has joined #openstack-cinder | 06:17 | |
| *** sgotliv has quit IRC | 06:17 | |
| *** pradipta has joined #openstack-cinder | 06:18 | |
| *** Miouge has quit IRC | 06:20 | |
| *** anshul has joined #openstack-cinder | 06:23 | |
| *** Miouge has joined #openstack-cinder | 06:27 | |
| *** Miouge has quit IRC | 06:32 | |
| *** agarciam has joined #openstack-cinder | 06:33 | |
| *** Longgeek has quit IRC | 06:34 | |
| *** Longgeek has joined #openstack-cinder | 06:34 | |
| *** deepakcs has joined #openstack-cinder | 06:38 | |
| *** kaisers1 has joined #openstack-cinder | 06:43 | |
| *** rhefner has quit IRC | 06:43 | |
| openstackgerrit | wanghao proposed openstack/cinder: Support volume backup of qcow2 format volume https://review.openstack.org/182276 | 06:44 |
| *** markus_z has joined #openstack-cinder | 06:47 | |
| *** emagana has joined #openstack-cinder | 06:50 | |
| *** daneyon has joined #openstack-cinder | 06:50 | |
| *** Miouge has joined #openstack-cinder | 06:50 | |
| *** emagana has quit IRC | 06:54 | |
| *** daneyon has quit IRC | 07:00 | |
| *** daneyon has joined #openstack-cinder | 07:01 | |
| *** tshefi has joined #openstack-cinder | 07:03 | |
| *** daneyon has quit IRC | 07:06 | |
| *** afazekas has joined #openstack-cinder | 07:12 | |
| *** yamada-h has joined #openstack-cinder | 07:16 | |
| *** vilobhmm has quit IRC | 07:24 | |
| *** haomaiw__ has joined #openstack-cinder | 07:24 | |
| *** haomaiwang has quit IRC | 07:27 | |
| *** Zhongjun has joined #openstack-cinder | 07:28 | |
| *** dulek has joined #openstack-cinder | 07:29 | |
| *** leopoldj has joined #openstack-cinder | 07:32 | |
| openstackgerrit | Vincent Hou proposed openstack/cinder-specs: Volume migration improvement for L https://review.openstack.org/186327 | 07:33 |
| *** tobe has quit IRC | 07:37 | |
| openstackgerrit | Vincent Hou proposed openstack/cinder-specs: Volume migration improvement for L https://review.openstack.org/186327 | 07:37 |
| *** jistr has joined #openstack-cinder | 07:39 | |
| *** tobe has joined #openstack-cinder | 07:39 | |
| *** lpetrut has joined #openstack-cinder | 07:43 | |
| *** dylan has quit IRC | 07:43 | |
| *** emagana has joined #openstack-cinder | 07:44 | |
| *** chlong has quit IRC | 07:45 | |
| *** nihilifer has joined #openstack-cinder | 07:46 | |
| *** emagana has quit IRC | 07:49 | |
| *** yamada-h_ has joined #openstack-cinder | 07:55 | |
| *** yamada-h has quit IRC | 07:55 | |
| *** yamada-h_ has quit IRC | 07:57 | |
| *** yamada-h has joined #openstack-cinder | 07:57 | |
| *** markvoelker has joined #openstack-cinder | 08:01 | |
| *** jordanP has joined #openstack-cinder | 08:01 | |
| *** markvoelker has quit IRC | 08:06 | |
| *** sgotliv has joined #openstack-cinder | 08:12 | |
| *** tobe has quit IRC | 08:22 | |
| *** sgotliv has quit IRC | 08:22 | |
| *** sgotliv has joined #openstack-cinder | 08:22 | |
| *** tobe has joined #openstack-cinder | 08:23 | |
| *** ajayaa has joined #openstack-cinder | 08:36 | |
| ajayaa | Hi guys. Does Cinder work with Keystone v3 api in Icehouse? | 08:37 |
| ajayaa | In my test environment it is using v2.0 api of Keystone and when it tries to verify a v3 token with v2.0 api, Keystone gives a 401. | 08:38 |
| *** emagana has joined #openstack-cinder | 08:38 | |
| *** sgotliv has quit IRC | 08:41 | |
| *** emagana has quit IRC | 08:43 | |
| *** sgotliv has joined #openstack-cinder | 08:45 | |
| *** yogeshwars1 has joined #openstack-cinder | 08:46 | |
| *** Longgeek has quit IRC | 08:50 | |
| *** afazekas_ has joined #openstack-cinder | 08:55 | |
| *** Longgeek has joined #openstack-cinder | 08:56 | |
| *** afazekas has quit IRC | 08:59 | |
| *** e0ne has joined #openstack-cinder | 09:02 | |
| *** afazekas_ has quit IRC | 09:17 | |
| *** IanGovett has joined #openstack-cinder | 09:23 | |
| *** mdbooth has quit IRC | 09:25 | |
| *** emagana has joined #openstack-cinder | 09:32 | |
| openstackgerrit | Vipin Balachandran proposed openstack/cinder: Fix capacity weigher help messages https://review.openstack.org/187147 | 09:33 |
| openstackgerrit | Vipin Balachandran proposed openstack/cinder: Fix doc in deprecated simple scheduler https://review.openstack.org/187148 | 09:36 |
| *** emagana has quit IRC | 09:37 | |
| *** aix has joined #openstack-cinder | 09:43 | |
| *** afazekas has joined #openstack-cinder | 09:45 | |
| *** e0ne is now known as e0ne_ | 09:48 | |
| *** markvoelker has joined #openstack-cinder | 09:49 | |
| *** e0ne_ is now known as e0ne | 09:50 | |
| *** markvoelker has quit IRC | 09:54 | |
| openstackgerrit | Vipin Balachandran proposed openstack/cinder: Add missing unit test for goodness weigher https://review.openstack.org/187157 | 09:54 |
| openstackgerrit | wanghao proposed openstack/cinder: Non-admin user to query volume filter by az https://review.openstack.org/156034 | 09:55 |
| *** lixiaoy1 has quit IRC | 09:55 | |
| *** dims has joined #openstack-cinder | 09:55 | |
| *** mdbooth has joined #openstack-cinder | 10:02 | |
| openstackgerrit | Ph. Marek proposed openstack/cinder: Re-add DRBD driver. https://review.openstack.org/178573 | 10:09 |
| *** heyun has quit IRC | 10:13 | |
| openstackgerrit | Vipin Balachandran proposed openstack/cinder: VMware: Create volume backing in specific clusters https://review.openstack.org/186664 | 10:13 |
| *** zz_naotok has quit IRC | 10:15 | |
| *** openstackstatus has quit IRC | 10:20 | |
| *** openstack has quit IRC | 10:20 | |
| *** openstack has joined #openstack-cinder | 10:27 | |
| *** emagana has quit IRC | 10:31 | |
| *** Miouge has quit IRC | 10:31 | |
| openstackgerrit | Pranali Deore proposed openstack/cinder: Made error message consistent across all apis https://review.openstack.org/186688 | 10:38 |
| *** nlevinki_ has joined #openstack-cinder | 10:42 | |
| *** danieru has joined #openstack-cinder | 10:42 | |
| *** nlevinki has quit IRC | 10:43 | |
| *** danieru_ has quit IRC | 10:43 | |
| *** mitz has quit IRC | 10:43 | |
| *** mitz has joined #openstack-cinder | 10:43 | |
| *** bswartz has quit IRC | 10:59 | |
| *** Miouge has joined #openstack-cinder | 11:02 | |
| *** emagana has joined #openstack-cinder | 11:20 | |
| *** e0ne has quit IRC | 11:21 | |
| *** emagana has quit IRC | 11:25 | |
| *** tobe has quit IRC | 11:25 | |
| *** rhefner has joined #openstack-cinder | 11:28 | |
| *** e0ne has joined #openstack-cinder | 11:30 | |
| *** aix has quit IRC | 11:34 | |
| *** markvoelker has joined #openstack-cinder | 11:35 | |
| *** markvoelker has quit IRC | 11:40 | |
| *** deepakcs has quit IRC | 11:42 | |
| dulek | geguileor: Hi! | 11:42 |
| dulek | geguileor: I'm in the process of refactoring api.create_volume. | 11:42 |
| geguileor | dulek Hi | 11:43 |
| dulek | geguileor: I've just noticed a lot of "TODO(harlowja): what happens if the status changes after this | 11:43 |
| dulek | initial volume status check occurs???" | 11:43 |
| dulek | This is when creating a volume from another volume. | 11:43 |
| dulek | Any idea if this is planned to be fixed somehow? | 11:44 |
| geguileor | dulek: You got me with my mind in another problem and I'm a little bit lost | 11:44 |
| geguileor | dulek: Can you point me to one of those cases you mention, please? | 11:45 |
| dulek | https://github.com/openstack/cinder/blob/master/cinder/volume/flows/api/create_volume.py#L123 | 11:45 |
| dulek | Simple situation - we're doing this check, then snapshot changes state. | 11:46 |
| dulek | This is an obvious race condition, but I wonder if your efforts to get atomic state changes cover also that place. | 11:47 |
| *** aix has joined #openstack-cinder | 11:48 | |
| geguileor | dulek: I don't think that would be covered by atomic state changes | 11:48 |
| geguileor | dulek: That would have to do with resource locking | 11:48 |
| geguileor | dulek: Like not allowing deletion of used resources | 11:48 |
| geguileor | dulek: In some places this is currently handled with synchronized, which will have to change | 11:49 |
| dulek | Okay, I get it. | 11:49 |
| dulek | Probably changing snapshot's status to some *ing value would also do the job. | 11:49 |
| geguileor | dulek: Since you are working now just to refactor the Taskflow part to see if there are benefits | 11:49 |
| geguileor | dulek: I would leave those comments and maybe add a note to Etherpad | 11:50 |
| dulek | Okay, I'll take a look if this kind of races are harmful and how it can be approached. | 11:50 |
| dulek | Thanks for clarifying! :) | 11:50 |
| ekarlso | how in devstack with cinder can I have it share a volume ? | 11:51 |
| geguileor | dulek: I think ING states are not a solution for this | 11:51 |
| geguileor | dulek: Because you could be creating multiple volumes from the same snapshot at the same time | 11:51 |
| geguileor | dulek: And then you would need to have an additional counter to check if you can ffree the snapshot | 11:52 |
| dulek | geguileor: Okay, you're right. | 11:52 |
| geguileor | dulek: I had the same initial idea :) | 11:52 |
| dulek | ekarlso: What do you mean by sharing? | 11:52 |
| ekarlso | dulek: use a existing vol group | 11:53 |
| dulek | ekarlso: Something like that: http://paste.openstack.org/show/252664/ | 11:56 |
| *** ronis has joined #openstack-cinder | 11:56 | |
| dulek | ekarlso: ? | 11:56 |
| dulek | ekarlso: In cinder.conf | 11:56 |
| *** rushiagr is now known as rushiagr_away | 11:57 | |
| ekarlso | dulek: I mean, in devstack how is there a way to say CINDER_VOLUME_NAME=fooo | 11:58 |
| geguileor | ekarlso: Are you trying to do multi backend? | 11:58 |
| ekarlso | geguileor: no, i just have a pre-existing volume group "rootvg" that's on a raid and I want cinder in devstack to use it | 11:59 |
| geguileor | Oh, ok | 11:59 |
| geguileor | ekarlso: Are you using local.conf or localrc file? | 11:59 |
| ekarlso | geguileor: local.conf | 12:00 |
| geguileor | ekarlso: Only 1 backend? | 12:00 |
| ekarlso | geguileor: yep | 12:00 |
| geguileor | Then it's basically what dulek said: Add this at the end of your local.conf http://paste.openstack.org/show/252665/ | 12:02 |
| geguileor | ekarlso: That should do the trick | 12:02 |
| ekarlso | ah | 12:03 |
| dulek | ekarlso, geguileor: Yes, I've just sent options from cinder.conf instead of local.conf | 12:03 |
| *** ociuhandu has joined #openstack-cinder | 12:03 | |
| geguileor | dulek: The line he was looking for was just [[post-config|$CINDER_CONF]] | 12:04 |
| *** ganso_ has joined #openstack-cinder | 12:04 | |
| geguileor | Adding that to local.conf will make DevStack use next sections for cinder configuration | 12:04 |
| *** Zhongjun has quit IRC | 12:09 | |
| *** akerr has joined #openstack-cinder | 12:14 | |
| *** emagana has joined #openstack-cinder | 12:15 | |
| *** belmoreira has joined #openstack-cinder | 12:15 | |
| *** Miouge has quit IRC | 12:17 | |
| *** emagana has quit IRC | 12:19 | |
| *** ociuhandu has quit IRC | 12:20 | |
| *** rmesta has joined #openstack-cinder | 12:31 | |
| *** markvoelker has joined #openstack-cinder | 12:34 | |
| *** sks has quit IRC | 12:38 | |
| *** IanGovett1 has joined #openstack-cinder | 12:39 | |
| *** IanGovett has quit IRC | 12:41 | |
| *** timcl has joined #openstack-cinder | 12:41 | |
| *** chlong has joined #openstack-cinder | 12:49 | |
| *** Zhongjun has joined #openstack-cinder | 12:59 | |
| *** sgotliv_ has joined #openstack-cinder | 12:59 | |
| *** sgotliv has quit IRC | 13:00 | |
| *** Miouge has joined #openstack-cinder | 13:02 | |
| *** yamada-h has quit IRC | 13:03 | |
| *** eharney has quit IRC | 13:03 | |
| *** ajayaa has quit IRC | 13:04 | |
| openstackgerrit | Pranali Deore proposed openstack/cinder: Validate name and description string https://review.openstack.org/182994 | 13:08 |
| *** primechuck has joined #openstack-cinder | 13:08 | |
| *** emagana has joined #openstack-cinder | 13:09 | |
| openstackgerrit | Pranali Deore proposed openstack/cinder: Return 400 error instead of 500 for invalid body https://review.openstack.org/186669 | 13:09 |
| *** dustins has joined #openstack-cinder | 13:13 | |
| *** emagana has quit IRC | 13:13 | |
| *** Yogi1 has joined #openstack-cinder | 13:14 | |
| *** jaypipes has joined #openstack-cinder | 13:16 | |
| *** ajayaa has joined #openstack-cinder | 13:20 | |
| *** xyang1 has joined #openstack-cinder | 13:20 | |
| *** merooney has joined #openstack-cinder | 13:22 | |
| *** annegentle has joined #openstack-cinder | 13:23 | |
| openstackgerrit | Vincent Hou proposed openstack/cinder: Implement the update_migrated_volume for the drivers https://review.openstack.org/180873 | 13:25 |
| smcginnis | Finally back to the land of the living. I've got some scrollback to read through! :) | 13:32 |
| *** superdan is now known as dansmith | 13:34 | |
| *** cbader has joined #openstack-cinder | 13:35 | |
| *** e0ne has quit IRC | 13:35 | |
| *** rushil has joined #openstack-cinder | 13:42 | |
| *** merooney has quit IRC | 13:43 | |
| *** dhellmann has left #openstack-cinder | 13:44 | |
| *** eharney has joined #openstack-cinder | 13:44 | |
| *** rushil has quit IRC | 13:47 | |
| tbarron | smcginnis: Welcome back! | 13:48 |
| openstackgerrit | Szymon Wróblewski proposed openstack/cinder: Tooz locks https://review.openstack.org/183537 | 13:49 |
| smcginnis | tbarron: Thanks! Feel like I've missed a lot. | 13:53 |
| *** rmesta has quit IRC | 13:54 | |
| tbarron | smcginnis: we didn't do too much damage while you were gone. | 13:54 |
| openstackgerrit | Vincent Hou proposed openstack/cinder-specs: Volume migration improvement for L https://review.openstack.org/186327 | 13:55 |
| *** jaypipes has quit IRC | 13:55 | |
| *** mriedem has joined #openstack-cinder | 13:55 | |
| *** merooney has joined #openstack-cinder | 13:56 | |
| *** ajayaa has quit IRC | 13:57 | |
| *** rushil has joined #openstack-cinder | 13:58 | |
| smcginnis | ;) | 14:00 |
| *** annegentle has quit IRC | 14:02 | |
| *** emagana has joined #openstack-cinder | 14:03 | |
| *** annegentle has joined #openstack-cinder | 14:03 | |
| *** thangp has joined #openstack-cinder | 14:07 | |
| *** ankit_ag has quit IRC | 14:08 | |
| *** emagana has quit IRC | 14:08 | |
| *** merooney has quit IRC | 14:15 | |
| *** merooney has joined #openstack-cinder | 14:16 | |
| openstackgerrit | Szymon Wróblewski proposed openstack/cinder: Tooz locks https://review.openstack.org/183537 | 14:20 |
| *** anuragpalsule has joined #openstack-cinder | 14:21 | |
| *** agarciam has quit IRC | 14:21 | |
| *** agarciam has joined #openstack-cinder | 14:21 | |
| *** rmesta has joined #openstack-cinder | 14:25 | |
| *** merooney has quit IRC | 14:26 | |
| *** krtaylor has quit IRC | 14:30 | |
| *** mtanino has joined #openstack-cinder | 14:31 | |
| *** merooney has joined #openstack-cinder | 14:31 | |
| *** annegentle has quit IRC | 14:32 | |
| *** breitz has joined #openstack-cinder | 14:32 | |
| *** crose has joined #openstack-cinder | 14:34 | |
| *** anuragpalsule has quit IRC | 14:35 | |
| *** anuragpalsule1 has joined #openstack-cinder | 14:35 | |
| openstackgerrit | John Griffith proposed openstack/cinder: Use SolidFire snapshots for Cinder snapshots https://review.openstack.org/186963 | 14:35 |
| *** krtaylor has joined #openstack-cinder | 14:42 | |
| *** nlevinki_ has quit IRC | 14:50 | |
| *** jaypipes has joined #openstack-cinder | 14:51 | |
| *** julim has joined #openstack-cinder | 14:52 | |
| *** e0ne has joined #openstack-cinder | 14:52 | |
| *** e0ne is now known as e0ne_ | 14:52 | |
| *** Yogi11 has joined #openstack-cinder | 14:55 | |
| *** e0ne_ is now known as e0ne | 14:56 | |
| *** afazekas has quit IRC | 14:57 | |
| openstackgerrit | Szymon Wróblewski proposed openstack/cinder: Tooz locks https://review.openstack.org/183537 | 14:58 |
| *** Yogi1 has quit IRC | 14:59 | |
| *** Guest83839 has joined #openstack-cinder | 15:00 | |
| flip214 | thingee: I'm not available on Wednesday, but I wanted to tell you that I think we've nearly arrived with the CI for DRBD. Unless there's something else to do that I don't know about yet... | 15:01 |
| *** dmz has quit IRC | 15:02 | |
| *** merooney has quit IRC | 15:05 | |
| *** jungleboyj has joined #openstack-cinder | 15:07 | |
| *** emagana has joined #openstack-cinder | 15:09 | |
| *** emagana has quit IRC | 15:09 | |
| *** emagana has joined #openstack-cinder | 15:09 | |
| *** Zhongjun has quit IRC | 15:09 | |
| *** emagana has quit IRC | 15:10 | |
| *** hemnabeer is now known as hemna | 15:11 | |
| *** lpetrut has quit IRC | 15:11 | |
| *** emagana has joined #openstack-cinder | 15:11 | |
| *** Yogi11 has quit IRC | 15:11 | |
| *** emagana has quit IRC | 15:12 | |
| openstackgerrit | Szymon Wróblewski proposed openstack/cinder: POC: Tooz locks demo https://review.openstack.org/185646 | 15:12 |
| *** merooney has joined #openstack-cinder | 15:12 | |
| *** emagana has joined #openstack-cinder | 15:13 | |
| *** emagana has quit IRC | 15:14 | |
| *** Yogi1 has joined #openstack-cinder | 15:18 | |
| *** david-ly_ is now known as david-lyle | 15:21 | |
| *** Yogi1 has quit IRC | 15:25 | |
| *** emagana has joined #openstack-cinder | 15:25 | |
| mtanino | jgriffith: I posted a spec for enhancement of volume copy during volume migration which we discussed last week. | 15:27 |
| *** dannywilson has joined #openstack-cinder | 15:27 | |
| mtanino | jgriffith: Could you give me a feedback for it? https://review.openstack.org/#/c/186209/ Thanks. | 15:28 |
| *** tshefi has quit IRC | 15:29 | |
| *** leopoldj has quit IRC | 15:30 | |
| *** akshai has joined #openstack-cinder | 15:30 | |
| *** jdurgin has joined #openstack-cinder | 15:31 | |
| smcginnis | flip214: That should be the last roadblock as far as I'm aware. https://wiki.openstack.org/wiki/Cinder/how-to-contribute-a-driver | 15:31 |
| *** anuragpalsule1 has quit IRC | 15:32 | |
| flip214 | smcginnis: I do hope so, too ;) | 15:33 |
| jungleboyj | smcginnis: Welcome back. | 15:34 |
| smcginnis | jungleboyj: Thanks! | 15:34 |
| flip214 | smcginnis: Now I only have to find out what to do to get my driver tempest run using that devstack plugin | 15:34 |
| smcginnis | jungleboyj: Now to unbury myself from the email avalanche. | 15:34 |
| Swanson | smcginnis: Congrats on fooling the core! | 15:35 |
| *** nihilifer has quit IRC | 15:35 | |
| smcginnis | Swanson: Hah! :-P | 15:35 |
| jungleboyj | smcginnis: Yeah, I don't think I am totally out of that from the Summit yet. | 15:35 |
| jungleboyj | smcginnis: How were the karate seminars? | 15:35 |
| smcginnis | jungleboyj: Very good. Exhausted today though, so catching up is going to be a lot of fun. | 15:36 |
| openstackgerrit | John Griffith proposed openstack/cinder: Return provider_id in SolidFire model update https://review.openstack.org/187233 | 15:36 |
| jungleboyj | :-) | 15:36 |
| *** daneyon has joined #openstack-cinder | 15:36 | |
| *** harlowja_at_home has joined #openstack-cinder | 15:38 | |
| *** Yogi1 has joined #openstack-cinder | 15:38 | |
| *** merooney has quit IRC | 15:39 | |
| *** dims has quit IRC | 15:41 | |
| *** dhellmann has joined #openstack-cinder | 15:42 | |
| *** dims_ has joined #openstack-cinder | 15:42 | |
| dhellmann | thingee: I would like to change the "drivers" for https://launchpad.net/os-brick to the cinder-drivers team (it's set to "Walter Boring" now). Do you have any objections? | 15:43 |
| *** patrickeast has joined #openstack-cinder | 15:43 | |
| anteaya | hemna: ^^ | 15:43 |
| hemna | doh | 15:43 |
| hemna | what did I miss ? | 15:43 |
| openstackgerrit | Sean Chen proposed openstack/cinder: Tintri Cinder Volume driver https://review.openstack.org/185148 | 15:43 |
| jungleboyj | Ruh Roh. | 15:43 |
| thingee | dhellmann: fine with me | 15:43 |
| dhellmann | thingee, hemna : not a big deal, I can fix it. | 15:44 |
| dhellmann | anteaya: thanks :-) | 15:44 |
| hemna | dhellmann, oh yah, it should be owned by cinder | 15:44 |
| anteaya | dhellmann: welcome | 15:44 |
| hemna | I couldn't fix it after I created it | 15:44 |
| hemna | thanks a bunch! | 15:44 |
| anteaya | hemna: not sure you missed anything, dhellmann is just getting things in line | 15:44 |
| dhellmann | hemna: I think after you change the administrator to a group you're not in, you lose privileges | 15:44 |
| hemna | I followed the instructions that I saw, and then realized I lost the ability to move it to cinder | 15:45 |
| anteaya | hemna: you can review this patch if you want: https://review.openstack.org/#/c/187212/ | 15:45 |
| hemna | I figured it would get fixed eventually | 15:45 |
| hemna | :P | 15:45 |
| hemna | anteaya, ok will do | 15:45 |
| dhellmann | hemna: eventually has come! :-) | 15:45 |
| anteaya | dhellmann: otherwise known as 'eventually' | 15:45 |
| hemna | heheh | 15:45 |
| hemna | :P | 15:45 |
| anteaya | hemna: thanks :) | 15:45 |
| *** jdurgin has quit IRC | 15:47 | |
| *** dhellmann has left #openstack-cinder | 15:48 | |
| *** e0ne is now known as e0ne_ | 15:50 | |
| jgriffith | mtanino: minor request and a couple observations/questions | 15:51 |
| *** leeantho has joined #openstack-cinder | 15:51 | |
| mtanino | jgriffith: sure | 15:51 |
| *** e0ne_ is now known as e0ne | 15:51 | |
| jgriffith | mtanino: let me know if the note about explicitly pointint out the use of capability makes sense | 15:51 |
| jgriffith | mtanino: if you think it's already covered well enough that's fine | 15:52 |
| jgriffith | mtanino: other than that the other two comments are just sort of nits, and I don't really care about them :) | 15:52 |
| mtanino | jgriffith: I will check and comment back. Thank you :) | 15:52 |
| *** _cjones_ has joined #openstack-cinder | 15:53 | |
| *** garthb has joined #openstack-cinder | 15:56 | |
| *** lpetrut has joined #openstack-cinder | 15:58 | |
| thingee | dulek: wrt https://review.openstack.org/#/c/183947/ | 16:00 |
| thingee | dulek: this is not a UI just fyo | 16:00 |
| thingee | fyi* | 16:00 |
| thingee | dulek: volume type management has existed | 16:00 |
| thingee | dulek: I still don't agree this is a good reason to store the metadata, as again, I don't want it stored. | 16:01 |
| thingee | backends change | 16:01 |
| *** dulek has quit IRC | 16:01 | |
| thingee | they get updated. | 16:01 |
| thingee | How often is a good question argument, but regardless, I don't want capabilities stored. | 16:01 |
| *** _cjones_ has quit IRC | 16:02 | |
| *** _cjones_ has joined #openstack-cinder | 16:02 | |
| *** bkopilov has quit IRC | 16:02 | |
| *** Longgeek has quit IRC | 16:02 | |
| thingee | jgriffith: I had a bit of questions for you yesterday wrt https://review.openstack.org/#/c/183947/ | 16:02 |
| *** Longgeek has joined #openstack-cinder | 16:02 | |
| *** agarciam has quit IRC | 16:03 | |
| jgriffith | thingee: scrolling back... | 16:03 |
| thingee | not sure if you saw them in the backlog, but could hash them out with you again if you have time | 16:03 |
| jgriffith | thingee: bahh... eavesdrop it is | 16:03 |
| jgriffith | thingee: yesterday? | 16:03 |
| thingee | jgriffith: I can just repeat and maybe be more clear :) | 16:03 |
| jgriffith | thingee: LOL.. yeah, there is always that :) | 16:03 |
| thingee | jgriffith: so take compression for example https://review.openstack.org/#/c/183947/2/specs/liberty/get-vol-type-extra-specs.rst,cm | 16:04 |
| thingee | Was compression_type meant to be a standard key in the capability? | 16:04 |
| jgriffith | thingee: AIR yes | 16:05 |
| *** belmoreira has quit IRC | 16:05 | |
| thingee | jgriffith: and does the vendor choose what those values are? | 16:05 |
| thingee | could be empty if they just say compression true? | 16:05 |
| jgriffith | thingee: for the compression_type yes | 16:06 |
| *** Yogi1 has quit IRC | 16:06 | |
| jgriffith | thingee: yeah, empty would be acceptable/fine | 16:06 |
| jgriffith | thingee: IIRC we agreeed the interesting key is "cinder_compression: True|False" | 16:06 |
| jgriffith | for scheduling | 16:06 |
| thingee | I'm wondering if it's a good idea for us to have vendor keys and the ability for vendors to choose what are the values at that level as well. Doesn't seem predictable. | 16:07 |
| jgriffith | thingee: we wanted to avoid having the scheduler filter on backend/internal keys | 16:07 |
| jgriffith | thingee: so I def think vendor-keys still exist and are useful | 16:07 |
| thingee | Definitely | 16:07 |
| jgriffith | thingee: but I was hoping to first just cover the "well-defined" cinder_ keys | 16:07 |
| jgriffith | thingee: know what I mean? | 16:08 |
| jgriffith | thingee: but honestly if you have a better idea and a clean way to mash all this together without being a nightmare :) I'm open to that for sure | 16:08 |
| jgriffith | I just don't know how well that works | 16:08 |
| *** Guest83839 has quit IRC | 16:08 | |
| *** bkopilov has joined #openstack-cinder | 16:08 | |
| jgriffith | so the idea was "return whatever crap you want" but we have "cinder_*" to designate the mandatory well defined keys | 16:09 |
| jgriffith | thingee: make sense? | 16:09 |
| *** annegentle has joined #openstack-cinder | 16:09 | |
| thingee | I agree. I'm just asking if you think vendors being able to define what compression_type and vendor keys are could lead to us not knowing what is actually a standard set of keys/values. If they're just defining vendor_keys, sure, they just define that. I can always assume the backend is going to pick what vendor_keys are. But if they also choose what the | 16:10 |
| thingee | values are up the structure with compression_type... now I'm not sure what cinder standardizes on and what vendors pick. Know what I mean? | 16:10 |
| *** jistr has quit IRC | 16:11 | |
| jgriffith | thingee: not sure... so my little brain just said (from Cinder's perspective) | 16:11 |
| jgriffith | thingee: "Give me a backend that supports compression" | 16:11 |
| jgriffith | done | 16:11 |
| jgriffith | that's all I care about, how you do it, what type etc that's all up to you | 16:11 |
| thingee | ok | 16:11 |
| jgriffith | because that's not a "standard" thing right? | 16:11 |
| thingee | I get compression=true/false | 16:11 |
| thingee | that's a standard. | 16:12 |
| *** bkopilov has quit IRC | 16:12 | |
| jgriffith | thingee: right, so what are you trying to solve with the "options" key? | 16:12 |
| thingee | I'm asking you because it was in your example. I'm trying to understand how this fits in. | 16:12 |
| jgriffith | thingee: sure, I'm just saying I'm not exactly clear on what issue you're seeing here | 16:13 |
| thingee | Ok let me start over | 16:13 |
| jgriffith | thingee: know what I mean? Trying to make sure there's not a problem that I didn't think of | 16:13 |
| thingee | we have vendor_keys...these are defined by the backend | 16:13 |
| thingee | we agree? | 16:13 |
| jgriffith | thingee: wait... let's try something different first | 16:13 |
| *** jordanP has quit IRC | 16:14 | |
| jgriffith | thingee: We have "capabilities" | 16:14 |
| jgriffith | thingee: there are two types of capabilities: | 16:14 |
| *** mdenny has joined #openstack-cinder | 16:14 | |
| jgriffith | thingee: 1. well-defined (cinder_*) | 16:14 |
| jgriffith | thingee: 2. vendor-specific | 16:14 |
| *** annegentle has quit IRC | 16:14 | |
| thingee | yup with you so far | 16:14 |
| jgriffith | thingee: agreed? | 16:14 |
| jgriffith | Ok... cool | 16:14 |
| jgriffith | so the idea is that the scheduler will be able to filter on anything that is of capability-category '1' | 16:15 |
| *** markus_z has quit IRC | 16:15 | |
| jgriffith | thingee: so anything with a prefix of "cinder_" is a base feature/capability | 16:15 |
| jgriffith | thingee: BUT, any key in a capability that doesn't include that is a vendor-unique thing | 16:16 |
| *** rushil has quit IRC | 16:16 | |
| jgriffith | thingee: it's only applicable to that backend ONCE a request is actually filtered to that backend via the "well-defined" parent key | 16:16 |
| jgriffith | or capability itself. | 16:16 |
| jgriffith | thingee: so sticking with the compression example | 16:16 |
| jgriffith | thingee: anything under "options' can/should be ignored by the world | 16:17 |
| jgriffith | thingee: except the driver itself | 16:17 |
| *** asselin_ has quit IRC | 16:17 | |
| jgriffith | thingee: does that make sense? Or does that make it more confusing :) | 16:17 |
| thingee | yes, but I do have a question | 16:18 |
| jgriffith | sure... *mabye* I have an answer :) | 16:18 |
| jgriffith | hopefully :) | 16:18 |
| thingee | jgriffith: https://gist.github.com/j-griffith/fd27297d79e0bb2c3e3d | 16:19 |
| thingee | so driver can have options however they want?... we don't define a structure in there? | 16:20 |
| jgriffith | thingee: so that proposal/gist is a little outdated | 16:20 |
| jgriffith | thingee: it was "shot down" remember? | 16:20 |
| jgriffith | thingee: but it does define a structure FWIW | 16:21 |
| hemna | thingee, can't we 'validate' the dictionary in the volume manager? | 16:21 |
| jgriffith | thingee: capability-type and default | 16:21 |
| hemna | and puke if they don't match the format? | 16:21 |
| hemna | dunno | 16:21 |
| hemna | json schema is good for that | 16:21 |
| hemna | you can actually validate a dict with it. | 16:22 |
| hemna | (if we wanted to go that route) | 16:22 |
| thingee | ok, so lets I define options how I want. How do I communicate that I have qos related keys min_iops, max_iops.... versus I have a qos related capability qos_persona (oh dear god): {'generic', 'foo', 'bar'} | 16:22 |
| *** Apoorva has joined #openstack-cinder | 16:22 | |
| jgriffith | thingee: well, I have th atin the gist | 16:22 |
| jgriffith | thingee: line 20 | 16:22 |
| thingee | {} was mean to be[] | 16:22 |
| jgriffith | vendor_keys | 16:22 |
| thingee | right, but qos_person is a list of options...only one can be chosen. min_iops, max_iops are possible keys that can have values set | 16:23 |
| jgriffith | qos_person ? | 16:23 |
| openstackgerrit | Anurag Palsule proposed openstack/cinder: Send pool free and total capacity metrics to ceilometer https://review.openstack.org/180447 | 16:23 |
| thingee | qos_persona...see my message a few up | 16:23 |
| thingee | jgriffith: all hypothetically. | 16:24 |
| jgriffith | thingee: well, I'm not sure why that matters | 16:24 |
| jgriffith | thingee: so let's stick with my idea (ie the gist) | 16:24 |
| thingee | jgriffith: cool | 16:25 |
| openstackgerrit | Earle F. Philhower, III proposed openstack/cinder: Implement Cinder Volume driver for HGST Solutions https://review.openstack.org/186580 | 16:25 |
| jgriffith | thingee: the scheduler looks at capabilities that have "capability_type == well-defined" | 16:25 |
| thingee | cinder_* | 16:25 |
| jgriffith | thingee: No, that's not in my gist :) | 16:25 |
| thingee | :) | 16:25 |
| thingee | I'm with ya | 16:25 |
| jgriffith | thingee: or my original example | 16:25 |
| jgriffith | so... | 16:25 |
| jgriffith | the "base" here for say qos is a capability "quality_of_service" | 16:26 |
| jgriffith | the scheduler picks up on it because it has the key "capability_type == well_defined" | 16:26 |
| jgriffith | the scheduler uses the "default" key only, which is "does the backend support this or not" | 16:27 |
| jgriffith | and that's all | 16:27 |
| jgriffith | the "vendor_keys" in the "options" section are just that, vendor specific keys | 16:27 |
| jgriffith | persona is a completely different thing, and purely vendor special stuff | 16:28 |
| jgriffith | that honestly I don't know what the hypothetical you mentioned could/would be, but I shouldn't care | 16:28 |
| jgriffith | and neither should the scheduler or anybody else in Cinder | 16:28 |
| * thingee feels like he can't communicate anything today | 16:29 | |
| thingee | yea persona isn't important | 16:29 |
| jgriffith | thingee: I'ms orry | 16:29 |
| jgriffith | sorry | 16:29 |
| thingee | but I get what you're saying | 16:29 |
| jgriffith | so FWIW I still like my original proposal that's there in the gist :) | 16:29 |
| jgriffith | and I still contend that it works really well :) | 16:30 |
| thingee | what's different from your original proposal besides the prefix? | 16:30 |
| jgriffith | thingee: not sure :) | 16:31 |
| asselin | nikeshm, don't know how to improve speed.....the dib images just cache quite a bit..... | 16:31 |
| thingee | jgriffith: I think it's the same pretty much | 16:31 |
| *** pradipta has joined #openstack-cinder | 16:31 | |
| thingee | jgriffith: also https://review.openstack.org/#/c/150511/ | 16:31 |
| *** bkopilov has joined #openstack-cinder | 16:31 | |
| *** ronis has quit IRC | 16:32 | |
| jgriffith | thingee: yeah, I think it is pretty close | 16:32 |
| *** julim has quit IRC | 16:32 | |
| *** tsekiyama has joined #openstack-cinder | 16:33 | |
| thingee | xyang1: for thick_provision addition, I would raise it in the standard capability spec. | 16:33 |
| jgriffith | thingee: I'm sorry I'm not following the question/issue | 16:33 |
| thingee | ^ | 16:33 |
| jgriffith | thingee: xyang1 wouldn't that just be "thin-provision=False" :) | 16:34 |
| thingee | jgriffith: +1 | 16:34 |
| thingee | that was my comments in the spec | 16:34 |
| thingee | jgriffith: no I think it's just me forgetting how the scheduler will see all this. | 16:34 |
| *** alexpilotti has joined #openstack-cinder | 16:35 | |
| thingee | jgriffith: however, I'm still a little worried about my initial question on allowing vendors to pick what is set in compression_type and options. | 16:35 |
| jgriffith | thingee: oh... so in your spec it would just see the keys: "cinder_*" | 16:35 |
| jgriffith | thingee: that's all | 16:35 |
| thingee | jgriffith: yes | 16:35 |
| jgriffith | thingee: or in my gist it was the ones with the "well_defined" value in them | 16:35 |
| jgriffith | thingee: oh... | 16:36 |
| thingee | jgriffith: I updated it based on the summit discussion and what winston-1 came up with | 16:36 |
| jgriffith | thingee: ok, why are you worried about that? | 16:36 |
| jgriffith | thingee: the options part? | 16:36 |
| *** timcl has quit IRC | 16:36 | |
| jgriffith | thingee: For the record, I've always said AFAIC vendor unique stuff doesn't belong here at all | 16:36 |
| thingee | jgriffith: so if vendors set options, that's fine. If we in addition say they set compression_type, that's now making it unpredictable of what vendors set. | 16:36 |
| thingee | I agree. that's why I'm asking you about this | 16:37 |
| jgriffith | thingee: why? I mean... the parent keys is 'compression=True|False' | 16:37 |
| jgriffith | who cares how they do it, just so long as they do it :) | 16:37 |
| jgriffith | thingee: OHHHH | 16:37 |
| jgriffith | thingee: ok, thin IMHO dump all the vendor unique stuff :) | 16:37 |
| jgriffith | thingee: come back to it later if we must | 16:38 |
| jgriffith | thingee: but I know hemna and the folks at HP aren't fans of that | 16:38 |
| *** timcl has joined #openstack-cinder | 16:38 | |
| jgriffith | s/thin/then/ :) | 16:38 |
| xyang1 | thingee: jgriffith , for the same pool, it could report thin provision=true and thick_provision=true | 16:38 |
| jgriffith | xyang1: oh yeah :) | 16:38 |
| jgriffith | xyang1: but.... | 16:38 |
| *** harlowja_at_home has quit IRC | 16:38 | |
| *** harlowja_at_home has joined #openstack-cinder | 16:38 | |
| jgriffith | xyang1: it can have options T|F | 16:39 |
| jgriffith | xyang1: so it can do both | 16:39 |
| jgriffith | xyang1: it's right there in thingee 's spec | 16:39 |
| jgriffith | xyang1: https://review.openstack.org/#/c/183947/2/specs/liberty/get-vol-type-extra-specs.rst,cm L#302 | 16:39 |
| xyang1 | jgriffith: Let me check | 16:40 |
| tbarron | thingee: jgriffith: Vendor A has min_iops, max_iops, burst_iops and reports these as qos-keys whereas Vendor B has only max_iops, which it reports. Both report qos_support=True. | 16:40 |
| *** thangp has quit IRC | 16:41 | |
| *** afazekas has joined #openstack-cinder | 16:41 | |
| thingee | tbarron: yeah that's true...could led to problems later once a cinder volume service receives it. | 16:41 |
| openstackgerrit | Patrick East proposed openstack/cinder-specs: Generic image volume cache functionality https://review.openstack.org/182520 | 16:41 |
| openstackgerrit | Patrick East proposed openstack/cinder-specs: Cinder internal tenant https://review.openstack.org/186232 | 16:41 |
| xyang1 | jgriffith: It says thin_provisioning=true,false | 16:41 |
| thingee | lead* | 16:41 |
| tbarron | Are you saying that the scheduler ignores the keys and will choose Backend A or Backend B indifferently if a volumetype with any kind of QOS is requested. | 16:41 |
| tbarron | ? | 16:42 |
| jgriffith | xyang1: so what do you get when you set thin-prov = False ? | 16:42 |
| jgriffith | xyang1: you get "thick" right? Or do you get a Unicorn? | 16:42 |
| xyang1 | jgriffith: maybe a unicorn:) | 16:42 |
| jgriffith | tbarron: yes, just like it does today | 16:42 |
| jgriffith | tbarron: have to start somewhere | 16:42 |
| thingee | tbarron: currently yes. I would like to hear what jgriffith thinks though. I think that might be a problem | 16:42 |
| thingee | ah ok | 16:42 |
| jgriffith | tbarron: we have to start "somewhere" | 16:43 |
| tbarron | similary for compression: vendor A has compression-type zlib or gzip, vendor B has inline or background with incompressible data detection as another option | 16:43 |
| xyang1 | jgriffith: So the pool for vnx supports both thin and thick | 16:43 |
| tbarron | jgriffith: ok | 16:43 |
| jgriffith | tbarron: frankly I don't really think this is as contentious or difficult as we've made it out to be for the past year and a half | 16:43 |
| jgriffith | tbarron: it would be awesome if we at least 'started' something that was on the same level of what we have today | 16:43 |
| jgriffith | tbarron: and then we could grow it | 16:44 |
| tbarron | tbarron: I agree with the goal, uncontentious, move forwards, etc. Am just clarifying for myself. | 16:44 |
| thingee | jgriffith: it's just not what the admin is going to expect in making extra specs that we're suppose to guarantee something | 16:44 |
| jgriffith | rather than 'what if'ing' it to death every time it comes up | 16:44 |
| jgriffith | thingee: I don't understand why not | 16:44 |
| xyang1 | jgriffith: So depdending on what you specify in volume type, it will create a thin or thick, but driver reports both as true | 16:44 |
| jgriffith | xyang1: ? | 16:44 |
| tbarron | Do we need to set expectations for how the backend will handle extra specs/qos-specs that it can't truly support? | 16:44 |
| jgriffith | xyang1: so... look: | 16:44 |
| thingee | if I say min_iops and that's not honored, because of cinder_quality_of_service=true, we have a problem | 16:44 |
| jgriffith | xyang1: cinder create --volume-type THICK | 16:45 |
| thingee | jgriffith: ^ | 16:45 |
| tbarron | That is, if I can't support min-iops do I fail the create, or silently ignore and provision with the keys I understand? | 16:45 |
| jgriffith | scheduler says... ok, give me any backend that reports "false" in the options for thin-provisioning | 16:45 |
| thingee | xyang1: fwiw, I see what jgriffith is proposing and I think it supports your case. | 16:45 |
| jgriffith | tbarron: what do you do right now? | 16:45 |
| jgriffith | tbarron: ie what if somebody sends you the qos specs for a SolidFire volume? | 16:46 |
| tbarron | jgriffith: I fail the create right now. | 16:46 |
| jgriffith | tbarron: well then there's your answer :) | 16:46 |
| tbarron | I fail it if I see any keywords for backend qos that don't make sense to me. | 16:46 |
| jgriffith | tbarron: and IMO that's the right thing to do (even though I'm not doing it) :( | 16:46 |
| tbarron | jgriffith: but I think some drivers silently move ahead (kinda like your retype issue) | 16:47 |
| thingee | tbarron: seems right to me | 16:47 |
| xyang1 | jgriffith: thingee oh, I see, maybe that will work | 16:47 |
| * jgriffith says 'truuust me' :) | 16:47 | |
| thingee | jgriffith: I see this as a problem if that's your current proposal. I would not want the scheduler to pick a backend. | 16:47 |
| *** thangp has joined #openstack-cinder | 16:47 | |
| *** harlowja_still_a has joined #openstack-cinder | 16:47 | |
| tbarron | So maybe that expectation should be stated in the spec so bugs can be filed on drivers that don't do it. | 16:47 |
| jgriffith | thingee: ok, then rewrite it however you like | 16:47 |
| thingee | if it can't support the min_iops. | 16:47 |
| thingee | jgriffith: :) | 16:47 |
| jgriffith | thingee: honestly I bowed out of this a while back for this very reason. And you specifically said you had a solution and wanted to work on it | 16:48 |
| jgriffith | thingee: I'm fine with that | 16:48 |
| tbarron | thingee: I see jgriffith's point that the scheduler already has this limitation and this is still a step forward. | 16:48 |
| *** rushil has joined #openstack-cinder | 16:48 | |
| jgriffith | thingee: my only point is that if you're trying to introduce standard-capabilities, vendor-capabilities, and fix ALL of the existing problems, and thwart ANY future problems all at the same time you'll never succeeed | 16:49 |
| tbarron | In the mean time, some drivers will have bugs that should be flushed out. | 16:49 |
| thingee | jgriffith: I know you're focused on the scheduler aspect. My main reason of wanting to work on this was making the keys clear to an admin creating extra specs. | 16:49 |
| jgriffith | tbarron: +1 | 16:49 |
| jgriffith | thingee: ok, cool by me | 16:49 |
| tbarron | And eventually a phase2 could make the scheduler keywrord aware. | 16:49 |
| jgriffith | thingee: but then don't say "what if the scheduler sends it to the wrong backend" | 16:49 |
| *** afazekas has quit IRC | 16:49 | |
| jgriffith | thingee: that's contradictory to what you just said :) | 16:49 |
| jgriffith | see what I'm saying? | 16:50 |
| tbarron | if phase2 can be done now, then so much the better! | 16:50 |
| jgriffith | we tend to get into this contradictory statements rather than just focusing on one piece and solving it | 16:50 |
| * thingee is going to assume he's not making sense to people today and will just continue to update the spec | 16:50 | |
| tbarron | thingee: stop making sense | 16:50 |
| jgriffith | thingee: nahh... don't say that | 16:51 |
| * thingee starts listening to talking heads | 16:51 | |
| *** harlowja_at_home has quit IRC | 16:51 | |
| *** merooney has joined #openstack-cinder | 16:51 | |
| jgriffith | thingee: you're bringing up some valid points | 16:51 |
| thingee | stop making sense stop make sense | 16:51 |
| jgriffith | thingee: and I think people understand what you're saying | 16:51 |
| jgriffith | thingee: I'm just saying don't try and "fix the world" | 16:51 |
| jgriffith | thingee: especially in the case of things that aren't "broken" :) | 16:51 |
| tbarron | I'm reminded of karaoke santa singing psycho killer | 16:51 |
| tbarron | qu'est que c'est | 16:52 |
| jgriffith | Qu'est-ce que c'est | 16:52 |
| jgriffith | Fa-fa-fa-fa-fa-fa-fa-fa-fa-far better | 16:52 |
| tbarron | jgriffith: outspelled me! | 16:52 |
| jgriffith | tbarron: make a note, will likely NEVER happen again :) | 16:52 |
| tbarron | jgriffith: heh | 16:53 |
| jgriffith | alright, I gotta leave for a bit | 16:53 |
| jgriffith | bbl | 16:53 |
| tsekiyama | Is the Jenkins' pep8 for cinderclient broken now? I get pkg_resources.VersionConflict error for pbr in https://review.openstack.org/#/c/180361/ | 16:53 |
| jgriffith | thingee: sorry if I just made things worse, I don't know what else to do on this | 16:53 |
| thingee | Yeah I just have questions and I can't express them for some reason | 16:53 |
| *** ajayaa has joined #openstack-cinder | 16:55 | |
| *** e0ne has quit IRC | 16:55 | |
| tsekiyama | ^^ Looks like other changes are failing too so far https://review.openstack.org/#/q/project:openstack/python-cinderclient,n,z | 16:56 |
| *** harlowja_still_a has quit IRC | 17:01 | |
| *** merooney has quit IRC | 17:02 | |
| *** sgotliv_ has quit IRC | 17:04 | |
| *** merooney has joined #openstack-cinder | 17:06 | |
| *** Yogi1 has joined #openstack-cinder | 17:08 | |
| *** merooney has quit IRC | 17:10 | |
| *** annegentle has joined #openstack-cinder | 17:10 | |
| *** merooney has joined #openstack-cinder | 17:10 | |
| openstackgerrit | Gorka Eguileor proposed openstack/cinder: Fix cinder concurrency issues on rtstool https://review.openstack.org/187260 | 17:11 |
| *** vilobhmm has joined #openstack-cinder | 17:13 | |
| *** ajayaa has quit IRC | 17:14 | |
| *** vilobhmm1 has joined #openstack-cinder | 17:15 | |
| thingee | jgriffith: I appreciate your help though | 17:16 |
| *** merooeny_ has joined #openstack-cinder | 17:17 | |
| *** annegentle has quit IRC | 17:17 | |
| *** vilobhmm has quit IRC | 17:17 | |
| *** merooney has quit IRC | 17:21 | |
| *** annegentle has joined #openstack-cinder | 17:21 | |
| *** ronis has joined #openstack-cinder | 17:23 | |
| *** julim has joined #openstack-cinder | 17:27 | |
| *** julim has quit IRC | 17:27 | |
| *** julim has joined #openstack-cinder | 17:28 | |
| *** julim has quit IRC | 17:28 | |
| *** julim has joined #openstack-cinder | 17:28 | |
| *** julim has quit IRC | 17:29 | |
| *** julim has joined #openstack-cinder | 17:29 | |
| *** julim has quit IRC | 17:30 | |
| *** julim has joined #openstack-cinder | 17:31 | |
| *** merooeny_ has quit IRC | 17:33 | |
| *** jdurgin has joined #openstack-cinder | 17:34 | |
| *** julim has quit IRC | 17:35 | |
| *** julim has joined #openstack-cinder | 17:36 | |
| *** julim has quit IRC | 17:36 | |
| *** julim has joined #openstack-cinder | 17:37 | |
| *** julim has quit IRC | 17:37 | |
| *** vilobhmm1 has quit IRC | 17:38 | |
| *** julim has joined #openstack-cinder | 17:38 | |
| *** vilobhmm has joined #openstack-cinder | 17:38 | |
| *** julim has quit IRC | 17:38 | |
| *** julim has joined #openstack-cinder | 17:38 | |
| *** julim has quit IRC | 17:39 | |
| *** julim has joined #openstack-cinder | 17:39 | |
| *** afazekas has joined #openstack-cinder | 17:40 | |
| *** julim has quit IRC | 17:40 | |
| *** julim has joined #openstack-cinder | 17:41 | |
| *** julim has quit IRC | 17:41 | |
| nikeshm | hemna: hi now FC and iSCSI are both working on https://review.openstack.org/#/c/177665/5,can i expect one more +2 :) | 17:41 |
| *** julim has joined #openstack-cinder | 17:41 | |
| *** julim has quit IRC | 17:42 | |
| *** julim has joined #openstack-cinder | 17:42 | |
| *** julim has quit IRC | 17:42 | |
| *** julim has joined #openstack-cinder | 17:43 | |
| *** julim has quit IRC | 17:43 | |
| *** julim has joined #openstack-cinder | 17:43 | |
| *** julim has quit IRC | 17:44 | |
| *** julim has joined #openstack-cinder | 17:44 | |
| *** julim has quit IRC | 17:44 | |
| *** julim has joined #openstack-cinder | 17:45 | |
| tbarron | thingee: jgriffith: jungleboyj: I've got a couple kilo backports that it would be good to clear when you have a chance. | 17:45 |
| *** julim has quit IRC | 17:45 | |
| *** julim has joined #openstack-cinder | 17:46 | |
| tbarron | thingee: jgriffith: jungleboyj: https://review.openstack.org/#/c/184154/ fixes a help string and is necessary for kilo config doc to be corrected (the doc autogenerates config option tables). | 17:46 |
| *** julim has quit IRC | 17:46 | |
| *** julim has joined #openstack-cinder | 17:47 | |
| *** julim has quit IRC | 17:47 | |
| *** afazekas has quit IRC | 17:47 | |
| tbarron | thingee: jgriffith: jungleboyj: https://review.openstack.org/#/c/185724/ fixes an issue in the field that our customers hit. It's scope is limited to our NFS drivers. | 17:47 |
| tbarron | thingee: jgriffith: jungleboyj: I see jungleboyj has +2s on these already. | 17:47 |
| *** julim has joined #openstack-cinder | 17:47 | |
| *** julim has quit IRC | 17:48 | |
| thingee | tbarron, jungleboyj, jgriffith: unsure about not supporting over subscription in kilo is a bug. A lot of drivers don't. I won't be voting for it. | 17:49 |
| *** julim has joined #openstack-cinder | 17:49 | |
| *** julim has quit IRC | 17:49 | |
| tbarron | thingee: I see. Our customers certainly think it's broken though. They can configure the option and we don't honor it. | 17:50 |
| thingee | tbarron, jungleboyj, jgriffith otherwise I should be back porting every driver that wants to do this. | 17:50 |
| *** julim has joined #openstack-cinder | 17:50 | |
| thingee | That's not what back ports are for. | 17:50 |
| *** julim has quit IRC | 17:50 | |
| thingee | You're working around the point of back ports. | 17:50 |
| tbarron | thingee: to be clear, this isn't xing/winston oversubscription, it's old-fashioned NFS oversubscription, there since day zero. | 17:50 |
| tbarron | not the new feature | 17:51 |
| *** julim has joined #openstack-cinder | 17:51 | |
| tbarron | thingee: just want to get the case out there clearly, but I won't persist in arguing further on IRC. I do respect whatever decision you come to. | 17:51 |
| thingee | tbarron: I'll take a closer look, but last I looked this was to support this new over subscription way by reporting the appropriate keys. | 17:52 |
| thingee | Bbl currently in transit. | 17:52 |
| tbarron | thingee: I try to distinguish between those two in the bug, thanks for looking. | 17:53 |
| thingee | tbarron: I don't need you distinguish anything. I read the patch and what is in xyang1's spec. | 17:54 |
| tbarron | thingee: agreed | 17:54 |
| thingee | For oversubscription. | 17:54 |
| tbarron | thingee: you'll see I'm working with old, pre-xyang-spec keys in this fix. | 17:55 |
| nikeshm | what is this any mallware someone continuosly login and logout | 17:55 |
| tbarron | I do play to follow with a liberty-only implementation of the xyang spec. | 17:55 |
| tbarron | for our drivers. But that's another thing. | 17:55 |
| *** julim has quit IRC | 17:56 | |
| thingee | tbarron: cool I'll take a closer look when I have a chance. | 17:56 |
| tbarron | That one will deprecate these older keys/config options for one release, as discussed with eharney, and introduce the new way. | 17:56 |
| tbarron | new way having been introduced in kilo, but we wont' try to backport support of that to kilo :-) | 17:57 |
| tbarron | thingee: thanks, nuf said by me on this for now | 17:57 |
| *** zhenguo has quit IRC | 17:57 | |
| jungleboyj | tbarron: thingee I will look again. | 17:59 |
| tbarron | jungleboyj: ty! | 18:03 |
| *** pradipta has quit IRC | 18:03 | |
| *** alexpilotti has quit IRC | 18:03 | |
| *** aix has quit IRC | 18:04 | |
| *** lcurtis has joined #openstack-cinder | 18:05 | |
| *** julim has joined #openstack-cinder | 18:05 | |
| *** rushil has quit IRC | 18:06 | |
| *** rushil has joined #openstack-cinder | 18:08 | |
| jungleboyj | tbarron: Hmmm, ok ... I see you have addressed what appears to be thingee s concern in the commit message. I don't honestly have a strong feeling on this one. | 18:09 |
| *** lpabon has joined #openstack-cinder | 18:09 | |
| *** anshul has quit IRC | 18:09 | |
| openstackgerrit | Jon Bernard proposed openstack/cinder: WIP: Add support for generic volume migration https://review.openstack.org/187270 | 18:13 |
| *** IanGovett1 has quit IRC | 18:14 | |
| openstackgerrit | Mitsuhiro Tanino proposed openstack/cinder-specs: Efficient volume copy for volume migration https://review.openstack.org/186209 | 18:15 |
| winston-d | thingee, jgriffith: clarification - CapabilitiesFilter is able to act on anything being put to type extra_specs, either without any scope or unders the scope of 'capabilities:' | 18:21 |
| jgriffith | winston-d: yeah, I was trying to leave that part out for now :) | 18:22 |
| jgriffith | winston-d: in other words, try and just get the "standard" stuff done first | 18:22 |
| jgriffith | winston-d: then build on it | 18:22 |
| jgriffith | winston-d: it's causing way too much confusion to try and get everybody's heads wrapped around it right now I htink | 18:22 |
| jgriffith | think | 18:22 |
| jgriffith | winston-d: but maybe that's why it's confusing | 18:23 |
| winston-d | thingee, jgriffith: for example - if one put 'FancyCompressionMagicNeeded = True' into extra spec, and the FancyStor driver reports that, then schedule will only let FancyStor pass this filter. | 18:23 |
| jgriffith | winston-d: right, as long as it's not scoped.. correct? | 18:24 |
| jgriffith | winston-d: which is same as we have currently | 18:24 |
| winston-d | jgriffith: right, as long as it's not scoped, or scoped under 'capabilities:' | 18:24 |
| jgriffith | winston-d: if you put something in there that isn'r reported and leave the scoping out the filter scheduler says "umm.. sorry, no valid host found" | 18:24 |
| winston-d | correct | 18:25 |
| *** eharney has quit IRC | 18:25 | |
| jgriffith | winston-d: yeah, which would be "phase 2" IMHO if we got the standard stuff ironed out and submitted | 18:25 |
| jgriffith | winston-d: and eventually maybe just get rid of scoped keys altogether :) | 18:25 |
| winston-d | scoped key can be a easy trick for admin to put vendor specific/driver-consumed-only stuff into type extra specs. | 18:27 |
| winston-d | jgriffith: burst_iops for SF for example. | 18:27 |
| jgriffith | winston-d: yeah... so I need to look at updating SF driver to just go directly through filter like that | 18:27 |
| jgriffith | winston-d: it would be nice to not need to specify volume_backend_name any longer | 18:28 |
| jgriffith | winston-d: one less thing to type in :) | 18:28 |
| nikeshm | thingee: thanks for removing -2,hope sooner will get +2,i am going to upload new patches for lenovo and hpmsa,might be today or tomorrow which are dependent on https://review.openstack.org/#/c/177665/5 | 18:28 |
| jgriffith | winston-d: the only thing I had a bit of a struggle with there was the case where I require all 3 (I don't want to set defaults) | 18:29 |
| *** eharney has joined #openstack-cinder | 18:29 | |
| jgriffith | winston-d: but maybe just a simple algorithm based on percentages off of what's provided woudl be ok | 18:29 |
| jgriffith | winston-d: Ohhhhhhh.... even better | 18:29 |
| jgriffith | winston-d: let the admin pass a formula in config (or use default) that does it for us | 18:29 |
| winston-d | jgriffith: yeah, volume_backend_name is the last thing Avisay and I would like to see in type extra spec | 18:29 |
| jgriffith | winston-d: hmm... Ok | 18:29 |
| *** julim has quit IRC | 18:30 | |
| * jgriffith adds another work item to his list :) | 18:30 | |
| winston-d | SF driver 3.0 | 18:30 |
| jgriffith | winston-d: yeah, I think so | 18:30 |
| jgriffith | winston-d: there's a crap load of things on my list now that can be "mo-better" | 18:31 |
| jgriffith | winston-d: particularly since we don't have anybody below version 6 any more :) | 18:31 |
| *** cknight has joined #openstack-cinder | 18:31 | |
| jgriffith | winston-d: makes my life much easier!!! | 18:31 |
| *** merooney has joined #openstack-cinder | 18:33 | |
| winston-d | jgriffith: yeah, i'm sure that helps a lot from supporting perspective. | 18:34 |
| jgriffith | winston-d: indeed! | 18:35 |
| *** krtaylor has quit IRC | 18:37 | |
| *** dannywilson has quit IRC | 18:46 | |
| *** dannywilson has joined #openstack-cinder | 18:46 | |
| *** merooney has quit IRC | 18:46 | |
| openstackgerrit | Patrick East proposed openstack/cinder-specs: Generic image volume cache functionality https://review.openstack.org/182520 | 18:47 |
| openstackgerrit | Patrick East proposed openstack/cinder-specs: Cinder internal tenant https://review.openstack.org/186232 | 18:47 |
| *** patrickeast has quit IRC | 18:50 | |
| *** patrickeast_ has joined #openstack-cinder | 18:50 | |
| *** anshul has joined #openstack-cinder | 18:50 | |
| *** julim has joined #openstack-cinder | 18:50 | |
| *** rushil has quit IRC | 18:54 | |
| *** Yogi1 has quit IRC | 18:56 | |
| *** Apoorva has quit IRC | 18:56 | |
| thingee | tbarron: responded back to that review. you were right. | 19:00 |
| thingee | tbarron: sorry about that | 19:00 |
| thingee | tbarron: I'll wait to see if there are any objections, and merge it at some point | 19:00 |
| *** annegentle has quit IRC | 19:00 | |
| *** rushil has joined #openstack-cinder | 19:02 | |
| *** Yogi1 has joined #openstack-cinder | 19:02 | |
| tbarron | thingee: thanks for looking! | 19:06 |
| *** e0ne has joined #openstack-cinder | 19:10 | |
| thrawn01 | /join #graphite | 19:12 |
| thrawn01 | oops. ther must be a space there | 19:12 |
| jungleboyj | thingee: Yay, I am glad that was your impression. I had looked at it and not had the same concern. | 19:13 |
| *** erlon has joined #openstack-cinder | 19:18 | |
| openstackgerrit | Tomoki Sekiyama proposed openstack/python-cinderclient: Support host-attach of volumes https://review.openstack.org/180361 | 19:21 |
| openstackgerrit | Tomoki Sekiyama proposed openstack/python-cinderclient: Bump hacking to >=0.10,<0.11 to fix failure of gate pep8 https://review.openstack.org/187295 | 19:21 |
| *** merooney has joined #openstack-cinder | 19:22 | |
| *** patrickeast_ has quit IRC | 19:23 | |
| *** e0ne has quit IRC | 19:24 | |
| openstackgerrit | Scott DAngelo proposed openstack/python-cinderclient: Nova volume-detach --force requires API https://review.openstack.org/187305 | 19:33 |
| *** annegentle has joined #openstack-cinder | 19:33 | |
| *** annegentle has quit IRC | 19:35 | |
| *** annegentle has joined #openstack-cinder | 19:35 | |
| *** anshul has quit IRC | 19:37 | |
| thingee | jgriffith, jungleboyj, xyang1 https://review.openstack.org/#/c/187295/1 someone mind +2ing this? | 19:38 |
| *** daneyon_ has joined #openstack-cinder | 19:39 | |
| openstackgerrit | Tomoki Sekiyama proposed openstack/python-cinderclient: Support host-attach of volumes https://review.openstack.org/180361 | 19:40 |
| openstackgerrit | Tomoki Sekiyama proposed openstack/python-cinderclient: Bump hacking to >=0.10.0,<0.11 to fix failure of gate pep8 https://review.openstack.org/187295 | 19:40 |
| *** belmoreira has joined #openstack-cinder | 19:41 | |
| *** jungleboyj has quit IRC | 19:41 | |
| *** dannywilson has quit IRC | 19:42 | |
| *** daneyon has quit IRC | 19:42 | |
| thingee | jgriffith: oh right jenkins :P ... this new gerrit interface I swear... | 19:42 |
| tsbot_ | ^Fixed a little bit.. it should be 0.10.0, not 0.10 | 19:43 |
| thingee | tsbot_: I thought that too, but test_requirements.txt says 0.10 | 19:43 |
| thingee | I know requirements says otherwise | 19:43 |
| thingee | confusing | 19:43 |
| *** jungleboyj has joined #openstack-cinder | 19:43 | |
| tsekiyama | global-requirements.txt says 0.10.0, and test-requirements.txt says 0.10... confusing. | 19:44 |
| *** Longgeek has quit IRC | 19:50 | |
| *** Apoorva has joined #openstack-cinder | 19:50 | |
| *** dannywilson has joined #openstack-cinder | 19:51 | |
| *** annegentle has quit IRC | 19:51 | |
| *** patrickeast has joined #openstack-cinder | 19:53 | |
| *** belmoreira has quit IRC | 19:56 | |
| *** merooney has quit IRC | 20:00 | |
| *** jungleboyj has quit IRC | 20:04 | |
| openstackgerrit | John Griffith proposed openstack/cinder: Add secondary account capability to SolidFire https://review.openstack.org/187311 | 20:07 |
| *** merooney has joined #openstack-cinder | 20:08 | |
| vilobhmm | hemna, jgriffith : the connection_info field in nova.block_device_mapping gets set when we first time try to attach a volume or it should be initialized even before that…looking into https://github.com/openstack/nova/blob/master/nova/virt/block_device.py#L237 which will call volume/manager.py initialize_connection... | 20:11 |
| vilobhmm | but not able to figure that out...need some help | 20:12 |
| *** merooney has quit IRC | 20:12 | |
| *** merooney has joined #openstack-cinder | 20:17 | |
| *** merooney has quit IRC | 20:18 | |
| *** annegentle has joined #openstack-cinder | 20:19 | |
| *** Lee1092 has quit IRC | 20:21 | |
| *** crose has quit IRC | 20:21 | |
| jgriffith | vilobhmm: not following what you're asking? | 20:21 |
| *** belmoreira has joined #openstack-cinder | 20:21 | |
| jgriffith | vilobhmm: I think you're looking for the higher up reserve call | 20:22 |
| *** krtaylor has joined #openstack-cinder | 20:22 | |
| *** merooney has joined #openstack-cinder | 20:22 | |
| vilobhmm | so there is nova.block_device_mapping which also consist of volume information and also details like connection_info, device_name etc details | 20:23 |
| jgriffith | vilobhmm: yes | 20:23 |
| vilobhmm | I want to know when the connection_info is set ? (which has basically the route to the remote storage like protocol used etc….) | 20:24 |
| vilobhmm | is it during first attach ? or something before attach ? | 20:24 |
| vilobhmm | because i see my volumes getting stuck in attching state and the reason for that I can see that nova.block_device_mapping has no information about the connection_info | 20:24 |
| jgriffith | vilobhmm: look at nova/compute/api.py: def _attach_volume | 20:24 |
| jgriffith | vilobhmm: there's a reserve section in there that I *think* is what you're looking for | 20:25 |
| jgriffith | but not sure | 20:25 |
| *** merooney has quit IRC | 20:26 | |
| jgriffith | vilobhmm: I have no idea what/why your driver is or isn't working | 20:26 |
| jgriffith | vilobhmm: or even "what" your driver is for that matter :) | 20:26 |
| vilobhmm | will look into https://github.com/openstack/nova/blob/master/nova/compute/api.py#L2999 | 20:27 |
| vilobhmm | reserve_block_device_name | 20:27 |
| vilobhmm | so if i understand you correctly you mean to say the connection_info is set as part of reserve_block_device_name. Is that correct jgriffith ? | 20:28 |
| jgriffith | vilobhmm: depends | 20:29 |
| jgriffith | vilobhmm: what is your driver? | 20:29 |
| vilobhmm | rbd driver | 20:29 |
| jgriffith | vilobhmm: Ok, so what problem are you having? | 20:29 |
| vilobhmm | not able to attach volumes to an instance as they are getting stuck in "attaching" state | 20:30 |
| tsekiyama | OK, now passed the CI tests for https://review.openstack.org/#/c/187295/ | 20:30 |
| jgriffith | vilobhmm: what's different from your code vs what's in master and passing in the gate etc? | 20:30 |
| jgriffith | vilobhmm: most of that info is going to come from the provider_* info | 20:31 |
| vilobhmm | jgriffith : this is not master | 20:31 |
| jgriffith | vilobhmm: well obviously | 20:31 |
| jgriffith | vilobhmm: my point is that you have a branch that you changed *something* | 20:31 |
| vilobhmm | nope | 20:31 |
| vilobhmm | i am using upstream code just for a poc | 20:32 |
| jgriffith | vilobhmm: "upstream" where? I thought OpenStack/Master "was" upstream? | 20:32 |
| vilobhmm | icehouse | 20:32 |
| vilobhmm | yes you r right | 20:32 |
| jgriffith | vilobhmm: so it may just be that you don't have your stuff configured | 20:32 |
| jgriffith | correctly | 20:32 |
| jgriffith | vilobhmm: which is obviously what you're probably asking :) | 20:33 |
| vilobhmm | nope | 20:33 |
| jgriffith | vilobhmm: Ok | 20:33 |
| jgriffith | vilobhmm: well, I tried | 20:33 |
| vilobhmm | i am not asking some config thing here I want to understand | 20:33 |
| jgriffith | vilobhmm: ok | 20:33 |
| vilobhmm | when does nova.block_Device_mapping gets the connection_info as I was not able to figure it out myself... | 20:34 |
| *** julim has quit IRC | 20:34 | |
| vilobhmm | aw…will figure out | 20:34 |
| vilobhmm | thanks jgriffith :) | 20:35 |
| *** jungleboyj has joined #openstack-cinder | 20:36 | |
| jgriffith | vilobhmm: https://github.com/openstack/nova/blob/stable/icehouse/nova/compute/api.py#L2792 | 20:36 |
| jgriffith | vilobhmm: so I'm not sure, but that's what i was saying earlier "might" be what you're looking for | 20:36 |
| jgriffith | vilobhmm: but not entirely sure as you've guessed by now ;) | 20:37 |
| vilobhmm | sure | 20:37 |
| vilobhmm | that helped | 20:37 |
| vilobhmm | thnx | 20:38 |
| jgriffith | vilobhmm: np, sorry I don't have a better response, maybe hemna will | 20:38 |
| *** lpabon has quit IRC | 20:38 | |
| hemna | uh uh, what'd I do ? | 20:38 |
| jgriffith | vilobhmm: is tracing out attach on RBD | 20:39 |
| jgriffith | vilobhmm: had some questions | 20:39 |
| hemna | ok | 20:39 |
| hemna | against Icehouse ? | 20:39 |
| vilobhmm | yes | 20:39 |
| vilobhmm | https://github.com/openstack/nova/blob/stable/icehouse/nova/compute/api.py#L2789 | 20:39 |
| hemna | huh ok, that's ancient, but ok | 20:39 |
| jgriffith | hemna: it hasn't changed really | 20:40 |
| jgriffith | hemna: besides.. it's only a year :) | 20:40 |
| *** belmoreira has quit IRC | 20:40 | |
| hemna | for the most part afaik, nova takes what cinder's initialize_connection gives it, and passes it to the libvirt volume driver for the transport | 20:40 |
| vilobhmm | ok hemna | 20:41 |
| vilobhmm | so thats what is missing i guess | 20:41 |
| hemna | sorry if I'm not anwering the question | 20:41 |
| vilobhmm | cinder is not able to pass the right info | 20:41 |
| vilobhmm | you are | 20:41 |
| hemna | the connection_info goes directly into the libvirt volume's attach_volume call | 20:42 |
| vilobhmm | hemna : I don't see connection_info in nova.block_device_mapping which I guess is sent by cinder's initialize_connection call | 20:42 |
| vilobhmm | sure | 20:42 |
| *** ronis has quit IRC | 20:42 | |
| hemna | then it's the responsibility of the libvirt volume driver to discover the volume and get a block device mapping and pass that back, which gets injected into the BDM | 20:42 |
| vilobhmm | which means cinder's initialize_connection call is failing | 20:42 |
| hemna | I don't think nova saves the entire part of cinder's connection_info into the BDM | 20:43 |
| hemna | the libvirt volume driver uses the connection_info to discover the block device in the host | 20:43 |
| hemna | then passes that back to get saved | 20:43 |
| vilobhmm | not entire but something of this sort …{"driver_volume_type": "rbd", "serial": "a3a00411-7d23-4697-bb2b-367c9a23f115", "data": {"secret_type": "ceph", "device_path": null, "name": "volumes/volume-a3a00411-7d23-4697-bb2b-367c9a23f115", "secret_uuid": "31ba14e8-93bc-47ca-af4f-369e479a9ee2", "qos_specs": null, "hosts": [….] | 20:44 |
| jgriffith | smcginnis: yeah, so https://goo.gl/X294vJ | 20:44 |
| hemna | so it's entirely dependent on the libvirt volume driver to figure out how to discover that block device as it shows up for the protocol | 20:44 |
| vilobhmm | sure | 20:44 |
| openstackgerrit | OpenStack Proposal Bot proposed openstack/cinder: Updated from global requirements https://review.openstack.org/186112 | 20:44 |
| jgriffith | smcginnis: prints a benign message on the tox args | 20:44 |
| jgriffith | smcginnis: but when you run you should notice on {0} - {3} | 20:44 |
| jgriffith | smcginnis: In other words it works, although it's not very elegant | 20:45 |
| jgriffith | smcginnis: until we get cinder stable at 8, figured we could deal with that | 20:45 |
| jgriffith | smcginnis: then just lift the limit altogether | 20:45 |
| vilobhmm | hemna : that was helpful...thanks | 20:45 |
| *** rushil has quit IRC | 20:45 | |
| hemna | vilobhmm, sure | 20:45 |
| jgriffith | smcginnis: although, actaully, I'm not sure I see the same thing you pointed out | 20:46 |
| *** merooney has joined #openstack-cinder | 20:46 | |
| nikeshm | hemna: hi | 20:46 |
| hemna | nikeshm, hey | 20:46 |
| *** rushil has joined #openstack-cinder | 20:47 | |
| * hemna has been slammed doing 'paperwork' todya | 20:47 | |
| hemna | bleh | 20:47 |
| nikeshm | hemna: now FC and iSCSI both are working on https://review.openstack.org/#/c/177665/5. can you give +2,i have to upload patches for hpmsa today or tommorow which are dependent on these,if it is merged,then will easy to set up CI for hpmsa drivers | 20:47 |
| hemna | nikeshm, ok I'll take a look | 20:48 |
| nikeshm | thanks | 20:48 |
| *** Yogi1 has quit IRC | 20:49 | |
| *** merooney has quit IRC | 20:49 | |
| smcginnis | jgriffith: Looks like I get that message because one of the unit tests is failing. | 20:51 |
| smcginnis | jgriffith: It does appear to be harmless having the extra "concurrency" parameter. | 20:52 |
| smcginnis | jgriffith: Just a little confusing maybe. | 20:52 |
| smcginnis | jgriffith: Any way to get it to not pass the default --concurrency=6 if a differency level is passed in? | 20:53 |
| *** cknight has quit IRC | 20:58 | |
| *** merooney has joined #openstack-cinder | 20:59 | |
| *** merooney has quit IRC | 21:04 | |
| smcginnis | jgriffith: Your paste shows the same command on line 5. | 21:05 |
| *** dustins has quit IRC | 21:05 | |
| smcginnis | jgriffith: I believe I just get the InvocationError on the end because one of the UTs fail. | 21:05 |
| smcginnis | jgriffith: You do have the note in tox.ini, so I guess I'm fine with it. | 21:07 |
| smcginnis | jgriffith: Just wasn't immediately obvious and expected. | 21:07 |
| *** thangp has quit IRC | 21:07 | |
| nikeshm | hemna: ohhh slammed with paperwork, | 21:08 |
| smcginnis | jgriffith: OK, you got my very first +2. ;) | 21:08 |
| *** cpallares has quit IRC | 21:08 | |
| *** eharney has quit IRC | 21:09 | |
| hemna | smcginnis, wielding his new abilities! :) | 21:10 |
| smcginnis | hemna: I feel drunk with power. Mwahahahahah. | 21:10 |
| hemna | :P | 21:11 |
| *** dannywilson has quit IRC | 21:11 | |
| *** dannywilson has joined #openstack-cinder | 21:13 | |
| *** cpallares has joined #openstack-cinder | 21:14 | |
| *** bswartz has joined #openstack-cinder | 21:18 | |
| *** cpallares has quit IRC | 21:19 | |
| tsekiyama | jgriffith: Could you look at https://review.openstack.org/#/c/187295/ again? needs last +A. | 21:21 |
| *** lpetrut has quit IRC | 21:21 | |
| *** lpetrut has joined #openstack-cinder | 21:22 | |
| *** lpetrut has quit IRC | 21:22 | |
| *** timcl has quit IRC | 21:26 | |
| *** akerr has quit IRC | 21:26 | |
| *** cpallares has joined #openstack-cinder | 21:29 | |
| *** erlon has quit IRC | 21:31 | |
| *** afazekas has joined #openstack-cinder | 21:34 | |
| *** IlyaG has joined #openstack-cinder | 21:35 | |
| *** mriedem is now known as mriedem_away | 21:39 | |
| openstackgerrit | Davanum Srinivas (dims) proposed openstack/cinder: [WIP] Check if we can get rid of oslo copy of middleware https://review.openstack.org/187348 | 21:43 |
| *** daneyon_ has quit IRC | 21:47 | |
| *** e0ne has joined #openstack-cinder | 21:51 | |
| *** akshai has quit IRC | 21:51 | |
| *** annegentle has quit IRC | 21:51 | |
| jgriffith | tsekiyama: ping | 21:53 |
| tsekiyama | jgriffith: pong | 21:53 |
| jgriffith | tsekiyama: hey, had some questions about https://goo.gl/yRnqsb | 21:54 |
| *** afazekas has quit IRC | 21:54 | |
| jgriffith | tsekiyama: I was actually looking at something like this to allow attach in Ironic as well, or "any" user of the client for that matter, so not just Nova | 21:54 |
| *** vilobhmm has quit IRC | 21:54 | |
| jgriffith | tsekiyama: but I had a question how you envision usage here? | 21:54 |
| *** vilobhmm1 has joined #openstack-cinder | 21:54 | |
| jgriffith | tsekiyama: my plan was to actually create a policy entry to expose this to end-user | 21:55 |
| jgriffith | tsekiyama: and return iqn and other relevant attach info | 21:55 |
| tsekiyama | jgriffith: one of use-case of this is glance backend | 21:55 |
| tsekiyama | jgriffith: to upload/download, glance-api will attach the volume using this feature to read/write the image data from volumes | 21:56 |
| jgriffith | tsekiyama: yeah, that's cool | 21:56 |
| jgriffith | tsekiyama: but my point was, do you think we should just be promoting this call and creating a policy for it? | 21:56 |
| jgriffith | tsekiyama: in other words, there seems to be a number of use cases | 21:56 |
| tsekiyama | jgriffith: yeah | 21:57 |
| jgriffith | tsekiyama: that would benefit from being able to say "cinder volume-attach" and getting back the connection info | 21:57 |
| jgriffith | tsekiyama: Ok, cool... so the follow up question :) | 21:57 |
| jgriffith | tsekiyama: shall we merge your patch and then I can do a follow up? Possibly moving that call? | 21:58 |
| jgriffith | tsekiyama: or do you have a better idea? | 21:58 |
| * jgriffith is looking for ideas/brainstorming :) | 21:58 | |
| tsekiyama | jgriffith: Some users may use this for attaching volumes to their own host *not* controlled by nova | 21:58 |
| jgriffith | tsekiyama: yes, exactly | 21:58 |
| jgriffith | tsekiyama: but they can't do that as it is right now | 21:58 |
| tsekiyama | jgriffith: so we need to add 'cinder volume-attach', right? | 21:59 |
| jgriffith | tsekiyama: so what i was thinking... is I could just add a command to shell.py | 22:00 |
| jgriffith | tsekiyama: and figure out what we want to do with policy | 22:00 |
| jgriffith | tsekiyama: but I was curious to get your opinion since you've been looking at this code recently | 22:00 |
| *** emagana has quit IRC | 22:00 | |
| jgriffith | tsekiyama: so right now it's burried in the os-attach call which isn't exposed | 22:00 |
| jgriffith | tsekiyama: so I'm proposing a follow up to your patch where I would expose it (for those that want it) | 22:01 |
| jgriffith | default being turned off in policy.json | 22:01 |
| jgriffith | tsekiyama: make sense? | 22:01 |
| tsekiyama | jgriffith: Yeah, and it might be more complex if we want to control the volume status like nova | 22:01 |
| jgriffith | tsekiyama: yeah, that's where things get a bit "ugly" :) | 22:01 |
| jgriffith | tsekiyama: my initial plan is that it can do two things, and ONLY two things: | 22:01 |
| guitarzan | one note, it feels a bit weird to return the connection info from this call | 22:01 |
| jgriffith | 1. Attach | 22:02 |
| jgriffith | 2. Detach | 22:02 |
| *** garthb has quit IRC | 22:02 | |
| jgriffith | guitarzan: yeah, we kinda talked about that in that meeting in Vancouver | 22:02 |
| guitarzan | this call should get made after you're already attached | 22:02 |
| jgriffith | guitarzan: and I know DuncanT hated it and said "whatever you do don't do that" :) | 22:02 |
| guitarzan | haha | 22:02 |
| guitarzan | this is just a courtesy call to cinder anyway | 22:02 |
| jgriffith | guitarzan: OK, that rmeinded me of soemthing.. as well as an idea I had | 22:03 |
| jgriffith | guitarzan: so brand new call, that gets the connection info and marks the volumes as in-use all in one | 22:03 |
| tsekiyama | jgriffith: +1 | 22:03 |
| mtanino | interesting. | 22:03 |
| *** garthb has joined #openstack-cinder | 22:04 | |
| jgriffith | tsekiyama: guitarzan the good thing about that is that it can be controlled independently from all the other cases via policy | 22:04 |
| guitarzan | I'm not sure if that will help or hurt the sync issues with nova | 22:04 |
| guitarzan | jgriffith: true | 22:04 |
| *** ganso_ has quit IRC | 22:04 | |
| jgriffith | guitarzan: not following... re the sync issues with Nova | 22:04 |
| jgriffith | guitarzan: My use case is "non" nova attach of volumes | 22:04 |
| guitarzan | jgriffith: yeah, no argument at all | 22:04 |
| guitarzan | the totally new api call is a very interesting idea | 22:05 |
| jgriffith | guitarzan: ie making Cinder a stand-alone project (or at least capable of being one) | 22:05 |
| jgriffith | guitarzan: ok... so maybe I'll see what I can put together | 22:05 |
| jgriffith | guitarzan: tsekiyama mtanino I'm just calling it "foo" for now since names are the hardest part ;) | 22:05 |
| *** Guest44 has joined #openstack-cinder | 22:05 | |
| guitarzan | great call there | 22:05 |
| tsekiyama | jgriffith: I'm proposing a code to attach volumes to glance-api. It calls a lot of cinder API's to control the status: https://review.openstack.org/#/c/166414/2/glance_store/_drivers/cinder.py | 22:05 |
| jgriffith | tsekiyama: oh, I hadn't seen that.... | 22:06 |
| tsekiyama | jgriffith: If this could be done in one-stop API, that would be nicer. | 22:06 |
| jgriffith | tsekiyama: so something that I proposed last week might throw a wrench into things | 22:06 |
| jgriffith | tsekiyama: I talked with patrickeast about image-caching in Cinder | 22:06 |
| jgriffith | tsekiyama: and suggested an internal cinder-tenant specifically for this | 22:07 |
| *** openstackgerrit has quit IRC | 22:07 | |
| jgriffith | tsekiyama: but if cinder backed glance might actually be a reality on the horizon it may make all of that work/effort kinda obsolete | 22:07 |
| tsekiyama | jgriffith: Hmm | 22:07 |
| *** openstackgerrit has joined #openstack-cinder | 22:08 | |
| *** Guest44 has quit IRC | 22:08 | |
| jgriffith | tsekiyama: and FYI, I'd recommend the same approach of a designated service tenant | 22:08 |
| *** Guest44 has joined #openstack-cinder | 22:08 | |
| jgriffith | tsekiyama: I'm not saying I wouldn't go foward with what you have, I'm saying what you have may be the "whole" solution that we're trying to work around here | 22:08 |
| tsekiyama | jgriffith: yeah I've added cinder_store_tenant_name etc. to glance-cinder backend so that it can put image volumes into service tenant | 22:08 |
| jgriffith | tsekiyama: oh! I missed that... cool | 22:09 |
| jgriffith | tsekiyama: Ahh... cidner_store_user_name | 22:09 |
| jgriffith | and cinder_storte_tenant_name | 22:09 |
| jgriffith | slick | 22:09 |
| jgriffith | I think :) | 22:09 |
| *** emagana has joined #openstack-cinder | 22:09 | |
| *** annegentle has joined #openstack-cinder | 22:09 | |
| tsekiyama | jgriffith: Using this backend, I believe we can use glance ACL features for sharing volumes-snapshot | 22:09 |
| jgriffith | tsekiyama: yep, that was my assumption :) | 22:10 |
| jgriffith | tsekiyama: otherwise the whole idea unravels quickly :) | 22:10 |
| jgriffith | tsekiyama: so how do you see this working with Nova/ephemeral instances? | 22:10 |
| *** emagana has quit IRC | 22:10 | |
| jgriffith | tsekiyama: suppose I could just read the code :) | 22:11 |
| *** Guest44 has left #openstack-cinder | 22:11 | |
| tsekiyama | jgriffith: Unfortunately not yet started.. Just having an idea to attaching volumes to nova-compute nodes instead of copying the images | 22:11 |
| jgriffith | tsekiyama: got ya | 22:12 |
| jgriffith | tsekiyama: so maybe some sort of an http proxy to a node that attaches and reads the bytes? | 22:12 |
| jgriffith | tsekiyama: of course that raises the other issue... qcow vs raw | 22:12 |
| tsekiyama | jgriffith: yeah, but nova can still use qcow format volumes as a external base images | 22:14 |
| *** krtaylor has quit IRC | 22:14 | |
| tsekiyama | jgriffith: Actually qemu supports reading from qemu format written directly on volumes | 22:15 |
| jgriffith | tsekiyama: but then what's the point? I mean... in that case Cinder and the backends don't gain anything really | 22:15 |
| jgriffith | tsekiyama: because it's filesystem layer | 22:15 |
| jgriffith | tsekiyama: although I guess if that's what people want "ok" but I think they're crazy :) | 22:16 |
| jgriffith | tsekiyama: there are much cheaper ways for storing images (object stores) | 22:16 |
| openstackgerrit | Patrick East proposed openstack/cinder-specs: Generic image volume cache functionality https://review.openstack.org/182520 | 22:16 |
| *** annegentle has quit IRC | 22:16 | |
| tsekiyama | jgriffith: At least we can skip copying the image data to each compute hosts | 22:17 |
| jgriffith | tsekiyama: oh, you're saying attach the volume and read from it? | 22:17 |
| patrickeast | jgriffith: tsekiyama: oo the cinder glance backend | 22:17 |
| tsekiyama | jgriffith: exactly | 22:17 |
| patrickeast | that would be pretty cool | 22:18 |
| *** chlong has quit IRC | 22:18 | |
| tsekiyama | I thought somebody proposed this idea several years ago... | 22:18 |
| tsekiyama | Ah this one: https://blueprints.launchpad.net/nova/+spec/effective-template-base-image-preparing | 22:19 |
| tsekiyama | At that time there are missing pieces like os-brick, multi-attach, host-attach, but now we all have them :) | 22:20 |
| patrickeast | so it attaches the volume and uses it as the base image? | 22:20 |
| tsekiyama | patrickeast: yes | 22:20 |
| *** alexpilotti has joined #openstack-cinder | 22:21 | |
| *** e0ne has quit IRC | 22:22 | |
| *** flip214 has quit IRC | 22:23 | |
| *** netapp has quit IRC | 22:23 | |
| jgriffith | tsekiyama: yes :) | 22:24 |
| jgriffith | tsekiyama: some crazy person :) | 22:24 |
| *** flip214 has joined #openstack-cinder | 22:24 | |
| *** flip214 has quit IRC | 22:24 | |
| *** flip214 has joined #openstack-cinder | 22:24 | |
| tsekiyama | ;) | 22:24 |
| jgriffith | tsekiyama: I don't mean Zhi Yan either... I mean myself :) | 22:24 |
| *** netapp has joined #openstack-cinder | 22:25 | |
| tsekiyama | jgriffith: any way this is one of use-case of host-attach | 22:25 |
| *** rbradbury has joined #openstack-cinder | 22:29 | |
| jgriffith | tsekiyama: sounds cool.. thanks for the explanation | 22:32 |
| tsekiyama | jgriffith: your comments for the patches are very welcome! | 22:33 |
| *** markvoelker_ has joined #openstack-cinder | 22:35 | |
| tsekiyama | FYI, glance-spec for downloading/uploading images from/to cinder volumes is here: https://review.openstack.org/#/c/183363/ | 22:35 |
| *** markvoelker has quit IRC | 22:37 | |
| *** lcurtis has quit IRC | 22:37 | |
| openstackgerrit | Eric Brown proposed openstack/cinder: Add bandit for security static analysis testing https://review.openstack.org/179568 | 22:50 |
| *** openstackgerrit has quit IRC | 22:51 | |
| *** openstackgerrit has joined #openstack-cinder | 22:51 | |
| *** garthb_ has joined #openstack-cinder | 22:53 | |
| *** garthb has quit IRC | 22:53 | |
| *** dims__ has joined #openstack-cinder | 22:53 | |
| vilobhmm1 | is there a cinder health check url that can be used while using VIP | 22:55 |
| vilobhmm1 | for verification | 22:55 |
| *** dims___ has joined #openstack-cinder | 22:56 | |
| *** zul has joined #openstack-cinder | 22:56 | |
| *** dims_ has quit IRC | 22:56 | |
| *** dims__ has quit IRC | 22:58 | |
| *** agentle1 has joined #openstack-cinder | 23:10 | |
| *** IlyaG has quit IRC | 23:11 | |
| *** krtaylor has joined #openstack-cinder | 23:12 | |
| openstackgerrit | Earle F. Philhower, III proposed openstack/os-brick: Add HGST volume type Connector https://review.openstack.org/186588 | 23:20 |
| *** rushil has quit IRC | 23:21 | |
| *** nikeshm has quit IRC | 23:22 | |
| *** nikeshm has joined #openstack-cinder | 23:24 | |
| openstackgerrit | Earle F. Philhower, III proposed openstack/cinder: Implement Cinder Volume driver for HGST Solutions https://review.openstack.org/186580 | 23:25 |
| *** agentle1 has quit IRC | 23:30 | |
| *** jdurgin has quit IRC | 23:31 | |
| openstackgerrit | Earle F. Philhower, III proposed openstack/os-brick: Add HGST volume type Connector https://review.openstack.org/186588 | 23:31 |
| openstackgerrit | Davanum Srinivas (dims) proposed openstack/cinder: Get rid of oslo copy of middleware https://review.openstack.org/187348 | 23:41 |
| openstackgerrit | Davanum Srinivas (dims) proposed openstack/cinder: Get rid of oslo copy of middleware https://review.openstack.org/187348 | 23:41 |
| *** chlong has joined #openstack-cinder | 23:41 | |
| openstackgerrit | Davanum Srinivas (dims) proposed openstack/cinder: Get rid of oslo-incubator copy of middleware https://review.openstack.org/187348 | 23:43 |
| *** agentle1 has joined #openstack-cinder | 23:49 | |
| *** agentle1 has quit IRC | 23:49 | |
| *** hemna is now known as hemnafk | 23:54 | |
| *** rmesta has quit IRC | 23:55 | |
| *** yamada-h has joined #openstack-cinder | 23:58 | |
| openstackgerrit | Tomoki Sekiyama proposed openstack/python-cinderclient: Add volume multi attach support https://review.openstack.org/85856 | 23:59 |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!