Wednesday, 2019-03-06

*** henriqueof has quit IRC00:01
*** henriqueof has joined #openstack-cinder00:02
*** sapd1 has quit IRC00:16
*** hoonetorg has quit IRC00:24
*** hoonetorg has joined #openstack-cinder00:27
*** brinzhang has joined #openstack-cinder00:31
*** dviroel_ has quit IRC00:37
*** markvoelker has joined #openstack-cinder00:49
*** itlinux has joined #openstack-cinder00:57
*** mriedem has quit IRC01:00
*** whoami-rajat has joined #openstack-cinder01:09
*** mmethot has joined #openstack-cinder01:15
*** markvoelker has quit IRC01:22
*** sdake has quit IRC01:29
*** itlinux_ has joined #openstack-cinder01:38
*** itlinux_ has quit IRC01:40
*** itlinux has quit IRC01:41
*** sdake has joined #openstack-cinder01:45
*** gnufied has quit IRC01:53
*** sdake has quit IRC02:05
*** markvoelker has joined #openstack-cinder02:19
*** sdake has joined #openstack-cinder02:28
*** sdake has quit IRC02:42
*** psachin has joined #openstack-cinder02:42
*** markvoelker has quit IRC02:52
*** sdake has joined #openstack-cinder03:20
*** sdake has quit IRC03:23
*** sdake_ has joined #openstack-cinder03:24
*** _alastor_ has quit IRC03:39
*** zhubx has quit IRC03:46
*** zhubx has joined #openstack-cinder03:47
*** markvoelker has joined #openstack-cinder03:49
*** zhubx has quit IRC03:57
*** sdake_ has quit IRC03:58
*** zhubx has joined #openstack-cinder03:58
*** markvoelker has quit IRC04:22
*** dave-mccowan has quit IRC05:03
*** vishakha has joined #openstack-cinder05:11
*** bhagyashris_ has joined #openstack-cinder05:11
*** ianychoi_ has joined #openstack-cinder05:24
*** ianychoi has quit IRC05:28
*** sdake has joined #openstack-cinder05:31
*** sdake has quit IRC05:31
*** hoonetorg has quit IRC05:33
*** sdake has joined #openstack-cinder05:38
*** bhagyashris_ has quit IRC05:43
*** hoonetorg has joined #openstack-cinder05:50
*** udesale has joined #openstack-cinder06:07
*** gkadam has quit IRC06:13
*** e0ne has joined #openstack-cinder06:31
*** v12aml has quit IRC06:32
*** v12aml has joined #openstack-cinder06:39
*** markvoelker has joined #openstack-cinder06:49
*** Luzi has joined #openstack-cinder06:53
*** artem_vasilyev has joined #openstack-cinder06:58
*** markvoelker has quit IRC07:21
*** ondrejme has joined #openstack-cinder07:35
ondrejmeHow could I remove block storage backend from cinder and also from database in Kolla?07:36
*** bhagyashris has joined #openstack-cinder07:38
whoami-rajatondrejme: you can remove the entry of backend from cinder.conf and restart the cinder services (also delete the associated volume type). For kolla database you can directly go to database (mysql) and remove the entry but that's not recommended. you can ask for the same in kolla channel.07:41
openstackgerritAbhishek Kekane proposed openstack/cinder-specs master: Support multiple backend of Glance  https://review.openstack.org/64126707:46
*** e0ne has quit IRC07:47
ondrejmewhoami-rajat: so the only thing needed is to delete volumes, volume type and restart cinder services?07:49
ondrejmewhoami-rajat: no need to work with database?07:49
*** e0ne has joined #openstack-cinder07:54
*** avishay has joined #openstack-cinder08:07
*** tkajinam has quit IRC08:09
avishayHi all, in our tests (running Ocata) we are seeing CleanableInUse errors sporadically in create_volume and create_snapshot08:13
avishayI saw a few bugs such as this one: https://bugs.launchpad.net/cinder/+bug/171227208:13
openstackLaunchpad bug 1712272 in Cinder "cinder get worker failed " [Undecided,Invalid] - Assigned to wlfightup (wlfightup)08:13
*** gkadam has joined #openstack-cinder08:14
avishay@geguileo wrote there that it is caused by an old version of pymysql, but we are running 0.7.9 and I verified that it contains the referenced fix08:14
avishay(Running one instance of cinder-volume)08:15
*** e0ne has quit IRC08:16
*** markvoelker has joined #openstack-cinder08:19
openstackgerritSilvan Kaiser proposed openstack/cinder master: Always run chmod privileged in remotefs drivers  https://review.openstack.org/64127708:19
*** pcaruana has joined #openstack-cinder08:29
*** markvoelker has quit IRC08:52
whoami-rajatondrejme:  no, the important part is to delete the entry in cinder.conf08:54
whoami-rajatsomething like this08:54
whoami-rajat[lvmdriver-1]08:54
whoami-rajatvolume_group=cinder-volumes-108:54
whoami-rajatvolume_driver=cinder.volume.drivers.lvm.LVMVolumeDriver08:54
whoami-rajatvolume_backend_name=LVM08:54
whoami-rajatondrejme: deleting volumes and volume types related to that backend will automatically wipe out entries in database.08:54
*** dpawlik has joined #openstack-cinder08:57
*** dpawlik has quit IRC09:17
*** lemko has joined #openstack-cinder09:27
*** markvoelker has joined #openstack-cinder09:49
*** bhagyashris has quit IRC09:54
*** luizbag has joined #openstack-cinder10:06
*** markvoelker has quit IRC10:22
*** e0ne has joined #openstack-cinder10:35
openstackgerritRajat Dhasmana proposed openstack/python-cinderclient master: Correct the 'get resource' behavior for non-admin user  https://review.openstack.org/64131010:37
*** artem_vasilyev has quit IRC10:41
*** dviroel_ has joined #openstack-cinder10:43
*** udesale has quit IRC10:57
*** zhubx has quit IRC10:59
*** zhubx has joined #openstack-cinder10:59
*** thgcorrea has joined #openstack-cinder11:00
*** FlorianFa has quit IRC11:01
*** FlorianFa has joined #openstack-cinder11:02
*** pcaruana has quit IRC11:04
*** erlon has joined #openstack-cinder11:09
*** erlon has quit IRC11:16
*** erlon has joined #openstack-cinder11:17
*** markvoelker has joined #openstack-cinder11:19
*** ianychoi_ is now known as ianychoi11:20
*** pcaruana has joined #openstack-cinder11:32
*** dave-mccowan has joined #openstack-cinder11:45
*** markvoelker has quit IRC11:52
*** erlon has quit IRC12:03
*** erlon has joined #openstack-cinder12:04
*** carlos_silva has joined #openstack-cinder12:08
*** enriquetaso has joined #openstack-cinder12:09
*** mvkr has joined #openstack-cinder12:13
*** henriqueof has quit IRC12:16
*** brinzhang has quit IRC12:31
*** udesale has joined #openstack-cinder12:38
*** markvoelker has joined #openstack-cinder12:49
*** avishay has quit IRC12:52
*** mchlumsky has joined #openstack-cinder13:02
*** raghavendrat has joined #openstack-cinder13:03
*** TheJulia_sick is now known as TheJulia13:05
raghavendrathttps://review.openstack.org/#/c/634119/ I have replied to reviewers comments. Not sure if I was able to convince with my explanation. Waiting for response from core reviewers so that we can logically conclude. If any core reviewer is available, appreciate if you can look and respond.13:12
*** lemko has quit IRC13:15
*** markvoelker has quit IRC13:22
hemnamornin13:25
openstackgerritsahid proposed openstack/os-brick master: multipathd: fix support for find_multipaths set to yes  https://review.openstack.org/63863913:28
openstackgerritsahid proposed openstack/os-brick master: multipath: add debug messages of connection threads  https://review.openstack.org/64135013:28
*** mriedem has joined #openstack-cinder13:34
*** arne_wiebalck has quit IRC13:41
*** arne_wiebalck_ has joined #openstack-cinder13:42
whoami-rajatraghavendrat: Hey, I think eharney and hemna  had queries regarding your patch. It would be better if you ask them.13:43
openstackgerritMerged openstack/cinder stable/queens: Fix multiattach set to false after retype  https://review.openstack.org/64106813:45
raghavendratThanks Rajat. I am waiting for eharney and hemna.13:48
hemnasup13:48
*** pcaruana has quit IRC13:51
*** eharney has joined #openstack-cinder13:51
*** sdake has quit IRC13:54
whoami-rajatraghavendrat: you can discuss it here too :)13:56
*** sdake has joined #openstack-cinder14:00
*** pcaruana has joined #openstack-cinder14:01
*** FlorianFa has quit IRC14:09
*** gkadam_ has joined #openstack-cinder14:16
*** sdake has quit IRC14:17
*** gkadam has quit IRC14:19
*** markvoelker has joined #openstack-cinder14:19
*** sdake has joined #openstack-cinder14:21
*** sdake has quit IRC14:24
openstackgerritjacky06 proposed openstack/python-cinderclient master: Update hacking version  https://review.openstack.org/64139514:35
*** arne_wiebalck_ is now known as arne_wiebalck14:35
*** arne_wiebalck has quit IRC14:35
*** arne_wiebalck_ has joined #openstack-cinder14:37
*** sdake has joined #openstack-cinder14:38
*** markvoelker has quit IRC14:52
*** gkadam__ has joined #openstack-cinder15:00
*** sdake has quit IRC15:00
*** gkadam_ has quit IRC15:03
*** sdake has joined #openstack-cinder15:06
*** lseki has joined #openstack-cinder15:06
openstackgerritEric Harney proposed openstack/cinder master: Revert "Use native python truncate for privsep"  https://review.openstack.org/64141315:09
openstackgerritEric Harney proposed openstack/cinder master: Revert "Remove truncate from rootwrap filters"  https://review.openstack.org/64141415:09
*** Luzi has quit IRC15:09
*** gnufied has joined #openstack-cinder15:27
*** udesale has quit IRC15:37
openstackgerritMatthew Booth proposed openstack/cinder master: DNM: Getting extra logs for a volume type delete race  https://review.openstack.org/64142515:50
*** lpetrut has joined #openstack-cinder15:50
*** markvoelker has joined #openstack-cinder15:50
*** pcaruana has quit IRC15:53
*** lseki has quit IRC16:00
*** erlon has quit IRC16:00
*** luizbag has quit IRC16:00
*** thgcorrea has quit IRC16:01
*** luizbag has joined #openstack-cinder16:01
*** erlon has joined #openstack-cinder16:03
*** pcaruana has joined #openstack-cinder16:06
*** sdake has quit IRC16:09
*** sdake has joined #openstack-cinder16:13
*** _alastor_ has joined #openstack-cinder16:17
*** markvoelker has quit IRC16:22
*** arne_wiebalck_ has quit IRC16:32
*** arne_wiebalck_ has joined #openstack-cinder16:32
*** imacdonn has quit IRC16:36
*** imacdonn has joined #openstack-cinder16:36
*** gkadam__ has quit IRC16:42
*** sdake has quit IRC16:48
*** sdake has joined #openstack-cinder16:48
*** pcaruana has quit IRC16:55
*** sdake has quit IRC16:56
*** arne_wiebalck_ has quit IRC16:56
hemnaso whoami-rajat fwiw, we have a customer that's running into problems with the 3PAR driver related to standing up a lot of VMs (100) and creating volumes from snaps17:02
hemnabecause by default the 3PAR driver creates clones for every volume from snap operation17:02
hemnawhich is why I hated that idea from the beginning17:03
*** henriqueof has joined #openstack-cinder17:03
hemnait's crazy fast to create a vol from snap in 3PAR17:03
hemnabut cloning is very slow17:03
hemnathe problem with vols from snaps is that you can't delete the snap, while those volumes exist (dependency)17:03
hemnabut I never understood why that was a problem from cinder's perspective17:04
whoami-rajathemna:  ok.17:05
hemnaand the main usecase is to create a lot of new vms at once17:05
hemnawhich means vols from snaps17:05
hemnaand doing a convert to base on every vol from snap is....s l o w17:05
hemnathen you run into ssh connection issues17:06
hemnaand lots of stuff stops working17:06
*** eharney has quit IRC17:06
hemnaso, that's why I think the default in that new patch should be set to False.17:06
hemnameaning don't convert to base by default.17:06
hemnaand if you REALLY want to have a base volume, then you can create a new vol type and set that new attribute in there.17:07
hemnaour customer wants to create 100 140G volumes from snaps17:08
hemnathat's a lot of bits to copy (for convert_to_base_volume)17:08
hemnait looks like eharney disagrees though17:09
*** angela-s has joined #openstack-cinder17:10
openstackgerritWalter A. Boring IV (hemna) proposed openstack/cinder master: Add get_driver_options method  https://review.openstack.org/63525517:10
whoami-rajathemna: so if we create a volume type with convert_to_base as False and use it instead to create volumes. what problems do we face in this case?17:13
*** eharney has joined #openstack-cinder17:13
angela-ssmcginnis, jungleboyj: Hi Sean, Jay, could you take a look at this cherry pick for Pike? Thanks! https://review.openstack.org/#/c/640568/17:13
*** lpetrut has quit IRC17:14
*** lpetrut has joined #openstack-cinder17:17
hemnathe problem with not creating a base volume, is one of dependency17:17
hemnayou can't delete the snapshot17:18
hemnauntil all of the volumes from that snap have either been deleted or made into a base volume.17:18
hemnaIMHO I think that's a better tradeoff17:18
hemnaas then you can get the 3PAR to act more cloud like.17:18
hemnasuch as being able to create 100 volumes very quickly17:19
*** markvoelker has joined #openstack-cinder17:19
hemnathis is how I originally wrote the driver, because I knew of the 3PAR limitations for creating cloned volumes.17:20
hemnabut this patch changed all that, in an attempt to make rally happy during delete.17:21
hemnahttps://github.com/openstack/cinder/commit/5f2074e1092f07c3420877fa00d1c49efde6e42017:21
whoami-rajathemna:  ohh17:22
*** ntpttr has quit IRC17:22
*** lseki has joined #openstack-cinder17:22
whoami-rajathemna: i thought the initial functionality while implementing the driver was to convert to base volume17:23
hemnabut that patch causes all sorts of problems with rally if you try and create 100 volumes of any size17:23
whoami-rajatso i remember raghavendrat  argued to keep support of the current functionality that wasn't intended at all.17:24
hemnaoriginally the 3PAR driver didn't clone the volume during create from snap.17:24
hemnabecause of all the stuff I mentioned above17:25
hemnait creates a rw snapshot17:25
hemnawhich is basically just a metadata change on the parent snap17:25
hemnacrazy fast17:25
hemnaand it would only convert to a base volume if the new volume was larger than the original snap17:25
hemnahttps://github.com/openstack/cinder/blob/liberty-eol/cinder/volume/drivers/san/hp/hp_3par_common.py#L168417:26
whoami-rajathemna: makes sense. now i understand your concerns regarding the patch.17:27
hemnaunforunately I have a lot of history with the 3PAR :P17:27
whoami-rajathemna: probably the author is just trying to keep the current (but not intended) functionality.17:28
hemnayah17:28
hemnaI think so17:28
whoami-rajathemna: yes, a lot17:29
whoami-rajathemna:  so, the fast solution coming into my mind is17:29
whoami-rajathemna: since the patch is just adding a new extra spec, and your concern is a little bit different from the point of patch, we merge this and then change the default to false in the next PS.17:30
hemnaPS ?17:30
whoami-rajathemna: or add a comment about all the things you explained currently for the usecase and set the option to false as default with that comment.17:31
whoami-rajathemna: oh sorry, followup.17:31
hemnawe have a customer that can make use of that patch today :P17:31
hemnaI just wanted the default to be changed, such that large volume creations work out of the box17:31
hemnaas they used to.17:32
*** lpetrut has quit IRC17:32
*** e0ne has quit IRC17:34
whoami-rajathemna: yes, but the change that shouldn't have merged and this patch brought up the same Case again. Most probably we can summarize the benefit for setting the convert_to_base to false as default in a comment, and add the following to current patch. does that seem like the right approach?17:35
hemnaeither way there should be some documentation associated with this patch to explain this17:36
hemnaand the cost benefits of both True and False for customers.17:37
hemnaI have to go get some lunch..bbiab17:37
whoami-rajathemna: is this worth discussing why the cloning is slower compared to other drivers?17:37
whoami-rajathemna: ok see ya! Thanks for your insights.17:37
hemnawell, that's a 3PAR specific comparison vs. other backends17:38
hemna3par just is very slow in cloning17:38
whoami-rajathemna: ok, better not to dig more into that.17:40
*** mmethot has quit IRC17:44
*** mmethot has joined #openstack-cinder17:45
*** markvoelker has quit IRC17:52
*** psachin has quit IRC18:05
*** irclogbot_0 has joined #openstack-cinder18:11
whoami-rajathemna: I'm really not sure why rally jobs were failing, it is initially deleting the volumes then snapshot[1] so the dependency shouldn't cause problem, do we need any wait_for_resource_delete func to make these operations synchronous, not sure. the bug doesn't seem like a bug now, sometime it may or may not fail IMO.18:20
whoami-rajatAlso i found your review regarding the patch[2], looking at the rally job makes me think we can revert this change.18:20
whoami-rajat[1] https://github.com/openstack/rally-openstack/blob/master/rally_openstack/scenarios/cinder/volumes.py#L834-L83518:20
whoami-rajat[2] https://review.openstack.org/#/c/421739/3/cinder/volume/drivers/hpe/hpe_3par_common.py@221518:20
*** luizbag_ has joined #openstack-cinder18:21
*** luizbag__ has joined #openstack-cinder18:25
whoami-rajathemna: and if we are going forward with reverting this change, we can easily set the value default=False for convert_to_base.18:27
*** luizbag_ has quit IRC18:27
*** luizbag has quit IRC18:27
*** irclogbot_0 has quit IRC18:36
openstackgerritMerged openstack/cinder master: Update minimal acceptable glanceclient  https://review.openstack.org/64103718:46
openstackgerritMerged openstack/cinder master: Update required version of taskflow  https://review.openstack.org/64106518:46
*** markvoelker has joined #openstack-cinder18:50
*** enriquetaso has quit IRC18:54
*** markvoelker has quit IRC19:22
*** zhubx has quit IRC19:59
*** zhubx007 has joined #openstack-cinder19:59
*** luizbag__ has quit IRC20:01
*** e0ne has joined #openstack-cinder20:03
*** arne_wiebalck has joined #openstack-cinder20:06
openstackgerritMerged openstack/cinder stable/pike: Fix for auth version change in Brcd HTTP  https://review.openstack.org/64056820:06
*** markvoelker has joined #openstack-cinder20:19
*** e0ne has quit IRC20:21
*** lemko has joined #openstack-cinder20:29
*** vishakha has quit IRC20:35
hemnadid we get a bug filed about the filter issue?20:51
*** markvoelker has quit IRC20:53
smcginnisNot that I've seen.20:57
smcginnisThere may be an existing one for a very specific side effect of all of that, but not the actual real issue.20:57
*** erlon has quit IRC21:05
*** irclogbot_0 has joined #openstack-cinder21:08
*** e0ne has joined #openstack-cinder21:13
*** whoami-rajat has quit IRC21:18
*** lseki has quit IRC21:21
*** irclogbot_0 has quit IRC21:28
*** eharney has quit IRC21:29
angela-ssmcginnis, jungleboyj, hemna: thanks guys, Pike is done.  Now how about Ocata, man is this tedious. https://review.openstack.org/#/c/640567/ Thank you!21:36
*** markvoelker has joined #openstack-cinder21:50
*** mchlumsky has quit IRC21:53
*** angela-s has quit IRC21:56
*** henriqueof has quit IRC21:57
*** e0ne has quit IRC21:58
*** e0ne has joined #openstack-cinder22:01
*** e0ne has quit IRC22:01
*** carlos_silva has quit IRC22:07
*** bal has quit IRC22:11
*** irclogbot_0 has joined #openstack-cinder22:20
*** markvoelker has quit IRC22:22
*** irclogbot_0 has quit IRC22:34
*** irclogbot_0 has joined #openstack-cinder22:36
*** lemko has quit IRC22:38
*** irclogbot_0 has quit IRC22:44
*** rcernin has joined #openstack-cinder22:52
*** tkajinam has joined #openstack-cinder23:01
*** dave-mccowan has quit IRC23:06
*** markvoelker has joined #openstack-cinder23:19
*** markvoelker has quit IRC23:53
*** gnufied has quit IRC23:59

Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!