| *** marlinc is now known as Guest7471 | 00:18 | |
| *** osmanlicilegi is now known as Guest7485 | 00:18 | |
| opendevreview | 王昊 proposed openstack/cinder master: Support multiple filters together for volume time comparison filter https://review.opendev.org/c/openstack/cinder/+/740146 | 01:07 |
|---|---|---|
| opendevreview | Oleg proposed openstack/cinder master: PowerStore Driver - New format of connection properties for NVMeOF connector https://review.opendev.org/c/openstack/cinder/+/852261 | 01:16 |
| opendevreview | 王昊 proposed openstack/cinder master: Support to query snapshot filter by updated_at/created_at https://review.opendev.org/c/openstack/cinder/+/733062 | 01:26 |
| *** tobias-urdin5 is now known as tobias-urdin | 06:03 | |
| opendevreview | Alexander Malashenko proposed openstack/cinder master: PowerFlex driver - fix the display of the incorrect volume size on volume or snapshot creation. https://review.opendev.org/c/openstack/cinder/+/836973 | 07:50 |
| opendevreview | Alexander Deiter proposed openstack/cinder master: Infinidat: add support for revert to snapshot operation https://review.opendev.org/c/openstack/cinder/+/851996 | 09:20 |
| opendevreview | Raghavendra Tilay proposed openstack/cinder master: HPE 3PAR: Fix umanaged volumes & snapshots missing https://review.opendev.org/c/openstack/cinder/+/850836 | 10:36 |
| opendevreview | Alexander Deiter proposed openstack/cinder master: Infinidat: add support for manage/unmanage volume and snapshot https://review.opendev.org/c/openstack/cinder/+/852009 | 11:35 |
| *** dviroel_ is now known as dviroel | 11:38 | |
| *** elodille1 is now known as elodilles | 12:11 | |
| opendevreview | Nahim Alves de Souza proposed openstack/cinder master: NetApp ONTAP: Implemented volume replication functions on REST client https://review.opendev.org/c/openstack/cinder/+/852209 | 12:35 |
| opendevreview | Nahim Alves de Souza proposed openstack/cinder master: NetApp ONTAP: Implemented volume migration functions on REST client https://review.opendev.org/c/openstack/cinder/+/852210 | 12:35 |
| opendevreview | Nahim Alves de Souza proposed openstack/cinder master: NetApp ONTAP: Implemented revert to snapshot functions on REST client https://review.opendev.org/c/openstack/cinder/+/852211 | 12:35 |
| olegnest | rosmaita hemna: hello! Could you please reset your votes on issue https://review.opendev.org/c/openstack/cinder/+/797970? We have already got +2 from you, but when we changed the commit message all the votes were lost =( Thanks! | 12:48 |
| rosmaita | olegnest: ack | 12:48 |
| opendevreview | Gorka Eguileor proposed openstack/cinder master: DEMC: Add support for trim/discard https://review.opendev.org/c/openstack/cinder/+/849890 | 13:01 |
| geguileo | eharney: sorry, I had to update that patch ^ to include the new nvmeof driver :-( | 13:19 |
| rosmaita | olegnest: left some questions for you | 13:42 |
| opendevreview | Harsh Ailani proposed openstack/cinder master: [SVf] As part of Flashcopy 2.0 adding support for volumegroup https://review.opendev.org/c/openstack/cinder/+/852364 | 14:09 |
| *** dviroel is now known as dviroel|lunch | 15:29 | |
| opendevreview | Oleg proposed openstack/cinder master: PowerMax Driver - Fix for force flag https://review.opendev.org/c/openstack/cinder/+/797970 | 15:51 |
| opendevreview | Brian Rosmaita proposed openstack/cinder stable/yoga: DNM: checking flake8-logging-format==0.7.4 damage https://review.opendev.org/c/openstack/cinder/+/852454 | 16:19 |
| opendevreview | Brian Rosmaita proposed openstack/cinder stable/yoga: DNM: checking flake8-logging-format==0.7.4 damage https://review.opendev.org/c/openstack/cinder/+/852454 | 16:23 |
| opendevreview | Brian Rosmaita proposed openstack/cinder stable/xena: DNM: checking flake8-logging-format==0.7.4 damage https://review.opendev.org/c/openstack/cinder/+/852456 | 16:26 |
| opendevreview | Brian Rosmaita proposed openstack/cinder stable/wallaby: DNM: checking flake8-logging-format==0.7.4 damage https://review.opendev.org/c/openstack/cinder/+/852457 | 16:30 |
| *** dviroel|lunch is now known as dviroel | 16:38 | |
| opendevreview | Brian Rosmaita proposed openstack/cinder stable/xena: Pin flake8-logging-format to 0.6.0 https://review.opendev.org/c/openstack/cinder/+/852458 | 16:39 |
| rosmaita | yoga seems OK, master requires some fixes, this one should unstick yoga ^^ | 16:39 |
| rosmaita | i mean xena | 16:40 |
| opendevreview | Brian Rosmaita proposed openstack/cinder master: Address G004 in flake8-logging-format 0.7.4 https://review.opendev.org/c/openstack/cinder/+/852461 | 17:12 |
| rosmaita | should fix pep-8 job in master ^^ | 17:12 |
| geguileo | whoami-rajat I broke pep8 job with https://review.opendev.org/c/openstack/cinder/+/847216 | 17:21 |
| geguileo | whoami-rajat: do you know if somebody has provided a patch to fix it? | 17:21 |
| whoami-rajat | geguileo, isn't the patch by brian fixing it? | 17:23 |
| geguileo | whoami-rajat: I'm just blind | 17:24 |
| whoami-rajat | https://review.opendev.org/c/openstack/cinder/+/852461 | 17:24 |
| jungleboyj | :-) | 17:24 |
| geguileo | yeah, that fixes it | 17:24 |
| whoami-rajat | great | 17:25 |
| whoami-rajat | we're good as long as we've a fix! | 17:25 |
| geguileo | I'll also update the backport to yoga of the patch that broke that | 17:25 |
| geguileo | rosmaita: thanks for fixing that :-) | 17:31 |
| opendevreview | Gorka Eguileor proposed openstack/cinder stable/yoga: Log when waiting to acquire coordinator lock https://review.opendev.org/c/openstack/cinder/+/848821 | 17:42 |
| admin1 | hi all .. i added a nfs nbackend in cinder and I get Active-Active configuration is not currently supported by driver cinder.volume.drivers.nfs.NfsDriver . what does this mean ? | 18:01 |
| admin1 | how/whats sets active-active in cinder ? | 18:02 |
| geguileo | admin1: this active-active refers to being able to run 2 or more cinder-volume services with the same configuration and managing volumes for the same cinder backend | 18:13 |
| geguileo | admin1: drivers need to make sure that they support active-active because the feature is turned on in code | 18:13 |
| geguileo | admin1: once the driver supports the feature then it's a matter of deploying and configuring multiple cinder-volume services | 18:14 |
| geguileo | a DLM needs to be deployed (Distributed Lock Manager) and configured in cinder as the coordinator | 18:14 |
| geguileo | and then in cinder.conf the cluster configuration option needs to be set to the same value in these cinder-volume services | 18:15 |
| admin1 | geguileo, i mean why is it trying to look for active-active in nfs ? | 18:18 |
| admin1 | https://gist.githubusercontent.com/a1git/b0ddc0b718a3fb89d28b037d89f1583f/raw/34e0992495db81e7b591570946f4296981867a4e/gistfile1.txt | 18:18 |
| admin1 | i had 2 backends .. then added nfs .. and then it rejected nfs | 18:18 |
| geguileo | admin1: this is NOT about NFS being active-active, this is about cinder-volume | 18:18 |
| admin1 | so i need to have more than 1 cinder-volume container to add nfs type as a cinder backend ? | 18:20 |
| admin1 | i mean i have a cluster with only nfs cinder backend running fine .. but this is one where i am putting both ceph and nfs as backend .. they cannot co-exist ? | 18:22 |
| admin1 | sorry but i don't seem to get the core of the issue | 18:23 |
| geguileo | admin1: you can have cinder run rbd and nfs as different backends, that is fine | 18:25 |
| geguileo | admin1: and you can do it either on the same service (setting both in enabled_backends) | 18:25 |
| geguileo | admin1: or in 2 different services, one for RBD and another for NFS | 18:26 |
| admin1 | i have 1 controller and 1 nfs running fine ...this one is 1 controller + 2 rbd backends .. which was running fine .. then just now i had to add one more backend, NFS and then it rejected the nfs backend with that error . | 18:26 |
| geguileo | admin1: but you will only have 1 cinder-volume process for ALL the requests that come to RBD and another single process for ALL the requests to NFS | 18:26 |
| geguileo | admin1: now, if you want to horizontally scale the cinder-volume service you will need an active-active deployment | 18:26 |
| admin1 | i undertsand this, but its not a warning .. it rejected the nfs backend with that error | 18:27 |
| geguileo | where you can have 2 or more RBD Cinder backends managing the same volumes in the same pool | 18:27 |
| geguileo | admin1: because NFS doesn't support it | 18:27 |
| geguileo | by NFS I mean the Cinder NFS driver | 18:27 |
| geguileo | so of course it rejects it, because developers have not made it safe to run in an active-active deployment | 18:27 |
| admin1 | wait, so whats the solution ? i need to add 1 more controller (cinder-volume service) and it will run ? | 18:28 |
| geguileo | if you try with the RBD driver you will see that that one doesn't fail | 18:28 |
| geguileo | whaaaaaaaaaaaaaaaat? | 18:28 |
| admin1 | ok .. let me try from the begining .. what is the solution ? and that might help me undersand the issue | 18:29 |
| geguileo | no, the problem is that you are telling cinder that it's going to be running in active-active | 18:29 |
| geguileo | fix your cinder.conf file | 18:29 |
| admin1 | grep -ri active /etc/cinder/cinder.conf .. no hits | 18:30 |
| geguileo | the config option is "cluster" | 18:30 |
| geguileo | because you define a name of the cluster with that | 18:30 |
| geguileo | RBD supports it, but not NFS | 18:30 |
| admin1 | cluster = ceph | 18:30 |
| geguileo | where is that defined? which section? | 18:30 |
| admin1 | one moment please | 18:31 |
| admin1 | https://gist.github.com/a1git/fbf329c3027ea51278ef3c3c599d0dfa line 12 | 18:34 |
| admin1 | openstack-ansible generated conf .. | 18:35 |
| geguileo | admin1: that's the problem, the config is telling cinder that all backends belong to a cluster | 18:35 |
| geguileo | admin1: and cinder complains for NFS because that driver cannot belong to a cluster | 18:35 |
| geguileo | on how many hosts is the cinder-volume service running? | 18:36 |
| geguileo | admin1: I have to drop, it's 8:40 PM over here | 18:39 |
| geguileo | the problem is that, NFS cannot run with cluster variable set | 18:39 |
| geguileo | and if you run cinder-volume without cluster set, in active-passive where the switch is controller by something like pacemaker, then you need to set the backend_host for each of the enabled backends or host in the DEFAULT section | 18:41 |
| geguileo | otherwise when the service switches hosts the volumes will no longer be accessible | 18:41 |
| admin1 | i have just 1 controller in this one, and 3 in some others .. there is no pacemaker, but standard haproxy setup | 18:49 |
| admin1 | is it possible to say nfs does not belong to cluster ? | 18:50 |
| admin1 | or bypass cluster for just 1 backend ? | 18:50 |
| admin1 | geguileo, thanks .. i undersand it now | 18:56 |
| admin1 | and also know what variable to use to remove it and test it | 18:56 |
| admin1 | many thanks | 18:57 |
| *** tosky_ is now known as tosky | 19:25 | |
| *** dviroel is now known as dviroel|out | 21:22 | |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!