| *** keper7 has quit IRC | 00:22 | |
| *** brinzhang has quit IRC | 00:50 | |
| *** brinzhang has joined #openstack-cinder | 00:51 | |
| *** brinzhang_ has joined #openstack-cinder | 00:59 | |
| *** brinzhang has quit IRC | 01:03 | |
| *** ociuhandu has joined #openstack-cinder | 01:04 | |
| *** ociuhandu has quit IRC | 01:09 | |
| *** zhanglong has joined #openstack-cinder | 01:12 | |
| *** Liang__ has joined #openstack-cinder | 01:17 | |
| *** ociuhandu has joined #openstack-cinder | 02:34 | |
| *** ociuhandu has quit IRC | 02:39 | |
| *** psachin has joined #openstack-cinder | 03:17 | |
| *** lvbin01 has joined #openstack-cinder | 03:21 | |
| *** spatel has joined #openstack-cinder | 03:24 | |
| *** spatel has quit IRC | 03:25 | |
| *** ykarel|away is now known as ykarel | 04:20 | |
| *** brinzhang has joined #openstack-cinder | 04:36 | |
| *** brinzhang_ has quit IRC | 04:39 | |
| openstackgerrit | Raghavendra Tilay proposed openstack/cinder master: HPE 3PAR: Support duplicated FQDN in network https://review.opendev.org/713803 | 05:05 |
|---|---|---|
| *** andrebeltrami has quit IRC | 05:16 | |
| *** lvbin01 has quit IRC | 05:25 | |
| *** lvbin01 has joined #openstack-cinder | 05:26 | |
| *** brinzhang_ has joined #openstack-cinder | 05:33 | |
| openstackgerrit | Merged openstack/cinderlib master: Cleanup py27 support https://review.opendev.org/717487 | 05:36 |
| *** brinzhang has quit IRC | 05:36 | |
| *** udesale has joined #openstack-cinder | 05:41 | |
| *** ircuser-1 has quit IRC | 06:07 | |
| *** udesale has quit IRC | 06:09 | |
| *** udesale has joined #openstack-cinder | 06:13 | |
| *** kevinz has joined #openstack-cinder | 06:28 | |
| kevinz | Morning Cinder! | 06:29 |
| kevinz | I wonder do we support change volume backend after the volume backend is dead? | 06:29 |
| *** dpawlik has joined #openstack-cinder | 06:30 | |
| kevinz | we have cinder+ceph, and we removed the cinder volume service+Ceph osd in one host. but some volumes still have "os-vol-host-attr:host" to that cinder service. How could we migrate the volume to another host? | 06:31 |
| openstackgerrit | Xuchu Jiang proposed openstack/cinder master: Add an option to specify iSCSI IP version https://review.opendev.org/698630 | 06:40 |
| *** lvbin02 has joined #openstack-cinder | 06:46 | |
| *** lvbin01 has quit IRC | 06:46 | |
| *** lvbin02 is now known as lvbin01 | 06:46 | |
| *** throne82 has quit IRC | 06:58 | |
| *** zhanglong has quit IRC | 06:58 | |
| *** tosky has joined #openstack-cinder | 07:24 | |
| *** xuanyandong has joined #openstack-cinder | 07:31 | |
| *** ociuhandu has joined #openstack-cinder | 07:45 | |
| xuanyandong | whoami-rajat: Hello, thanks for your review, In order not to lose +2, Can i submit a new commit to fix it? https://review.opendev.org/#/c/705876/ | 07:51 |
| *** lvbin02 has joined #openstack-cinder | 07:52 | |
| *** muliby has joined #openstack-cinder | 07:52 | |
| whoami-rajat | xuanyandong, i was thinking of same but just because of commit message i didn't consider followup. i think now the releasenote is more important so if you submit a quick followup then i can merge it. | 07:54 |
| *** lvbin01 has quit IRC | 07:54 | |
| *** lvbin02 is now known as lvbin01 | 07:54 | |
| *** brinzhang has joined #openstack-cinder | 07:56 | |
| muliby | hello everyone, I'm writing a Cinder driver for the LightOS storage defined disaggregated storage soluion (https://www.lightbitslabs.com/products/lightos/). I think I found a bug with the thin provisioning support in Cinder. If I expose 20GB of storage *thinly-provisioned* and attempt to create volumes larger than that (say 70GB) rapidly, at some | 07:58 |
| muliby | point volume creation will be forbidden by the capacity filter because we do self.free_capacity_gb -= volume size in host_manager:consume_from_volume(), leading to negative free_capacity_gb | 07:58 |
| *** xuanyandong has quit IRC | 07:58 | |
| *** brinzhang_ has quit IRC | 07:59 | |
| muliby | The reason it takes multiple creation attempts is that *if* the host_manager gets updated stats from the driver, self.free_capacity_gb will be reset to the driver reported value | 07:59 |
| muliby | But if two creation attempts happen without an updated stats call inbetween, the second creation will fail, even though the system is thinly provisioned and no actual storage was used, so free_capacity_gb is still 20GB | 08:00 |
| *** xuanyd has joined #openstack-cinder | 08:01 | |
| muliby | I worked around it for now by setting free_capacity_gb to 'infinite', but it would be nicer to fix this properly, either by forcing updated stats on every volume creation (can I do that from the driver?) or by fixing the free_capacity_gb calculation | 08:01 |
| xuanyd | whoami-rajat: Ok, i will update it right now. | 08:01 |
| whoami-rajat | xuanyd, okay. you don't have to do it right away, i just wanted a confirmation as we will cut stable/ussuri soon and i don't want the wrong releasenote to publish in it. | 08:03 |
| whoami-rajat | muliby, are you using master or any stable branch? | 08:05 |
| *** brinzhang_ has joined #openstack-cinder | 08:05 | |
| muliby | whoami-rajat I have only tested the driver on queens, rocky, and stein so far (customer requests...) but confirmed that this code is identical in Cinder master (https://github.com/openstack/cinder/blob/e5fe9b881cc06d94a077efd0fa4c73abb1781766/cinder/scheduler/host_manager.py) | 08:06 |
| whoami-rajat | muliby, ok. the volume backend periodically reports the driver stats so i don't think its a good idea to change that | 08:08 |
| *** brinzhang has quit IRC | 08:09 | |
| whoami-rajat | muliby, modifying the calculations seems reasonable, you can open a launchpad bug and also propose a solution upstream if you have one | 08:09 |
| xuanyd | whoami-rajat: Sorry, I do n’t quite understand what you mean, what should I do next | 08:10 |
| muliby | whoami-rajat I thought of modifying the calculation to do something like this: self.free_capacity_gb = (volume size) / (max overprovision factor) | 08:10 |
| muliby | (for thinly provisioned systems, of course) | 08:11 |
| muliby | that should have been -= not = of course, i.e., temporarily reduce the free_capacity_gb not by the full volume size, since it's thinly provisioned, but rather by the volume size divided by the max overprovisioning factor, i.e., the amount of space it is likely to consume shortly after having been created | 08:12 |
| whoami-rajat | xuanyd, i've merged your patch, you just need to propose a followup now. | 08:13 |
| xuanyd | whoami-rajat: ok, thanks very much, i will submit a new path latter. | 08:14 |
| *** e0ne has joined #openstack-cinder | 08:16 | |
| whoami-rajat | xuanyd, great. just don't forget about it :) | 08:17 |
| *** tkajinam has quit IRC | 08:17 | |
| *** xinliang has joined #openstack-cinder | 08:18 | |
| kevinz | morning Cinder! I wonder do we support change volume backend after the volume backend is dead? we have cinder+ceph, and we removed the cinder volume service+Ceph osd in one host. but some volumes still have "os-vol-host-attr:host" to that cinder service. How could we migrate the volume to another host? | 08:19 |
| *** zhanglong has joined #openstack-cinder | 08:25 | |
| whoami-rajat | muliby, so i'm not totally familiar with the code but i think the free_capacity_gb term is calculated from max_over_subscription_ratio parameter defined in cinder.conf and not the actual free_capacity of the backend | 08:27 |
| muliby | whoami-rajat that's the way I would have expected it to be, but it seems incomplete -- here we reduce the entire volume size, which makes sense for a thick provisioned volume, but not for a thinly provisioned volume | 08:28 |
| whoami-rajat | muliby, ack. so you're pointing at if the new volume size is 10 GB and 2 GB is used then we should only subtract 2 from the free capacity | 08:30 |
| muliby | well, yes, but the problem here is that we don't know how much is actually going to be used soon until the next stats update, so I'm suggesting that if the volume size is 10GB and the max ovp factor is 20, we reduce only 0.5GB | 08:32 |
| muliby | which is a more accurate approximation of how much will be used shortly after creation | 08:33 |
| muliby | I'll open a launchpad bug and we'll take it from there, thanks for the help! | 08:33 |
| whoami-rajat | muliby, yeah, that seems like a reasonable issue | 08:34 |
| whoami-rajat | muliby, thanks for finding and reporting it. | 08:35 |
| whoami-rajat | kevinz, i think the correct way to migrate is, setup the new backend, retype (migrate) all existing volumes to the new backend and then remove the config of previous backend. | 08:37 |
| whoami-rajat | kevinz, i'm not really sure your case is supported. | 08:37 |
| Xuchu | whoami-rajat: for the thick provision, I think the capacity is dedicated for this volume, so I think cinder should subtract 10GB from free space. | 08:42 |
| Xuchu | for thin provision, cinder already has over_provision ratio, I think subtract 10GB is ok too | 08:44 |
| Xuchu | over-provision capacity is a virtual capacity, thin capacity is a virtual capacity | 08:44 |
| whoami-rajat | Xuchu, yep, we are discussing the scenario of only thin provisioning here | 08:45 |
| whoami-rajat | Xuchu, as it makes sense to use MOSR but theoretically the allocated capacity should only account for the filled up space in the volume | 08:46 |
| *** ykarel is now known as ykarel|lunch | 08:48 | |
| whoami-rajat | Xuchu, but we really can't account for that so whichever works, as i think now its the same thing to use free capacity * MOSR or free_cap = vol/MOSR | 08:48 |
| whoami-rajat | muliby, ^^ | 08:48 |
| *** ociuhandu has quit IRC | 08:50 | |
| *** ociuhandu has joined #openstack-cinder | 08:57 | |
| *** zhanglong has quit IRC | 09:03 | |
| *** zhanglong has joined #openstack-cinder | 09:06 | |
| *** martinkennelly has joined #openstack-cinder | 09:08 | |
| muliby | whoami-rajat in the capacity filter, we indeed check free capacity * MOSR vs. volume size, the problem is that once free_capacity goes negative, free capacity * MOSR is negative as well. And it goes negative because we do free_capacity -= volume size after accepting it. So doing free_capacity -= volume size / MOSR seems like a reasonable solution | 09:08 |
| *** lvbin02 has joined #openstack-cinder | 09:19 | |
| *** lvbin01 has quit IRC | 09:21 | |
| *** lvbin02 is now known as lvbin01 | 09:21 | |
| *** ajitha has joined #openstack-cinder | 09:23 | |
| *** xinliang has quit IRC | 09:29 | |
| *** ajitha has quit IRC | 09:30 | |
| *** zhanglong has quit IRC | 09:35 | |
| *** brinzhang has joined #openstack-cinder | 09:36 | |
| *** zhanglong has joined #openstack-cinder | 09:38 | |
| *** brinzhang_ has quit IRC | 09:39 | |
| *** ykarel|lunch is now known as ykarel | 09:41 | |
| *** xuanyd has quit IRC | 09:42 | |
| kevinz | whoami-rajat: Hi, thanks for reply. Yeah I think it is reasonable, but we have already removed some backend volume services already, is there an easy way to recover or migrate the volumes? | 09:46 |
| openstackgerrit | Xuchu Jiang proposed openstack/cinder master: Add an option to specify iSCSI IP version https://review.opendev.org/698630 | 09:46 |
| *** n-saito has quit IRC | 09:46 | |
| *** kaisers_ has joined #openstack-cinder | 09:47 | |
| whoami-rajat | muliby, as i said i'm not very familiar with that code but if that's the case then we should do free_capacity = free_capacity * MOSR and then free_capacity -= vol_size to minimize code change | 09:49 |
| whoami-rajat | kevinz, i don't have that much production experience to help you, maybe others could | 09:50 |
| kevinz | whoami-rajat: thx :-) | 09:50 |
| *** ykarel is now known as ykarel|meeting | 10:02 | |
| *** zhanglong has quit IRC | 10:08 | |
| *** Liang__ has quit IRC | 10:09 | |
| openstackgerrit | Merged openstack/cinder master: Add features for add backup id to volume metadata https://review.opendev.org/705876 | 10:27 |
| openstackgerrit | Rajat Dhasmana proposed openstack/cinder-tempest-plugin master: Add LVM+tgt tempest job https://review.opendev.org/537658 | 10:34 |
| *** ykarel|meeting is now known as ykarel | 10:37 | |
| *** vishalmanchanda has joined #openstack-cinder | 10:40 | |
| openstackgerrit | Rajat Dhasmana proposed openstack/cinder-tempest-plugin master: Add LVM+tgt tempest job https://review.opendev.org/537658 | 10:41 |
| tosky | whoami-rajat: another question (apart from the several questions in the reviews): is https://review.opendev.org/#/c/702495/ a more general version of https://review.opendev.org/#/c/620577/ ? | 11:17 |
| whoami-rajat | tosky, If i understand your question clearly, they both are different tests but they both require common functions from tempest scenario manager so if it gets merged once we just require to write new tests only. | 11:22 |
| tosky | right | 11:25 |
| *** lvbin02 has joined #openstack-cinder | 11:28 | |
| *** andrebeltrami has joined #openstack-cinder | 11:28 | |
| *** lvbin01 has quit IRC | 11:30 | |
| *** lvbin02 is now known as lvbin01 | 11:30 | |
| *** rosmaita has quit IRC | 11:40 | |
| *** ociuhandu has quit IRC | 11:41 | |
| *** ociuhandu has joined #openstack-cinder | 11:42 | |
| *** rosmaita has joined #openstack-cinder | 11:42 | |
| *** muliby has quit IRC | 11:42 | |
| *** ociuhandu has quit IRC | 11:47 | |
| *** tkajinam has joined #openstack-cinder | 12:08 | |
| *** yaawang has quit IRC | 12:08 | |
| hemna_ | morning | 12:16 |
| *** ociuhandu has joined #openstack-cinder | 12:45 | |
| *** lvbin02 has joined #openstack-cinder | 12:48 | |
| *** enriquetaso has joined #openstack-cinder | 12:48 | |
| *** lvbin01 has quit IRC | 12:50 | |
| *** lvbin02 is now known as lvbin01 | 12:50 | |
| *** enriquetaso has quit IRC | 12:51 | |
| *** ociuhandu has quit IRC | 12:52 | |
| *** ociuhandu has joined #openstack-cinder | 12:59 | |
| *** ociuhandu has quit IRC | 13:00 | |
| *** ociuhandu has joined #openstack-cinder | 13:01 | |
| *** udesale_ has joined #openstack-cinder | 13:03 | |
| *** enriquetaso has joined #openstack-cinder | 13:04 | |
| *** udesale has quit IRC | 13:05 | |
| *** Luzi has joined #openstack-cinder | 13:09 | |
| *** ociuhandu has quit IRC | 13:32 | |
| *** ociuhandu has joined #openstack-cinder | 13:33 | |
| *** yaawang has joined #openstack-cinder | 13:47 | |
| openstackgerrit | Sofia Enriquez proposed openstack/cinder-tempest-plugin master: Create new Ceph job with c-bak service enabled https://review.opendev.org/717379 | 13:50 |
| *** dave-mccowan has joined #openstack-cinder | 13:59 | |
| *** _kukacz has joined #openstack-cinder | 14:02 | |
| *** tkajinam has quit IRC | 14:03 | |
| *** tkajinam has joined #openstack-cinder | 14:03 | |
| openstackgerrit | Brian Rosmaita proposed openstack/cinder master: RBD: add support for revert-to-snapshot https://review.opendev.org/710334 | 14:07 |
| *** Luzi has quit IRC | 14:07 | |
| *** ajitha has joined #openstack-cinder | 14:08 | |
| *** dave-mccowan has quit IRC | 14:10 | |
| *** yaawang_ has joined #openstack-cinder | 14:11 | |
| *** yaawang has quit IRC | 14:14 | |
| openstackgerrit | Xuan Yandong proposed openstack/cinder master: Update test and releasenote for bp: add-volume-backup-id https://review.opendev.org/718121 | 14:18 |
| *** martinkennelly has quit IRC | 14:18 | |
| openstackgerrit | Xuan Yandong proposed openstack/cinder master: Update test and releasenote for bp: add-volume-backup-id https://review.opendev.org/718121 | 14:32 |
| *** xuanyd has joined #openstack-cinder | 14:33 | |
| *** tkajinam has quit IRC | 14:42 | |
| xuanyd | whoami-rajat: hello, the patch has commited, please review it, thanks very much,https://review.opendev.org/#/c/718121/ | 14:48 |
| whoami-rajat | xuanyd, great, i'm in a meeting right now, will review after that. | 14:49 |
| whoami-rajat | also zuul hasn't run yet | 14:50 |
| xuanyd | whoami-rajat: ok, thanks | 14:51 |
| *** xuanyd has quit IRC | 14:52 | |
| *** yaawang has joined #openstack-cinder | 14:53 | |
| *** yaawang__ has joined #openstack-cinder | 14:54 | |
| *** yaawang_ has quit IRC | 14:56 | |
| *** yaawang has quit IRC | 14:58 | |
| *** martinkennelly has joined #openstack-cinder | 15:03 | |
| *** ircuser-1 has joined #openstack-cinder | 15:11 | |
| *** rcernin has quit IRC | 15:13 | |
| rosmaita | any reviewer with an interest in cinderclient, please look at https://review.opendev.org/#/c/714572/ | 15:39 |
| rosmaita | e0ne eharney geguileo hemna_ jungleboyj smcginnis whoami-rajat ^^ | 15:39 |
| jungleboyj | rosmaita: Done. | 15:44 |
| smcginnis | If have (had) a few concerns on that one. | 15:45 |
| rosmaita | oops | 15:46 |
| rosmaita | smcginnis: not too late to stop it | 15:47 |
| jungleboyj | smcginnis: I just stopped it. | 15:47 |
| smcginnis | Eh, I'm looking through the results of codesearch.o.o and not seeing anyone using it that way. | 15:48 |
| smcginnis | My main concern was OSC or one of the packaging type projects like kolla or ansible. | 15:48 |
| smcginnis | Looks like maybe we're safe as far as the OpenStack hosted code goes. | 15:48 |
| smcginnis | jungleboyj: Go ahead with the +W, I think we're actually fine. | 15:49 |
| smcginnis | Thanks | 15:49 |
| jungleboyj | smcginnis: Okdie dokie. Thanks for checking. | 15:50 |
| *** dpawlik has quit IRC | 16:00 | |
| *** _kukacz has quit IRC | 16:01 | |
| *** udesale_ has quit IRC | 16:08 | |
| *** ykarel is now known as ykarel|away | 16:10 | |
| *** ociuhandu has quit IRC | 16:18 | |
| *** ociuhandu has joined #openstack-cinder | 16:19 | |
| *** lpetrut has joined #openstack-cinder | 16:27 | |
| *** brinzhang has quit IRC | 16:30 | |
| *** brinzhang has joined #openstack-cinder | 16:31 | |
| *** ociuhandu has quit IRC | 16:42 | |
| *** ociuhandu has joined #openstack-cinder | 16:43 | |
| *** ociuhandu has quit IRC | 16:48 | |
| *** psachin has quit IRC | 16:55 | |
| *** lpetrut has quit IRC | 16:56 | |
| openstackgerrit | Merged openstack/python-cinderclient master: Replace bypass_url with os_endpoint https://review.opendev.org/714572 | 17:18 |
| openstackgerrit | Merged openstack/python-cinderclient master: Remove autogen warning https://review.opendev.org/714577 | 17:19 |
| *** martinkennelly has quit IRC | 17:30 | |
| openstackgerrit | Sofia Enriquez proposed openstack/cinder-tempest-plugin master: WIP: Retype test added https://review.opendev.org/715566 | 17:47 |
| *** vishalmanchanda has quit IRC | 17:47 | |
| *** brinzhang_ has joined #openstack-cinder | 17:51 | |
| *** brinzhang has quit IRC | 17:53 | |
| *** kaisers_ has quit IRC | 17:59 | |
| *** e0ne has quit IRC | 18:04 | |
| *** e0ne has joined #openstack-cinder | 18:37 | |
| *** enriquetaso has quit IRC | 18:40 | |
| *** raunak12 has joined #openstack-cinder | 18:44 | |
| *** e0ne has quit IRC | 18:45 | |
| *** Lucas_Gray has joined #openstack-cinder | 19:14 | |
| *** Lucas_Gray has quit IRC | 19:22 | |
| *** Lucas_Gray has joined #openstack-cinder | 19:22 | |
| *** lvbin02 has joined #openstack-cinder | 19:49 | |
| *** lvbin01 has quit IRC | 19:49 | |
| *** lvbin02 is now known as lvbin01 | 19:49 | |
| *** ociuhandu has joined #openstack-cinder | 19:54 | |
| *** ociuhandu has quit IRC | 20:00 | |
| *** Lucas_Gray has quit IRC | 20:15 | |
| *** Lucas_Gray has joined #openstack-cinder | 20:21 | |
| openstackgerrit | Eric Harney proposed openstack/cinder master: validator: Replace InvalidInput exception w/ InvalidName https://review.opendev.org/708713 | 20:22 |
| *** Lucas_Gray has quit IRC | 20:42 | |
| *** Lucas_Gray has joined #openstack-cinder | 20:44 | |
| *** ajitha has quit IRC | 21:03 | |
| *** enriquetaso has joined #openstack-cinder | 21:04 | |
| *** ociuhandu has joined #openstack-cinder | 21:34 | |
| *** ociuhandu has quit IRC | 21:39 | |
| openstackgerrit | Brian Rosmaita proposed openstack/python-cinderclient master: Add release note for Ussuri cinderclient release. https://review.opendev.org/718234 | 21:39 |
| rosmaita | smcginnis: lmk if that addresses your concern about https://review.opendev.org/#/c/714572/ | 21:40 |
| rosmaita | jungleboyj: would also appreciate your feedback on https://review.opendev.org/718234 | 21:41 |
| *** enriquetaso has quit IRC | 21:51 | |
| *** rcernin has joined #openstack-cinder | 22:11 | |
| *** eharney has quit IRC | 22:20 | |
| *** Lucas_Gray has quit IRC | 22:36 | |
| *** tkajinam has joined #openstack-cinder | 22:56 | |
| *** yaawang__ has quit IRC | 23:16 | |
| *** yaawang__ has joined #openstack-cinder | 23:17 | |
| *** tosky has quit IRC | 23:26 | |
| *** raunak12 has quit IRC | 23:40 | |
| *** n-saito has joined #openstack-cinder | 23:48 | |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!