Tuesday, 2017-01-17

« Monday, 2017-01-16 Index Wednesday, 2017-01-18 »
*** catintheroof has joined #openstack-manila00:11
*** mtan_____ has joined #openstack-manila00:28
*** mtanino has quit IRC00:29
openstackgerritRodrigo Barbieri proposed openstack/manila: Add mountable snapshots support  https://review.openstack.org/34552600:35
*** lgreg has quit IRC01:01
*** catintheroof has quit IRC01:09
*** cknight has joined #openstack-manila01:31
*** mtan_____ has quit IRC01:57
*** ganso has quit IRC02:04
openstackgerritzhongjun proposed openstack/python-manilaclient: Add the ability to check tenant quota detail  https://review.openstack.org/36856702:13
*** gcb has joined #openstack-manila02:36
openstackgerritGoutham Pacha Ravi proposed openstack/manila: [DNM] Debug Data helper bulk access changes  https://review.openstack.org/42102802:44
openstackgerritClinton Knight proposed openstack/manila: Implement share revert to snapshot  https://review.openstack.org/34050203:04
*** gouthamr has quit IRC03:12
openstackgerritClinton Knight proposed openstack/python-manilaclient: Support share revert to snapshot in Manila client  https://review.openstack.org/34055603:18
*** cknight has quit IRC03:26
openstackgerritGoutham Pacha Ravi proposed openstack/manila: [DNM] Debug Data helper bulk access changes  https://review.openstack.org/42102804:06
openstackgerritMerged openstack/manila: Add MapR-FS native driver  https://review.openstack.org/35003504:33
*** houming has joined #openstack-manila05:17
*** houming has quit IRC05:20
*** houming has joined #openstack-manila05:21
openstackgerritSharat Sharma proposed openstack/manila-ui: Changed the home-page of manila-ui  https://review.openstack.org/38109605:38
*** nkrinner_afk is now known as nkrinner06:14
*** jcsp has quit IRC06:16
*** houming has quit IRC06:53
*** lpetrut has joined #openstack-manila06:58
*** houming has joined #openstack-manila07:00
*** dsariel has joined #openstack-manila07:05
*** houming has quit IRC07:06
*** houming has joined #openstack-manila07:07
*** tuanluong has joined #openstack-manila07:09
openstackgerritPony Chou proposed openstack/manila: Add QNAP Manila Driver  https://review.openstack.org/39470307:15
*** jprovazn has joined #openstack-manila07:22
*** jprovazn has quit IRC07:22
*** jprovazn has joined #openstack-manila07:23
*** lpetrut has quit IRC07:46
*** a-pugachev has joined #openstack-manila07:50
*** lpetrut has joined #openstack-manila07:53
*** a-pugachev has quit IRC07:57
*** dmellado has quit IRC08:02
*** dmellado has joined #openstack-manila08:04
openstackgerritHa Van Tu proposed openstack/manila: [api-ref] Refactor share metadata documentation  https://review.openstack.org/42111808:07
openstackgerritzhongjun proposed openstack/manila: Add manila-manage db purge command  https://review.openstack.org/30821208:20
openstackgerritHa Van Tu proposed openstack/manila: [api-ref] Refactor share migration documentation  https://review.openstack.org/42112908:28
*** tuanluong has quit IRC09:03
*** tuanluong has joined #openstack-manila09:04
*** a-pugachev has joined #openstack-manila09:18
openstackgerritzhongjun proposed openstack/manila: Fix metadata's soft-delete error when deleting shares  https://review.openstack.org/42117009:28
tommylikehuping vponomaryov09:36
*** ganso has joined #openstack-manila09:51
tommylikehuvponomaryov, I think the main cause of the bug you mentioned in the comment is another bug about share deletion, check here : https://review.openstack.org/#/c/308212/09:56
*** yumiriam has joined #openstack-manila09:57
openstackgerritRodrigo Barbieri proposed openstack/python-manilaclient: Implement Share Migration Ocata improvements  https://review.openstack.org/40630609:58
openstackgerritRodrigo Barbieri proposed openstack/manila: Mountable snapshots scenario tests  https://review.openstack.org/41200110:00
openstackgerritRodrigo Barbieri proposed openstack/manila: Add mountable snapshots support to HNAS driver  https://review.openstack.org/41147410:00
*** alyson_ has joined #openstack-manila10:05
openstackgerritTina Tang proposed openstack/manila: [Unity driver] VLAN enhancement  https://review.openstack.org/41003710:22
*** zengyingzhe has quit IRC10:30
*** zengyingzhe has joined #openstack-manila10:30
openstackgerritRamana Raja proposed openstack/manila: WIP: Refactor and rename CephFSNativeDriver  https://review.openstack.org/42120110:34
openstackgerritMerged openstack/manila-image-elements: Suppress some bashate warnings  https://review.openstack.org/42091110:35
*** tommylikehu has quit IRC10:40
openstackgerritMerged openstack/manila-image-elements: Enables end user to pick share protocol  https://review.openstack.org/40041110:40
openstackgerritRodrigo Barbieri proposed openstack/python-manilaclient: Implement Share Migration Ocata improvements  https://review.openstack.org/40630610:47
*** tommylikehu has joined #openstack-manila10:48
*** lpetrut has quit IRC11:06
openstackgerritMerged openstack/manila: tests: remove useless variables in db_utils methods  https://review.openstack.org/42088711:08
*** ociuhandu has quit IRC11:23
*** houming has quit IRC11:28
*** houming has joined #openstack-manila11:35
openstackgerritLucio Seki proposed openstack/manila: Mountable snapshots scenario tests  https://review.openstack.org/41200111:36
openstackgerritAlyson proposed openstack/python-manilaclient: Add mountable snapshots support to manila client  https://review.openstack.org/34562511:40
*** JoseMello has joined #openstack-manila11:43
openstackgerritValeriy Ponomaryov proposed openstack/python-manilaclient: Add share group support to Manila client  https://review.openstack.org/33512011:43
openstackgerritValeriy Ponomaryov proposed openstack/manila: Rename consistency group modules to share groups  https://review.openstack.org/40986411:44
openstackgerritValeriy Ponomaryov proposed openstack/manila: Manila Share Groups  https://review.openstack.org/33509311:54
openstackgerritValeriy Ponomaryov proposed openstack/manila: [Tempest] Add functional tests for share groups feature  https://review.openstack.org/35526411:55
*** rraja has joined #openstack-manila11:55
*** tuanluong has quit IRC11:59
*** cdelatte has joined #openstack-manila11:59
*** timcl has joined #openstack-manila11:59
*** rraja has quit IRC12:00
*** lpetrut has joined #openstack-manila12:10
openstackgerritTommyLike proposed openstack/manila: Fix metadata's soft-delete error when deleting shares  https://review.openstack.org/42117012:26
*** mkoderer has quit IRC12:26
*** dsariel has quit IRC12:30
*** catintheroof has joined #openstack-manila12:32
*** gouthamr has joined #openstack-manila12:48
*** houming has quit IRC12:48
gansogouthamr: ping12:51
gouthamrganso: pong12:51
gansogouthamr: Hi Goutham, good morning12:51
gouthamrganso: Hey Rodrigo... 'Morning!12:52
gansogouthamr: I am still a bit confused by https://review.openstack.org/#/c/413783/4/manila/share/drivers/netapp/dataontap/cluster_mode/lib_base.py@179012:52
gansogouthamr: if the volume does not exist, I guess migration shouldn't even start, correct?12:53
gouthamrganso: a non-existent volume move operation. _get_volume_move_status only raises an error if the move job on the backend doesn't exist, i.e, possibly already canceled12:54
gansogouthamr: oh I get it now, already cancelled12:54
gansogouthamr: thanks12:55
gouthamrganso: np! thanks for reviewing.12:55
gansogouthamr: btw, I am very confused about non-disruptive vs nondisruptive12:55
gansogouthamr: english is not my native language so I am unsure, but I saw you comment my patches before correcting non-disruptive to nondisruptive12:56
gansogouthamr: so I am confused12:56
gouthamrganso: nondisruptive isn't a word.. we made it one word because it's easier for a CLI user..12:56
gouthamrganso: i corrected it in documentation where you were referring to the option perhaps..12:57
gansogouthamr: hum I do not agree with that, does not make that much easier for the CLI user, we have preserve-metadata, preserve-snapshot, etc12:58
*** gcb has quit IRC12:58
gansogouthamr: http://www.netapp.com/us/communities/tech-ontap/tot-ndo-1308.aspx12:59
gansogouthamr: so people use "nondisruptive" as a word12:59
gouthamrganso: sigh, you point me to the absolute source of truth and logic13:00
gansogouthamr: also "nondisruptively"13:00
*** rraja has joined #openstack-manila13:00
gansogouthamr: it is not the absolute source of truth, but it is funny it is from NetApp :P13:00
gouthamrganso: lol.. tech writers invent words.. i'll change it if there's anything else to change in that patch13:02
*** dustins has joined #openstack-manila13:09
*** eharney has joined #openstack-manila13:10
openstackgerritGoutham Pacha Ravi proposed openstack/manila: NetApp cDOT: Add Intra-vserver migration support  https://review.openstack.org/41378313:14
openstackgerritMerged openstack/manila-ui: Changed the home-page of manila-ui  https://review.openstack.org/38109613:17
*** ianychoi has quit IRC13:18
gansoI am going to review your patch later today, it is next in my queue13:22
gansogouthamr: ^13:22
gouthamrganso: access_rules? migration? both? :P13:22
gansogouthamr: access rules13:22
gansomigration looks fine13:22
gansogouthamr: btw, I would like to take a look at CI results migrating snapshots13:23
gouthamrganso: ah, np.. yes.. the latter depends on your ocata improvements patch13:23
gansogouthamr: but we don't have tests that do this13:23
vponomaryovvkmc, tbarron, markstur, ganso: https://bugs.launchpad.net/manila-image-elements/+bug/165712113:24
openstackLaunchpad bug 1657121 in manila-image-elements "Default set of protocols is expected to be nfs + cifs, but it is only nfs" [Critical,New]13:24
vkmcvponomaryov, fixing that, thanks for the quick report13:24
*** dsariel has joined #openstack-manila13:25
gouthamrganso: my CI isn't smart enough to process gerrit dependencies yet, i.e, lacks a "DependantPipelineManager" - Yogesh's writing some tests around your ocata improvements patch and running them against cDOT, he'll post test results on the patch..13:26
gansogouthamr: thanks!13:26
openstackgerritGoutham Pacha Ravi proposed openstack/manila: Refactor Access Rules APIs  https://review.openstack.org/36966813:47
openstackgerritTommyLike proposed openstack/manila: Fix metadata's soft-delete error when deleting shares  https://review.openstack.org/42117013:55
*** tommylikehu_ has joined #openstack-manila13:58
*** ianychoi has joined #openstack-manila14:13
*** ociuhandu has joined #openstack-manila14:16
*** gcb has joined #openstack-manila14:36
*** porrua has joined #openstack-manila14:40
*** dustins has quit IRC14:47
bswartzganso tbarron markstur vkmc: ping re https://review.openstack.org/#/c/400411/14:48
bswartzguys ^ this change seems to have broken our generic driver jobs14:48
vkmcbswartz, yes, vponomaryov reported the bug, I'm fixing it14:48
bswartzaccording to vponomaryov it was previously possible to support NFS and CIFS at the same time, but after this change it is not14:49
bswartzvkmc: this is somewhat urgent so I'm wondering if you plan to have a fix today or if we should temporarily revert your whole change?14:49
vkmcbswartz, I'm planning to have a fix now14:49
bswartzvkmc: like in the next few minutes? in an hour?14:50
vkmcbswartz, like in the next few minutes14:50
bswartzawesome ty14:50
vkmcanytime14:50
tommylikehuping gouthamr14:51
gouthamrtommylikehu: pong14:51
tommylikehuhey gouthamr I would like to have a short talk with about your comments here14:52
tommylikehuhttps://review.openstack.org/#/c/421170/2/manila/db/sqlalchemy/api.py14:52
tommylikehuI would like to keep the deletion logic in share_delete method rather than the share_instance_deletion method14:52
gouthamrtommylikehu: hmmm, then you'd rewrite share_delete to accept the same session14:53
gouthamrtommylikehu: i don't see any value for doing that though..14:53
tommylikehuit's a basic crud action for share14:54
tommylikehugouthamr: one session could fail because we use session.begin inside another14:57
gouthamrtommylikehu: yes, you'd like to bind the deletion to the same session, viz, include the line to delete metadata in the share_instance_delete method..14:58
gouthamrtommylikehu: the only place we use share_delete is in the share create API *if* we can't create on the database for some reason.. i don't see what can fail there though..14:59
tommylikehugouthamr, oh so we just back to your first comment14:59
*** JoseMello has quit IRC15:05
*** cknight has joined #openstack-manila15:05
openstackgerritTommyLike proposed openstack/manila: Add manila-manage db purge command  https://review.openstack.org/30821215:13
*** pcaruana has quit IRC15:15
*** nkrinner is now known as nkrinner_afk15:17
openstackgerritVictoria Martinez de la Cruz proposed openstack/manila-image-elements: Changes default behaviour for image creation script  https://review.openstack.org/42131715:22
vkmcvponomaryov, bswartz, tbarron ^15:23
vponomaryovtommylikehu: please, look at comment here -> https://review.openstack.org/#/c/308212/21/manila/db/sqlalchemy/api.py15:24
vponomaryovtommylikehu: even if it was not failing with one of tables, it would be huge list of messages using old approach15:25
vponomaryovtommylikehu: but it would be useful to log amount of deleted records per each DB table15:25
vponomaryovtommylikehu: post-factum15:25
vponomaryovtommylikehu: when they really got deleted15:25
tommylikehuvponomaryov: thanks15:25
*** gcb has quit IRC15:26
*** xyang_ has joined #openstack-manila15:27
vponomaryovtommylikehu: also what have you decided with gouthamr about https://review.openstack.org/#/c/421170/ ?15:28
openstackgerritTommyLike proposed openstack/manila: Fix metadata's soft-delete error when deleting shares  https://review.openstack.org/42117015:28
vponomaryovtommylikehu: it is depenency for "purge feature" , de-facto15:28
tommylikehuvponomaryov to delete the share metadata at the share_instance_delete method15:29
tommylikehuI would like to add the dependecy at the commit message15:30
openstackgerritClinton Knight proposed openstack/manila: Implement share revert to snapshot  https://review.openstack.org/34050215:31
*** akerr has joined #openstack-manila15:32
gansotommylikehu: dependency of what?15:40
vponomaryovganso: https://review.openstack.org/#/c/308212 depends on https://review.openstack.org/#/c/42117015:40
tommylikehuthe purge db feature patch is dependented on the bug fix one 42177015:40
gansotommylikehu: use a "Depends-on:" or rebase on top15:41
tommylikehuganso: thanks, adding it15:42
gansovponomaryov, tbarron, bswartz, gouthamr: shall we get the gate fix merged? https://review.openstack.org/#/c/42131715:42
*** david-lyle has joined #openstack-manila15:42
*** xyang_ has quit IRC15:42
vponomaryovvkmc: do you like "default" approach?15:43
vponomaryovthat you propose?15:43
vponomaryovI would like to see something more explicit15:43
vponomaryovlike "nfs_cifs"15:44
vponomaryovor "cifs_nfs"15:44
vponomaryovsupporting both15:44
vponomaryovfor ease of usage15:44
vponomaryovganso: ^15:44
gansovponomaryov: I am fine with "default"15:44
*** dustins has joined #openstack-manila15:45
vponomaryovganso: come on, you just don't care ))15:46
vponomaryovganso: have you ever built service image using our manila-image-elements? ))15:46
vponomaryovI would say "the more explicitness we have there the better"15:47
gansovponomaryov: the default is documented15:49
gansovponomaryov: user will use the same command, and produce the same result15:49
gansovponomaryov: I believe you are suggesting to change the internal variable value15:49
vponomaryovganso: the one that will be used with tox config15:50
vponomaryovto reduce the need to read docs15:50
gansovponomaryov: go ahead and -1 the patch then15:53
openstackgerritTommyLike proposed openstack/manila: Fix metadata's soft-delete error when deleting shares  https://review.openstack.org/42117015:53
*** xyang_ has joined #openstack-manila15:55
*** david-lyle has quit IRC15:55
bswartzganso: sorry I was AFK, looking now15:59
bswartzI'm happy with it if vponomaryov is -- I agree that "default" is less clear than "nfs+cifs"16:01
vponomaryovbswartz: I will not be a blocker not having allies here ))16:02
*** akerr has quit IRC16:02
*** rraja has quit IRC16:03
bswartzvponomaryov I'm your ally!16:03
tbarronvponomaryov: vkmc ganso bswartz I would like to be able to say 'nfs cifs' or 'zfs nfs' or any combination but current approach looks fine to unblock16:03
bswartzwho is opposed to nfs+cifs instead of default? vkmc?16:04
vponomaryovtbarron: it should be small change16:04
*** tommylikehu_ has quit IRC16:04
vponomaryovtbarron: to find substitution16:04
gansotbarron: you can, but you need to define 2 env_vars16:05
vponomaryovtbarron: and, looks like, it was better to have old approach16:05
vponomaryovtbarron: env var per protocol16:05
*** ociuhandu has quit IRC16:05
gansotbarron: like, MANILA_SHARE_PROTO="custom" and OPTIONAL_ELEMENTS="manila-zfs manila-cifs manila-nfs"16:06
gansovponomaryov: also, it seems elif is trying to concatenate but it does not actually work16:06
openstackgerritTommyLike proposed openstack/manila: Add manila-manage db purge command  https://review.openstack.org/30821216:06
gansovponomaryov: since it will always enter 1 elif, not more than one16:06
tbarronsorry, not paying that much attention, we're in real-time meeting elsewhere.  If this patch unblocks why not go with it and then refine in another patch?16:07
gansotbarron: yea, we could go with it and fix logic in another patch, I'm in for unblocking now16:07
vponomaryovok16:07
*** david-lyle has joined #openstack-manila16:07
gansogouthamr pulled the trigger16:08
gansogouthamr: xD16:08
*** absubram has joined #openstack-manila16:09
gouthamrsmokin' gun16:11
*** bswartz has quit IRC16:13
openstackgerritMerged openstack/manila-image-elements: Changes default behaviour for image creation script  https://review.openstack.org/42131716:14
marksturSo we don't have a good syntax for building w/ more than one protocol support other than the special default one16:15
marksturI'd say that is OK for now, until someone needs it.16:15
marksturI don't much like using "contains" and strings like cifs_nfs anyway.16:16
*** breitz has joined #openstack-manila16:17
*** xinyanzhang has quit IRC16:19
*** xinyanzhang has joined #openstack-manila16:19
marksturOh maybe that script already works with -s nfs -s cifs16:20
marksturnope16:20
vponomaryovmarkstur: so grown up and sooooo too optimistic ))16:24
*** carthaca_ has joined #openstack-manila16:30
*** bswartz has joined #openstack-manila16:31
*** dustins has quit IRC16:34
*** dustins_ has joined #openstack-manila16:34
marksturDid he just call me old?16:34
marksturIt's the dementia that makes me optimistic.  :)16:35
gouthamrganso: take note, markstur might want to join us after this16:36
marksturLOL.  Did someone say "allies"?16:36
vponomaryovgouthamr: join club of hating vponomaryov? )16:36
markstur</oppression>16:36
gansogouthamr: lol indeed, markstur is welcome to be a member of the club16:37
gouthamrvponomaryov: lol :D hate is a strong word16:37
vponomaryovgouthamr: strong "enough" ? )16:37
gouthamrvponomaryov: yeah, we're going to get you to buy in ATL16:38
marksturI've always been a big fan of vponomaryov, but if I'm drafted by another team, then what can I do16:38
vponomaryovmarkstur: every epoch ends sometime16:38
marksturit's his philosophical wisdom that I just can't argue with16:40
vponomaryovand control shot in the head: "every epoch ends sometime", so appreciate every minute you live in it16:42
vponomaryovliek you do with manila project ^_^16:42
vponomaryovtommylikehu: here?16:43
vponomaryovone small comment left in https://review.openstack.org/#/c/308212 to be addressed so we could merge it16:43
marksturthat first part was kind of dark.  I'm gonna try to forget that and do the "appreciate every minute..." thing today16:43
vponomaryovgouthamr: are you ok with https://review.openstack.org/#/c/421170/5 ?16:45
vkmcsorry, I'm having some connection issues16:47
vkmcso... I was planning to look for a way to enable users to pick more than one proto16:48
vkmcusing the nfs_cifs approach seemed not to be a good idea for now16:48
vkmcsince it doesn't scale very well16:49
*** dustins_ has quit IRC16:52
*** dustins has joined #openstack-manila16:52
marksturvkmc: Not sure there is a best style to follow. With help being the only other flag it could just be cmd [-h] [opts...], but if you want to allow for more options you'd either keep the variable opts (protocols) at the end or I guess choose between along multiple -s or probably -s nfs,cifs,zfs.  At least the comma thing is what the "mount" command does so it aligns with something16:54
marksturs/along/allowing/16:55
vponomaryovgouthamr: do you have any preference merging one of your following changes first - https://review.openstack.org/#/c/419163 and https://review.openstack.org/#/c/369668 ?16:55
vkmcmarkstur, yes, I was looking into that :) I was looking for a more fancy way of doing it since the checks for arrays in bash can get  very complex with just a few combinations16:56
marksturI'd use getopt16:57
gouthamr vponomaryov:  https://review.openstack.org/#/c/369668 can go in first17:03
marksturvkmc: and you would just be concatenating onto OPTIONAL_ELEMENTS so you could do that in your current loop or with getopt. You'd just need to change the default to be set when needed instead of set and overwritten.17:04
vponomaryovgouthamr: ok17:05
marksturvkmc: but still it is speculative if no one needs it yet and could be simply  cmd [-h] [OPTIONAL_ELEMENTS]17:05
markstursomeone out there probably has future plans for more opts tho or has strong feelings about command flags17:05
*** a-pugachev has quit IRC17:17
*** yumiriam has quit IRC17:18
*** yumiriam has joined #openstack-manila17:18
gansobswartz, vponomaryov, gouthamr, xyang_, markstur, tbarron: Could you please review https://review.openstack.org/#/c/406305/ and https://review.openstack.org/#/c/345526/ when you have some review time? Thanks in advance17:23
gansocknight: ^17:24
*** porrua has quit IRC17:30
*** lpetrut has quit IRC17:34
*** ociuhandu has joined #openstack-manila17:37
gansobswartz: ping17:37
openstackgerritAlyson proposed openstack/python-manilaclient: Add mountable snapshots support to manila client  https://review.openstack.org/34562517:38
bswartzganso: I am reviewing that one, but my context keeps getting swapped17:45
bswartzganso: that change has been on the top of my list since last week17:46
gansobswartz: you're the PTL, you have context.is_admin :P (bad joke)17:46
gansobswartz: anyway, I have a question to ask17:48
bswartzok17:48
*** ociuhandu has quit IRC17:49
gansobswartz: so, today we migrate shares with "manila migration-start share_x ubuntu@host2#pool2" < driver-assisted-options >17:49
gansobswartz: I tested something yesterday, I did "manila migration-start share_x ubuntu@host2#pool1 < driver-assisted-options >17:50
gansobswartz: pool1 does not exist in host217:50
gansobswartz: the current behavior is that it gets to the scheduler, the scheduler gets a bit confused, but it ultimately says that the share cannot be migrated to ubuntu@host2#pool117:50
bswartzthat sounds good,except for the confused part17:51
gansobswartz: the API layer validates up to ubuntu@host2 part... but to know which pools are members of each host, it needs to talk to the scheduler17:51
bswartzit should be straightforward for the scheduler to determine that the destination is nonexistent17:51
bswartzokay I see17:51
bswartzwell do you think it would be better if the api knew which pools existed?17:52
bswartzor were you proposing some improvement on the scheduler side?17:52
gansobswartz: so unless I am missing something and all other existing methods are doing this ineficiently, we would need to send a synchronous RPC to the scheduler17:52
gansobswartz: in order to provide a response in the API layer17:52
gansobswartz: because, currently, it passes all API layer validations and sends the request to the scheduler, which becomes async and does not post results, the user will have to check for migration_error17:53
bswartzganso: I don't think there are any other cases of the API service looking at a pool name17:53
bswartzso what "other existing methods" are you referring to?17:53
gansobswartz: but sending a synchronous RPC is bad for this case, it is ok for "manila pool-list" because it does only that17:53
gansobswartz: pool-list is an existing method17:54
bswartzbut that doesn't take a pool name input does it?17:54
gansobswartz: no it lists all pools17:54
bswartzright17:54
bswartzso there's no other case where the API input validation needs to validate a pool name17:55
bswartzand API input validation is supposed to be fast17:55
gansobswartz: I am not aware of17:55
*** absubram has quit IRC17:55
gansobswartz: yes17:55
bswartzso ideally the api service would have a fast way of looking up valid pool names17:55
bswartzremind me how we validate host names17:56
gansobswartz: so, option 1) we remain async and do not communicate to the user the seemingly obvious incorrect pool error. Or 2) we make a synchronous call which is bad17:56
gansobswartz: host names are saved in the database17:56
gansobswartz: pool names are not17:56
bswartzhow hard would it be to change that?17:56
gansobswartz: change to what?17:56
bswartzto also make pool names stored in the database similar to host names17:57
bswartzbecause that's my first instinct17:57
gansobswartz: cannot be done in ocata17:57
bswartzfollow an existing pattern that works well17:57
gansobswartz: we would need to design how to store and keep it updated17:57
bswartzwhich table exactly?17:57
gansobswartz: we would need to figure out which table17:58
gansobswartz: or create one17:58
bswartzI mean which table are we pulling the host names out of17:58
bswartzis there only one?17:58
gansobswartz: services17:58
bswartzyeah that's what I feared17:59
gansobswartz: services cannot be mapped to pools 1:117:59
bswartzwell no ofc now17:59
gansobswartz: so we can't just add a column17:59
bswartzwe could add a subtable though17:59
*** timcl has quit IRC18:00
bswartzmainly it's a question of how to keep the table up to date, because pools can technically appear and disappear without warning18:00
gansobswartz: yes18:00
gansobswartz: the scheduler would need to keep it updated on every update_stats call18:00
*** porrua has joined #openstack-manila18:00
gansobswartz: which is how it keeps track of pools18:01
bswartzthat's not the only choice18:01
bswartzthe managers could be forced to do it18:01
*** timcl has joined #openstack-manila18:01
gansobswartz: yea there are other specific points in the code that invoke update_stats besides the periodic_task18:02
vponomaryovganso: what is wrong about doing synchronius call to scheduler?18:02
*** lpetrut has joined #openstack-manila18:02
vponomaryovganso: from your point of view?18:02
gansobswartz: but I still think the update_stats method would be a valid place to update it18:02
bswartzganso: well my specific thought was that the managers could be responsible for updating the pools table directly18:03
gansovponomaryov: it may take a while, services may not be in the same node, or have a long rabbit queue, be unresponsive, and the API layer will take longer to respond, and worse of all: does not scale well18:03
gouthamrwhat's wrong with it failing in the scheduler?18:03
bswartzgouthamr: that's probably what we'll have to do in ocata18:03
vponomaryovganso: not convinced18:04
bswartzbut it's suboptimal18:04
gansogouthamr: it does not communicate the error properly to the user18:04
gansogouthamr: we do not have user messages yet18:04
vponomaryovit is good enough in scope of current architecture18:04
bswartzoptimal is that the information exists in the DB and the api can validate requests without making any RPC call18:04
gouthamrganso: by user you mean administrator, who can see the logs?18:04
gansovponomaryov: API should be fast, and in a HA environment it has enough variables slowing it down18:05
gouthamrbswartz: +1 i agree it's optimal..18:05
gouthamrbut i don't want to slow down the API that we sped up for the very purpose18:05
vponomaryovganso: HA has no diff compared to non-HA setup in current case18:05
gansogouthamr: yes. We are already doing that for "no valid host found" errors, but in that case the input parameter is valid18:05
gouthamrit was a synchronous call before it was a cast..18:05
vponomaryovganso: we are talkign about "migration" API18:06
gouthamrthere's a lot of validation taht we can do in the API if we can get the scheduler info18:06
vponomaryovganso: I, as a user of this API, expect it to be slow enough anyway18:06
gouthamrtakes screenshot ^18:06
gansogouthamr: lol18:07
vponomaryovgouthamr: you will use it on court? ))18:07
*** absubram has joined #openstack-manila18:07
gansovponomaryov: it is mainly API semantics18:07
gansovponomaryov: anything that involves RPC can take up to 60 seconds18:07
gansovponomaryov: ^ synchronous RPCs18:07
gouthamrvponomaryov: nope, it's a good laugh for the boring times http://imgur.com/a/VtJR8 :D18:08
vponomaryovganso: it is not worse than common DB call18:08
gansovponomaryov: APIs that perform validation should be instantenous18:08
gansovponomaryov: DB call already slows things down18:09
vponomaryovganso: DB call is expensive operation too, compared to common python operational time18:09
gansovponomaryov: but the API is already doing loads of DB calls18:10
gansovponomaryov: it cannot be worse than a RPC18:11
vponomaryovganso: that is why call of scheduler is not problem here18:11
gouthamrvponomaryov ganso: any more concerns right away with https://review.openstack.org/#/c/369668 ? I can push up a patch before a slightly long meeting18:11
*** ociuhandu has joined #openstack-manila18:11
gansogouthamr: haven't finished yet, but you can push18:11
gouthamrsure. thanks for reviewing ganso, vponomaryov18:13
*** ociuhandu has quit IRC18:15
*** xinyanzhang has quit IRC18:18
*** xinyanzhang has joined #openstack-manila18:18
*** kambiz has quit IRC18:28
openstackgerritMerged openstack/manila: Fix metadata's soft-delete error when deleting shares  https://review.openstack.org/42117018:29
*** xyang_ has quit IRC18:31
*** david-lyle has quit IRC18:47
openstackgerritRodrigo Barbieri proposed openstack/manila: Implement Revert-to-snapshot in HNAS Driver  https://review.openstack.org/41192318:48
*** porrua has quit IRC18:57
*** porrua has joined #openstack-manila18:58
*** yumiriam has quit IRC19:00
*** xyang_ has joined #openstack-manila19:01
*** tommylikehu has quit IRC19:40
*** dsariel has quit IRC19:49
*** david-lyle has joined #openstack-manila19:50
*** timcl has quit IRC19:55
openstackgerritClinton Knight proposed openstack/manila: Implement share revert to snapshot  https://review.openstack.org/34050220:07
*** timcl has joined #openstack-manila20:13
*** xyang_ has quit IRC20:18
*** xyang_ has joined #openstack-manila20:20
*** dsariel has joined #openstack-manila20:23
*** absubram has quit IRC20:23
*** mtanino has joined #openstack-manila20:28
*** jprovazn has quit IRC20:31
*** lpetrut has quit IRC20:47
*** lpetrut has joined #openstack-manila20:49
*** absubram has joined #openstack-manila20:54
*** porrua has quit IRC21:00
*** alyson_ has quit IRC21:07
*** absubram has quit IRC21:11
*** lpetrut has quit IRC21:13
*** ganso has quit IRC21:14
*** cdelatte has quit IRC21:17
*** xyang_ has quit IRC21:18
*** xyang_ has joined #openstack-manila21:19
*** xyang_ has quit IRC21:36
*** xyang_ has joined #openstack-manila21:37
*** xyang_ has quit IRC21:38
*** xyang_ has joined #openstack-manila21:39
*** xyang_ has quit IRC21:40
*** timcl has quit IRC21:49
*** timcl has joined #openstack-manila21:50
*** cknight has quit IRC21:52
*** gouthamr has quit IRC22:01
*** harlowja has quit IRC22:01
openstackgerritGoutham Pacha Ravi proposed openstack/manila: NetApp cDOT: Add Intra-vserver migration support  https://review.openstack.org/41378322:10
*** dustins has quit IRC22:10
openstackgerritRodrigo Barbieri proposed openstack/manila: Implement Revert-to-snapshot in HNAS Driver  https://review.openstack.org/41192322:17
*** catintheroof has quit IRC22:20
*** xyang_ has joined #openstack-manila22:35
*** xyang_ has quit IRC22:36
*** david-lyle has quit IRC22:51
*** xyang_ has joined #openstack-manila22:57
*** xyang_ has quit IRC22:59
openstackgerritHelen Walsh proposed openstack/manila: VMAX manila plugin - Support for VMAX in Manila  https://review.openstack.org/40485923:00
*** tommylikehu_ has joined #openstack-manila23:11
*** tommylikehu_ has quit IRC23:11
*** tommylik_ has joined #openstack-manila23:11
*** tommylik_ has quit IRC23:12
*** xyang_ has joined #openstack-manila23:24
*** xyang_ has quit IRC23:26
*** gouthamr has joined #openstack-manila23:27
openstackgerritGoutham Pacha Ravi proposed openstack/manila: Refactor Access Rules APIs  https://review.openstack.org/36966823:51
« Monday, 2017-01-16 Index Wednesday, 2017-01-18 »

Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!