Friday, 2018-08-10

*** itlinux has joined #openstack-cinder00:06
itlinuxhi guys..00:06
itlinuxstill trying to figure out why the default does not work..00:07
openstackgerritErlon R. Cruz proposed openstack/cinder master: [NetApp SolidFire] Fix NetApp SolidFire SSL option
openstackgerritErlon R. Cruz proposed openstack/cinder master: NetApp SolidFire: Fix NetApp SolidFire SSL option
itlinuxhello guys, was talking to a couple of you.. what is the option that will make cinder create a volume under a specific backend..00:23
*** dklyle has quit IRC00:32
*** dklyle has joined #openstack-cinder00:32
*** markvoelker has quit IRC00:35
*** Nel1x has joined #openstack-cinder00:44
*** mtreinish has quit IRC00:45
*** mtreinish has joined #openstack-cinder00:46
*** tswanson has quit IRC00:47
*** tswanson has joined #openstack-cinder00:48
*** BrinZhang has joined #openstack-cinder00:50
*** mriedem has quit IRC00:52
*** zhaochao has joined #openstack-cinder00:55
*** mtreinish has quit IRC01:06
*** mtreinish has joined #openstack-cinder01:08
*** zhurong has joined #openstack-cinder01:23
*** markstur has quit IRC02:06
*** e0ne has joined #openstack-cinder02:16
*** e0ne has quit IRC02:18
*** zhurong has quit IRC02:23
openstackgerritErlon R. Cruz proposed openstack/cinder master: NetApp SolidFire: Fix NetApp SolidFire SSL option
erlonitlinux, you have to create a volume type, and set the volume_backend_name to the same as the volume_backend_name set in the backend definition02:31
openstackgerritYu Zhang proposed openstack/cinder master: IBM XIV: add method update_migrated_volume for xiv
*** psachin has joined #openstack-cinder02:48
*** tswanson has quit IRC02:50
openstackgerritMerged openstack/cinder stable/pike: VSA: Concurrent request handling in attachment
*** Nel1x has quit IRC03:04
itlinuxok let me try that..03:47
*** BrinZhang has quit IRC03:50
*** BrinZhang has joined #openstack-cinder03:51
itlinuxI have created the type that maps to the zone.. and it still goes to a diff zone when I boot up a vm from the UI03:52
itlinuxI am running PIKE03:55
itlinuxthe is no diff on the behaviors with or without volume type..03:55
*** ianychoi has joined #openstack-cinder03:56
itlinuxI can use the az from the UI and create a volume to the az and that works the probem it does not work when a vm books up and does not honor the Host AG AZ03:56
openstackgerritMerged openstack/cinder stable/rocky: Update .gitreview for stable/rocky
*** BrinZhang has quit IRC04:45
*** BrinZhang has joined #openstack-cinder04:46
*** itlinux has quit IRC04:51
*** gkadam has joined #openstack-cinder05:12
*** hoonetorg has joined #openstack-cinder05:25
*** nicolasbock has joined #openstack-cinder05:50
*** BrinZhang has quit IRC05:51
*** BrinZhang has joined #openstack-cinder05:52
*** brinzh has joined #openstack-cinder05:52
*** brinzh has quit IRC05:53
*** BrinZhang has quit IRC05:53
*** brinzhang has joined #openstack-cinder05:54
*** brinzhang has quit IRC06:22
*** brinzhang has joined #openstack-cinder06:23
*** pcaruana has joined #openstack-cinder06:43
*** evrardjp has joined #openstack-cinder07:06
*** rcernin has quit IRC07:07
*** brinzhang has quit IRC07:37
*** brinzhang has joined #openstack-cinder07:38
*** josecastroleon has quit IRC07:58
*** josecastroleon has joined #openstack-cinder07:59
*** e0ne has joined #openstack-cinder07:59
openstackgerritOpenStack Proposal Bot proposed openstack/cinder master: Imported Translations from Zanata
*** psachin has quit IRC08:13
*** psachin has joined #openstack-cinder08:27
*** e0ne has quit IRC08:49
openstackgerritErlon R. Cruz proposed openstack/cinder master: NetApp SolidFire: Fix NetApp SolidFire SSL option
*** e0ne has joined #openstack-cinder09:14
*** josecastroleon has quit IRC09:20
openstackgerritBoxiang Zhu proposed openstack/cinder master: relocate return of create_volume and _clone in rbd driver
*** e0ne has quit IRC09:27
*** KeithMnemonic has quit IRC09:33
*** e0ne has joined #openstack-cinder09:37
*** josecastroleon has joined #openstack-cinder09:38
*** e0ne has quit IRC10:21
*** dave-mccowan has joined #openstack-cinder10:55
*** dave-mccowan has quit IRC11:00
*** dave-mccowan has joined #openstack-cinder11:02
tobias-urdincan I run multiple cinder-volume if i specify host config option the same (and it has access to same storage etc) to increase the concurrency for cinder to create/delete volumes etc?11:06
*** erlon has quit IRC11:06
tobias-urdinI have a coordinator that can be shared between this two nodes, are locks or similar functionality shared then?11:06
*** nicolasbock has quit IRC11:19
*** mriedem has joined #openstack-cinder12:05
*** brinzhang has quit IRC12:07
*** mriedem has quit IRC12:17
*** luizbag has joined #openstack-cinder12:23
*** mchlumsky has joined #openstack-cinder12:26
*** abishop has joined #openstack-cinder12:35
*** psachin has quit IRC12:39
*** tpsilva has joined #openstack-cinder12:44
*** eharney has joined #openstack-cinder13:14
smcginnistobias-urdin: It's risky doing that. If one of the nodes are restarted while some task is running on the other node, as it comes back up it will see that task out there, think it's its own, and try to clean it up.13:17
openstackgerritXianjin Shao proposed openstack/cinder master: Add the all_tenants when query count of volumes
smcginnistobias-urdin: So they could interfere with each other.13:17
*** KeithMnemonic has joined #openstack-cinder13:19
KeithMnemonicI think this one is next. if someone can please review and W +1  queens and pike are updated and merged so now working on the ocata branch13:20
tobias-urdinsmcginnis: I see, thanks. Is there any recommended way to improve performance for creation of multiple volumes at the same time? The only way I see otherwise would be having a lot of cinder-volume services but they operate on the same Ceph pool?13:21
tobias-urdin(but not the same so they operate independently)13:21
smcginnistobias-urdin: I think there are some ways to do some tuning for that, but I haven't done that myself.13:23
smcginnistobias-urdin: We've merged the ability to run active/active services in a "cluster" so they are aware of each other and don't have this issues, but for the volume service we have not actually enabled any backend drivers to support it yet.13:23
smcginnisYou mention ceph. Maybe some of the Red Hat folks have some tips on performance optimizations there.13:24
smcginnisThere are usually a few in this channel, so we can see if they come around.13:24
smcginnistobias-urdin: In case it helps, here some docs that were just added that at least gives some background on that -
tobias-urdinsmcginnis: Thanks, looking forward to this feature being available and stable, been interested in it for years :)13:27
*** dpawlik has quit IRC13:30
*** dustins has joined #openstack-cinder13:31
*** dansmith is now known as SteelyDan13:31
smcginnisUnfortunately it has taken us years.13:31
*** jistr is now known as jistr|call13:46
*** jistr|call is now known as jistr14:00
*** e0ne has joined #openstack-cinder14:11
*** eharney has quit IRC14:13
*** cmart has joined #openstack-cinder14:19
*** eharney has joined #openstack-cinder14:26
*** e0ne has quit IRC14:39
*** jgriffith has quit IRC14:46
*** markstur has joined #openstack-cinder14:48
*** markstur has quit IRC14:48
*** e0ne has joined #openstack-cinder15:03
*** dpawlik has joined #openstack-cinder15:05
geguileotobias-urdin: what version of OpenStack are you working with?  What is the performance hit you are seeing?15:05
tobias-urdingeguileo: Queens, doing Rally tests against Cinder I've been told by a colleague that we're getting failures/timeouts when creating large amount of volumes at concurrently, they take too long.15:08
geguileotobias-urdin: empty volumes or volumes from images15:09
*** dpawlik has quit IRC15:09
tobias-urdinI'll email him and see if cans end over the Rally and Cinder logs that he hopefully collected. I would assume from images but not sure, that shouldn't involve Glance though right? With RBD backend it would just clone the volume in Ceph, but perhaps I'm wrong they are on different pools.15:10
*** e0ne has quit IRC15:10
geguileodepends on how things are setup15:10
geguileoalso, are you setting 'rbd_exclusive_cinder_pool' to True?15:11
geguileohow many volumes are being simultaneously being created?15:11
geguileodepending on the number performance could benefit from increasing the native thread pool15:11
*** cmart has quit IRC15:24
*** itlinux has joined #openstack-cinder15:24
itlinuxhello all..15:24
*** jgriffith has joined #openstack-cinder15:26
itlinuxI have cinder with AZ running and it works with volume-type as well but when I spin up a VM it does not respect the AZ from Nova.. The Nova AZ are the same in cinder15:26
itlinuxI have now some issue that if I do not specify the az the default cinder create 1 does fail15:27
*** mriedem has joined #openstack-cinder15:30
*** jgriffith has quit IRC15:31
*** jgriffith has joined #openstack-cinder15:31
openstackgerritMerged openstack/cinder master: Imported Translations from Zanata
*** rajinir has joined #openstack-cinder15:36
*** cmart has joined #openstack-cinder15:37
rajinirI got his customer asking for config parameter inside Openstack that volumes would go to recycle bin instead of deleting volumes.  Anyone?15:47
*** tswanson has joined #openstack-cinder15:52
*** cmart has quit IRC15:53
*** pocketprotector has quit IRC15:57
*** pocketprotector has joined #openstack-cinder15:59
geguileorajinir: there was a proposal to do that on the last PTG, I don't remember where we ended up with it16:01
geguileoitlinux: I don't think it's Cinder the one that has to respect the AZ, I believe it's Nova16:02
rajinirok, thanks geguileo>16:02
geguileorajinir: it should be in the etherpads16:02
rajinirgeguileo: ok will check16:02
geguileoitlinux: that's why in Nova you can configure it to support cross AZ attachments16:02
geguileoitlinux: the cinder create fails because we didn't have a way to define the default AZ in the volume type16:03
geguileoitlinux: I think we added the support recently16:03
geguileoitlinux: it's supported in Queens16:04
*** gkadam has quit IRC16:06
geguileorajinir: my bad, it was only for RBD:
geguileorajinir: originally the trash part was brought up in the cinder IRC meeting
geguileorajinir: line 62016:10
*** mriedem has quit IRC16:27
rajinir<geguileo>: any chance of revisiting this for all backends?16:30
imacdonnwhat would "gp to recycle bin" mean, in a generic sense?16:37
imacdonnI assume it'd have to be accompanied by some sort of "undelete" function16:39
imacdonnto be useful16:39
*** cmart has joined #openstack-cinder16:41
imacdonnthen you have to think about implication on things like quota ... does a volume "in the recycle bin" still consume quota ?16:44
imacdonnThe Ceph thing seems to be more of an "under the covers" option ... the value-add of that is not really clear, but it doesn't appear that the "recycle bin" aspect would be exposed to users16:46
*** rmcall has quit IRC16:50
*** hoonetorg has quit IRC17:04
itlinuxgeguileo: thanks.. but I am in PIKE :)17:06
itlinuxso now adding the allow_availability_zone_failback = true on all of them since I saw that if I create a non declear volume it fails, then try again fails then try again goes..17:07
itlinuxand works..17:07
itlinuxso that maybe the case ..17:07
itlinuxwell the other thing could be to create a flavor with it .. not sure that would work either though!17:09
itlinuxI see it fails when it tries to use the nova az in cinder.. even though I do not see it in the list..17:10
*** e0ne has joined #openstack-cinder17:11
*** N3l1x has joined #openstack-cinder17:13
rajinirimacdonn>  yes, an undelete, many things to consider17:17
imacdonnyeah, so that would be a fairly major architectural change to cinder, I think ... possibly doable, but not trivial17:17
*** hoonetorg has joined #openstack-cinder17:19
*** erlon has joined #openstack-cinder17:24
geguileorajinir: I'm not too sure myself about the whole "recycle bin" idea and there would need to be some developer who's interested in doing it17:25
geguileoitlinux: you say that you have to try 3 times and it goes?17:26
geguileoitlinux: how many API nodes do you have running?17:26
itlinuxI have HA mode..17:27
*** zhaochao has quit IRC17:27
geguileoitlinux: so 3 nodes, and you had to repeat the request 3 times?17:27
geguileoitlinux: I would say you don't have all your API nodes running the same configuration17:27
itlinuxbut that should not matter since I only have a main one and it depends on the config I notice that at time it does not work and at times it's just fine.. if I take the AZ off with one pure system it works every time..17:27
geguileoitlinux: if it's HA you don't have just a main API node, you should have 3 nodes running behind HAproxy17:28
itlinuxthis is standard OOO deployment so I don't see what it should fail17:28
*** eharney_ has joined #openstack-cinder17:30
*** erlon has quit IRC17:33
*** eharney has quit IRC17:33
itlinuxyes correct geguleo17:51
*** ianychoi has quit IRC18:10
*** dustins has quit IRC18:10
*** ianychoi has joined #openstack-cinder18:11
itlinuxwhat's the best way to reset the availability zone in cinder..18:13
itlinuxI am trying to get back and start again.. but I see additional az and I want them to remove them18:14
*** erlon has joined #openstack-cinder18:14
imacdonnitlinux: you may need to use 'cinder-manage service remove ...'18:21
itlinuxahh I do not think that I have that api..18:21
itlinuxya does not look like it's installed..18:22
itlinuxwhich python package is it: imacdonn:18:22
imacdonncinder-manage is part of the 'openstack-cinder' RPM in RDO .... it's used to db sync, so it should be installed already18:24
*** erlon has quit IRC18:24
itlinuxgot it..18:26
itlinuxwhat command will be looking like cinder-manage service remove ??18:27
imacdonnyou can do 'cinder-manage service list", then find the one that you want to do delete, and 'cinder-manage service remove <binary> <host>' (where <binary> should be 'cinder-volume')18:28
*** luizbag has quit IRC18:28
itlinuxok that's not going to help me since I do not have that package on the overcloud..18:28
imacdonnyou'll probably have to do it on the host that's running the cinder control services (API, scheduler)18:29
itlinuxwell I can remove it directly from the db.. but that's not a good option..18:30
itlinuxI can disable them...18:30
geguileoitlinux: it's probably installed int he undercloud18:48
geguileoitlinux: just connect to a controller node and it's probably there18:48
geguileopatrickeast: hemna here's a patch to improve multipathing disconnect if you have the time:
*** eharney_ is now known as eharney18:52
*** cmart has quit IRC18:58
*** dustins has joined #openstack-cinder19:02
*** pcaruana has quit IRC19:19
*** dpawlik has joined #openstack-cinder19:20
*** cmart has joined #openstack-cinder19:22
*** dpawlik has quit IRC19:24
*** cmart has quit IRC19:44
*** eharney has quit IRC19:45
*** gkadam has joined #openstack-cinder19:47
*** e0ne has quit IRC19:51
*** eharney has joined #openstack-cinder19:57
*** itlinux has quit IRC19:59
*** alysonR__ has quit IRC19:59
openstackgerritMatthew Edmonds proposed openstack/cinder master: Make solidfire code py3-compatible
*** itlinux has joined #openstack-cinder20:05
itlinuxhello imacdonn: you still around?20:05
*** cmart has joined #openstack-cinder20:24
*** eharney has quit IRC20:46
*** abishop has quit IRC20:47
*** wN has joined #openstack-cinder20:53
*** dustins has quit IRC20:54
*** mchlumsky has quit IRC21:11
*** dave-mccowan has quit IRC21:24
*** cmart has quit IRC21:29
*** gnufied__ has quit IRC21:37
*** cmart has joined #openstack-cinder21:42
*** imacdonn has quit IRC22:12
*** imacdonn has joined #openstack-cinder22:13
*** itlinux has quit IRC22:28
*** gkadam has quit IRC22:34
*** cmart has quit IRC22:35

Generated by 2.15.3 by Marius Gedminas - find it at!