Wednesday, 2016-08-10

« Tuesday, 2016-08-09 Index Thursday, 2016-08-11 »
*** yangyapeng has quit IRC00:06
*** gouthamr has joined #openstack-manila00:08
*** mtanino has quit IRC00:23
*** gouthamr_ has joined #openstack-manila00:43
*** gouthamr has quit IRC00:47
openstackgerritGoutham Pacha Ravi proposed openstack/manila: In-tree Install Guide  https://review.openstack.org/31715201:05
*** liyifeng has quit IRC01:06
*** huyang has quit IRC01:13
*** yangyapeng has joined #openstack-manila01:21
*** yangyape_ has joined #openstack-manila02:05
*** yangyapeng has quit IRC02:06
*** liyifeng has joined #openstack-manila02:07
*** cknight has joined #openstack-manila02:16
*** yangyapeng has joined #openstack-manila02:51
*** yangyape_ has quit IRC02:54
*** liyifeng has quit IRC03:19
*** liyifeng has joined #openstack-manila03:20
*** vbellur has joined #openstack-manila03:23
openstackgerritRodrigo Barbieri proposed openstack/manila: Share migration Newton improvements  https://review.openstack.org/32843103:25
openstackgerritMerged openstack/manila: Add support for CIFS shares in HNAS driver  https://review.openstack.org/34296003:30
*** martyt has joined #openstack-manila03:34
*** martyt has quit IRC03:35
*** martyt has joined #openstack-manila03:40
*** martyt has quit IRC03:45
*** shausy has joined #openstack-manila03:55
*** martyt has joined #openstack-manila04:03
*** merooney has quit IRC04:12
*** yangyapeng has quit IRC04:12
*** cknight has quit IRC04:13
*** gouthamr_ has quit IRC04:15
*** martyt has quit IRC04:22
*** martyt has joined #openstack-manila04:23
*** martyt has quit IRC04:25
*** permalac has quit IRC04:25
*** permalac has joined #openstack-manila04:26
*** martyt has joined #openstack-manila04:32
*** martyt has quit IRC04:37
*** pcaruana has quit IRC05:01
*** yangyapeng has joined #openstack-manila05:01
*** martyt has joined #openstack-manila05:04
*** martyt has quit IRC05:04
-openstackstatus- NOTICE: zuul is being restarted to reload configuration. Jobs should be re-enqueued but if you're missing anything (and it's not on http://status.openstack.org/zuul/) please issue a recheck in 30min.05:24
*** akapil has joined #openstack-manila05:37
*** akapil has quit IRC06:08
*** shausy has quit IRC06:15
openstackgerritzhongjun proposed openstack/manila: Add IPv6 ACL support in Manila  https://review.openstack.org/31232106:19
*** nherciu has joined #openstack-manila06:22
*** pgbridge has quit IRC06:36
*** nkrinner_afk is now known as nkrinner06:57
*** lpetrut has joined #openstack-manila07:01
*** akapil has joined #openstack-manila07:19
openstackgerritMarc Koderer proposed openstack/manila: Add MTU information in DB and API  https://review.openstack.org/34950607:21
*** pcaruana has joined #openstack-manila07:24
openstackgerritAlexey Ovchinnikov proposed openstack/manila: [DNM] Tests for scenario tests  https://review.openstack.org/34023607:41
*** gaurangt has joined #openstack-manila07:42
openstackgerritAlexey Ovchinnikov proposed openstack/manila: [DNM] Tests for scenario tests  https://review.openstack.org/34023607:43
*** nkrinner has quit IRC07:43
*** nkrinner has joined #openstack-manila07:46
*** lpetrut has quit IRC07:53
*** nkrinner has quit IRC07:53
*** lpetrut has joined #openstack-manila08:01
*** nkrinner has joined #openstack-manila08:06
*** akapil has quit IRC08:06
*** akapil has joined #openstack-manila08:07
*** liyifeng has quit IRC08:07
*** yangyapeng has quit IRC08:09
*** yangyapeng has joined #openstack-manila08:15
*** liyifeng has joined #openstack-manila08:17
*** lpetrut has quit IRC08:17
*** furlongm has joined #openstack-manila08:18
*** openstackgerrit has quit IRC08:18
*** openstackgerrit has joined #openstack-manila08:19
*** permalac has quit IRC08:40
openstackgerritGaurang Tapase proposed openstack/manila: Refactor GPFS driver for NFS ganesha support  https://review.openstack.org/32882908:55
*** akapil has quit IRC08:56
*** akapil has joined #openstack-manila09:02
*** kaisers has joined #openstack-manila09:15
openstackgerritAlexey Ovchinnikov proposed openstack/manila: [DNM] Tests for scenario tests  https://review.openstack.org/34023609:24
*** lpetrut has joined #openstack-manila09:24
*** kaisers has quit IRC09:31
*** kaisers has joined #openstack-manila09:32
openstackgerritYingzhe Zeng proposed openstack/manila: Implement replication support in huawei driver  https://review.openstack.org/33158609:36
*** akapil_ has joined #openstack-manila09:41
*** akapil has quit IRC09:41
*** wangqun has joined #openstack-manila09:42
openstackgerritzhongjun proposed openstack/manila: Fix share-instance-list command display error  https://review.openstack.org/35336310:09
*** yangyapeng has quit IRC10:15
*** liyifeng has quit IRC10:22
*** liyifeng has joined #openstack-manila10:35
*** mkoderer_ has joined #openstack-manila10:42
*** mkoderer_ has quit IRC10:42
*** permalac has joined #openstack-manila10:53
*** JoseMello has joined #openstack-manila11:01
*** ganso has joined #openstack-manila11:03
*** alyson_ has joined #openstack-manila11:03
*** dsariel has joined #openstack-manila11:14
*** tpsilva has joined #openstack-manila11:21
*** gaurangt has left #openstack-manila11:33
*** wangqun has quit IRC11:34
openstackgerritOpenStack Proposal Bot proposed openstack/manila: Updated from global requirements  https://review.openstack.org/35089511:34
openstackgerritRodrigo Barbieri proposed openstack/manila: Share migration Newton improvements  https://review.openstack.org/32843111:52
openstackgerritRodrigo Barbieri proposed openstack/manila: Fix Share Migration improper behavior for drivers  https://review.openstack.org/33226711:52
gansovponomaryov: ^11:52
gansovponomaryov: oops, I just saw your other comment11:52
gansovponomaryov: :\11:53
vponomaryovganso: I really hope it was last interface change11:54
vponomaryovganso: because I was about to upload ZFSonLinux implementation11:55
gansovponomaryov: have you looked at https://review.openstack.org/328431 ?11:55
vponomaryovno11:55
gansovponomaryov: I highly recommend you rebasing on top of that one11:55
vponomaryovwhy not this one?11:56
vponomaryovganso: anyone will be able to do small changes to ZFS driver such as "rename"11:57
gansovponomaryov: driver interface changes again, and you will need to add a condition for preserve-metadata and writable params. I am not familiar with how replication for ZFS works, but if you are not able to satisfy any of these properties, then you should raise exception.11:57
vponomaryovganso: in case other commit does such things11:57
gansovponomaryov: if your driver is able to satisfy both, then you can ignore them11:57
vponomaryovganso: it does not care about metadata11:58
*** timcl has joined #openstack-manila12:00
openstackgerritValeriy Ponomaryov proposed openstack/manila: [ZFSonLinux] Add share migration support  https://review.openstack.org/35341712:02
vponomaryovganso: also, I need to inform you, that tomorrow is my last working day before vacation and will return back only on 30th of August. It means, if you are interested in having first-party driver implementation of share migration, then you may want to review it at first priority12:04
vponomaryovuploaded without unit tests ^12:04
gansovponomaryov: I will try to take a look today12:04
gansovponomaryov: thanks for heads up12:05
openstackgerritMerged openstack/manila: Fix KeyError on err in unit test  https://review.openstack.org/35277712:06
openstackgerritTiago Pasqualini da Silva proposed openstack/manila: Add Hitachi HSP driver  https://review.openstack.org/32913412:10
openstackgerritAlexey Ovchinnikov proposed openstack/manila: [DNM] Tests for scenario tests  https://review.openstack.org/34023612:22
gansovponomaryov: I highly recommend looking at https://review.openstack.org/#/c/345142/ as well12:25
gansovponomaryov: if that patch merges, driver should be prepared to "retype" share12:25
gansovponomaryov: if driver cannot do it, should raise exception12:25
*** mtanino has joined #openstack-manila12:30
openstackgerritRamana Raja proposed openstack/manila: add access_key to share_access_map  https://review.openstack.org/34330612:35
openstackgerritRamana Raja proposed openstack/manila: add access_key to share_access_map  https://review.openstack.org/34330612:37
*** gaurangt has joined #openstack-manila12:55
*** mtanino has quit IRC12:56
*** bill_az has joined #openstack-manila13:00
*** xyang1 has joined #openstack-manila13:01
*** huyang has joined #openstack-manila13:04
*** gouthamr has joined #openstack-manila13:06
*** gouthamr has quit IRC13:06
*** gouthamr has joined #openstack-manila13:07
*** yangyapeng has joined #openstack-manila13:08
openstackgerritAlexey Ovchinnikov proposed openstack/manila: Fix for LV mounting issue in docker containers  https://review.openstack.org/35346313:10
*** akerr has joined #openstack-manila13:15
*** merooney has joined #openstack-manila13:16
*** porrua has joined #openstack-manila13:18
*** eharney has joined #openstack-manila13:18
*** vbellur has quit IRC13:21
*** harlowja has quit IRC13:23
openstackgerritRodrigo Barbieri proposed openstack/manila: Share migration Newton improvements  https://review.openstack.org/32843113:31
openstackgerritRodrigo Barbieri proposed openstack/manila: Fix Share Migration improper behavior for drivers  https://review.openstack.org/33226713:31
openstackgerritRodrigo Barbieri proposed openstack/manila: Add share type change to Share Migration  https://review.openstack.org/34514213:31
openstackgerritRodrigo Barbieri proposed openstack/manila: Share migration Newton improvements  https://review.openstack.org/32843113:36
openstackgerritRodrigo Barbieri proposed openstack/manila: Add share type change to Share Migration  https://review.openstack.org/34514213:36
*** dustins has joined #openstack-manila13:36
*** huyang has quit IRC13:41
*** dustins has quit IRC13:47
*** xyang_ has joined #openstack-manila13:48
*** dustins has joined #openstack-manila13:48
*** xyang_ has quit IRC13:53
*** xyang_ has joined #openstack-manila13:54
openstackgerritAlexey Ovchinnikov proposed openstack/manila: Container driver  https://review.openstack.org/30893013:56
*** akerr has quit IRC13:59
*** JoseMello has quit IRC14:00
*** akerr has joined #openstack-manila14:01
*** xyang_ has quit IRC14:02
openstackgerritTiago Pasqualini da Silva proposed openstack/manila: Add Hitachi HSP driver  https://review.openstack.org/32913414:03
*** vbellur has joined #openstack-manila14:04
*** xyang_ has joined #openstack-manila14:05
*** dsariel has quit IRC14:06
*** xyang_ has quit IRC14:08
*** xyang_ has joined #openstack-manila14:09
*** catintheroof has joined #openstack-manila14:09
*** liyifeng has quit IRC14:11
*** liyifeng has joined #openstack-manila14:12
*** nherciu has quit IRC14:12
*** gaurangt has quit IRC14:14
openstackgerritValeriy Ponomaryov proposed openstack/manila: [ZFSonLinux] Add share migration support  https://review.openstack.org/35341714:20
vponomaryovganso: here?14:21
gansovponomaryov: about to leave for lunch14:22
vponomaryovganso: don't you think thatk method  "migration_get_driver_info" is redundant if "src" host should get "dst" information in it and return it to its other method?14:22
*** zhongjun_ has quit IRC14:22
gansovponomaryov: not redundant14:22
gansovponomaryov: but mechanism is not ideal14:22
gansovponomaryov: we should not transfer password through RPCs14:22
gansovponomaryov: that's our problem with calling the dest host14:22
vponomaryovganso: why not redundant?14:23
*** zhongjun_ has joined #openstack-manila14:23
gansovponomaryov: because driver should implement its own logic to retrieve what is relevant14:23
vponomaryovganso: when src already has info because exactly src provides it there14:23
gansovponomaryov: netapp driver does that14:23
gansovponomaryov: for replication14:23
vponomaryovganso: netapp driver does what?14:23
vponomaryovganso: I am talking about migration-specific method14:23
vponomaryovganso: concrete method14:24
vponomaryovganso: I do not see the reason for it while it is called only src14:24
gansovponomaryov: for replication, netapp driver reads data from manila.conf that is related to other backends14:24
vponomaryovonly on src*14:24
gansovponomaryov: at midcycle we agreed to do the same for migration14:24
bswartz+114:25
vponomaryovganso: how such approach is related to presence of that method?14:25
vponomaryovganso: replication feature does not have it14:25
gansovponomaryov: only driver can do that14:25
gansovponomaryov: replication does not have an interface for that14:25
gansovponomaryov: but some drivers (such as netapp's) require an approach like this14:26
vponomaryovganso: Rodrigo, I am saying that this method is redundant, and the fact that dst host data is taken from config only proves it14:26
gansovponomaryov: redundant with what?14:26
vponomaryovganso: redundant as useless14:27
gansovponomaryov: 1) you either are saying that the information retrieved is not relevant. or 2) you are saying the information does not need to be retrieved because it is already in hand14:27
vponomaryovganso: driver can get all required info from config in required place14:27
gansovponomaryov: backends may have different passwords, IP addresses14:27
gansovponomaryov: for Hitachi share migration design, this is needed14:27
*** gaurangt has joined #openstack-manila14:27
bswartzevery backend should have the information for all the other backends which they can do driver assisted migration to/from14:28
vponomaryovganso: again, the fact, that driver gets info from config means that such methodis redundant14:28
vponomaryovganso: because this method is called on the same host where its result is used14:28
gansovponomaryov: so you are saying that the driver could do this in migration_start instead of that one?14:29
vponomaryovganso: what this?14:29
gansovponomaryov: are you saying that the driver could obtain the dest information by reading manila.conf in migration_start instead of migration_get_driver_info?14:30
*** mtanino has joined #openstack-manila14:30
vponomaryovganso: read config anywhere it requires some data14:30
vponomaryovganso; yes14:30
vponomaryovganso: so, either call that method on appropriate host or remove it14:30
gansovponomaryov: indeed, yes. But it is like this today because we would like to change it when we can overcome the problem of sending passwords through RPC14:30
vponomaryovganso: "current" approach does not need it14:31
vponomaryovganso: and it may be so for loong time14:31
gansovponomaryov: that method is obtaining that information, and then migration_check_compatibility is using. If I remove this method, you are going to do the same thing in other method, so what is the problem of doing a specific thing in its specific method?14:32
vponomaryovganso: then just provide backend name to specially dedicated method and expect boolean from there14:33
vponomaryovganso: so, driver, reading config, will just provide true/false answer to share manager14:34
gansovponomaryov: that's what migration_check_compatibility does14:34
gansovponomaryov: but if you do so, and if other drivers require that information, they will have to read it again in migration_start, and again in migration_continue14:35
vponomaryovganso: ok, I can just ignore it then14:35
gansovponomaryov: exactly, if you really don't need it, you can just ignore it14:35
gansovponomaryov: if you MUST have the method run in the destination backend, then you would currently have a limitation, which we cannot overcome at this moment14:36
*** cknight has joined #openstack-manila14:37
vponomaryovganso: yes, my vision of it is following: "do storage-related actions only by its driver-owner"14:38
vponomaryovganso: so, each driver should talk to other drivers, not their storages directly14:38
gansovponomaryov: in that case, ultimately all drivers would be performing migration as "copying", similar to fallback14:39
vponomaryovganso: no14:40
vponomaryovganso: why just "copying"?14:41
gansovponomaryov: because the destination backend will have to create an empty share and copy the contents14:41
gansovponomaryov: still, some backend is pushing data somewhere, or pulling data from somewhere, drivers will be talking to each other's storage14:42
vponomaryovganso: push data is only one action, we have lots of other actions14:43
vponomaryovganso: + getting up-to-date information and current state of a driver and its storage14:43
bswartzsorry I was AFK for a few minutes14:43
gansovponomaryov: what could be done is invoke migration_complete in both backends14:44
bswartzare you debating whether to initiate driver assisted migraiton from the source/dest side?14:44
vponomaryovbswartz: ganso says that involvement of both sides will lead to "fallback-like" approach14:44
bswartzit shouldn't matter14:45
gansovponomaryov: I said removal of possibility of one driver talking to other's storage will restrict to "fallback-like" approach14:45
bswartzdriver assisted migration will only ever succeed in the case when the driver knows how to talk to both backends, so whether it's the souce or the dest the result will be the same14:45
vponomaryovganso: you cannot remove such possibility14:45
vponomaryovganso: I am talking about adding possibility talk to dst driver14:45
vponomaryovganso: completely different thing14:46
ganso11:38 <vponomaryov> ganso: so, each driver should talk to other drivers, not their storages directly14:46
bswartzno14:46
vponomaryovganso: for all actions not related to data moving14:46
bswartzthat's not how replication or migration is currently designed14:46
vponomaryovand gettign up-to-date information14:47
bswartzthe current design assumes 1 drivers talks to everything14:47
bswartz1 driver*14:47
vponomaryovwhile it cannot talk to other driver too, it does not talk to everything14:47
*** gaurangt has left #openstack-manila14:47
gansoI did not disregard that possibility, that something like "migration_continue" or "migration_complete" could need to be invoked in dest backend as well14:47
gansobut I never had any real use case for that14:48
gansoif a vendor says it is required because else it will not be possible to do migration, then we can just go and add those calls14:48
bswartzganso: how could that happen?14:49
bswartzthe point of putting all the logic in 1 driver is to give implementors maximum flexibility14:49
vponomaryovI already feeeel that flexibility ))14:49
bswartzthe moment 2 drivers are involved you get less flexibility14:50
gansobswartz: I don't know, I don't have any use case. In Hitachi's share migration design, if the driver reads the configuration parameters related to the dest backend and has connectivity, it can do everything, it does not need to invoke something in the dest backend at all14:50
vponomaryovganso: you use APIs14:50
vponomaryovganso: not CLI14:50
vponomaryovganso: sure it is ok for you14:50
bswartzganso: right, and other vendors might do it all from the dest side and never need to talk to the source -- or some might need to talk to both14:50
gansovponomaryov: if you SSH, you can SSH into the dest storage and run commands there. Isn't what you do in ZFS?14:51
bswartzthere's no design we can pick which works for everyone, so we leave it open-ended14:51
vponomaryovganso: why do you propose implement "get_progress" here - https://review.openstack.org/#/c/353417/2/manila/share/drivers/zfsonlinux/driver.py when it is optional?14:51
vponomaryovganso: moreover it already works without it14:51
gansobswartz: yes, if anyone needs something that is not in the design, then we add. We cannot add everything assuming it will be needed14:51
gansovponomaryov: it is a suggestion for a friendlier message, that's all14:52
vponomaryovganso: ?14:52
vponomaryovganso: message where?14:52
gansovponomaryov: logs, and possibly user messages notification14:53
*** gaurangt has joined #openstack-manila14:53
gansovponomaryov: so it can be distinct from "NotImplementedError"14:53
vponomaryovganso: so, your -1 consists of only one comment - update of access rules?14:53
gansovponomaryov: yes, and update of migration_get_driver_info, which I assume you will remove in the next PS14:54
gansovponomaryov: as you said you would ignore14:54
vponomaryovganso: return empty dict14:54
gansovponomaryov: yup14:54
vponomaryovganso: and about update_access14:54
openstackgerritzzxwill proposed openstack/manila-ui: Correct typo  https://review.openstack.org/35354914:54
vponomaryovganso: you are saying it will update DB - my answer - it is false statement14:55
vponomaryovganso: call of update_access from share manger does not mean update of DB14:55
vponomaryovganso: "driver" does nto update DB states of rules14:55
vponomaryovganso: so, I am saying, that share manager should call update access if it cares about concurrency14:56
vponomaryovganso: via common interface of driver14:56
vponomaryovganso: but here we come to existing problem - how share manager should call dst driver method? ))14:57
gansovponomaryov: so you are proposing a new driver interface that detects if this will be necessary, and manager will invoke update_access from there, then handle restoring? Concurrency is handled through "migrating" status which blocks adding/removing access rules14:57
gansovponomaryov: also, list of access_rules that should exist, at any time when migration is taking place, is always supplied14:58
vponomaryovganso: look, we have interface, and now, we call it locally lots of times, it is not normal14:59
vponomaryovganso: and no need to decide usign separate method whether to apply it or not15:00
vponomaryovganso: just call update access with RO before migration and with original access levels after any migration result15:00
gansovponomaryov: cannot do that if driver can perform migration while writable15:00
vponomaryovganso: ok, I don't mind about that separate method that decides "do we need such call or not"15:02
openstackgerritMerged openstack/manila-ui: Delete openstack/common from the exclude list of flake8 in tox  https://review.openstack.org/35107315:03
gansovponomaryov: is concurrency the only problem with current approach?15:03
vponomaryovganso: why the only? ))15:03
openstackgerritMerged openstack/manila-ui: Update Django 1.10 tox env  https://review.openstack.org/35153215:03
gansovponomaryov: essentially, migration_start has all the tools it needs to do exactly what update_access does15:04
vponomaryovganso: it means make driver call its public interfaces from its other public interfaces15:04
vponomaryovganso: which means we have incorrect design because share manager is expected to call driver interfaces15:05
vponomaryovganso: at least, it is overhead15:05
gansovponomaryov: then it is incorrect when driver calls its own "create_share" when "creating a share from snapshot" as part of its implementation?15:05
vponomaryovganso: we do not make driver do it15:06
vponomaryovganso: i tis different15:06
vponomaryovganso: while it can implement either way, it is ok15:06
gansovponomaryov: ok15:07
gansovponomaryov: new driver interface for you this afternoon :)15:07
gansovponomaryov: migration_check_writable15:07
vponomaryovganso: you have 10 minutes ))15:07
vponomaryovganso: ^_^15:07
bswartzganso: you know that vponomaryov is going on vacation starting friday15:07
gansovponomaryov: you can see the implementation tomorrow15:08
bswartzso whatever isn't done w/ ZFS driver by then will fall on you and me to finish15:08
gansobswartz: ya he told me15:08
*** akapil_ has quit IRC15:12
*** david-lyle has quit IRC15:12
*** akapil has joined #openstack-manila15:12
*** david-lyle has joined #openstack-manila15:12
*** pgbridge has joined #openstack-manila15:20
*** vbellur has quit IRC15:22
*** akapil has quit IRC15:23
*** akapil has joined #openstack-manila15:23
*** pgbridge has quit IRC15:31
openstackgerritTiago Pasqualini da Silva proposed openstack/manila: Add Hitachi HSP driver  https://review.openstack.org/32913415:33
*** nkrinner is now known as nkrinner_afk15:34
*** vbellur has joined #openstack-manila15:37
*** pgbridge has joined #openstack-manila15:46
*** bill_az_ has joined #openstack-manila15:46
tpsilvacknight: ping15:51
cknighttpsilva: Hi, Tiago.15:51
tpsilvacknight: Hello Clinton, thank you for the review15:51
tpsilvacknight: could you check my responses before I upload another PS?15:51
openstackgerritMerged openstack/manila: Updated from global requirements  https://review.openstack.org/35089515:52
tpsilvacknight: just want to check if you agree with them or you suggest any changes... not sure I like the current implementation, I am open to suggestions :)15:52
cknighttpsilva: sure, I'm trying to keep up with you!15:53
tpsilvacknight: the GET methods are a little weird because this backend API is not fully compliant with REST patterns15:54
cknighttpsilva: OK, I'm not too concerned about weirdness, all platforms have some quirks.  The trick is to encapsulate all of that in your REST layer so the driver doesn't have to care.15:55
tpsilvacknight: yep15:55
tpsilvacknight: and about the provisioned_capacity_gb, I'm lost on what I should report there15:56
tpsilvacknight: Xing requested me to report that, so I implemented that private storage logic to get this info15:57
cknighttpsilva: According to Goutham, if a driver doesn't report it, the scheduler should be able to calculate it for you.  If you can efficiently query a source of truth (the backend) to get the number, that's great.  But trying to maintain a value incrementally over time is guaranteed to fail, so let's not do that.15:58
*** adrianofr_ has joined #openstack-manila15:58
cknighttpsilva: I'm not certain the scheduler has that logic yet (Cinder does), but that's the plan.15:59
tpsilvacknight: Ok, so aside from that, everything else is ok?16:02
tpsilvacknight: I am going to remove that logic and stop reporting provisioned_capacity_gb then16:02
cknighttpsilva: Give me a few to recheck your latest…16:02
tpsilvacknight: Ok, thanks16:02
*** timcl has quit IRC16:08
*** baojg has joined #openstack-manila16:10
*** nherciu has joined #openstack-manila16:11
gansobswartz, vponomaryov: so what are we doing about https://review.openstack.org/#/c/350647 ?16:17
vponomaryovganso: http://paste.openstack.org/show/553616/ - it is with latest PS of https://review.openstack.org/#/c/33226716:17
vponomaryovganso: something wrong is provided instead of dest_host16:18
vponomaryovganso: then error raises and manager hangs again trying to "complete"16:18
gansovponomaryov: docstring says dest_host contains capabilities as well16:19
gansovponomaryov: so you need dest_host['host']16:19
vponomaryovganso: you should have been a lawyer16:25
gansovponomaryov: I would be a terrible lawyer16:26
vponomaryovganso: best logic encryptor ever, you would win each court16:26
vponomaryovganso: ok, what about calling "complete" after error?16:27
vponomaryovganso what docstring I should read about it? ))16:27
gansovponomaryov: I tried to reproduce it yesterday, I could not16:28
*** pcaruana has quit IRC16:29
gansovponomaryov: https://review.openstack.org/#/c/332267/46/manila/share/manager.py16:30
gansovponomaryov: LN 727, if exception is raised, it goes to 73716:30
gansovponomaryov: same for LN 76016:31
gansovponomaryov: I am trying to understand from where that RPC is coming from16:31
vponomaryovganso: Rodrigo, error was raised by migration_get_driver_info16:32
vponomaryovganso: and it is in different place than you pointed out16:33
gansovponomaryov: ok. Still correct16:35
gansovponomaryov: it seems to me it is invoking fallback when your driver errors16:35
gansovponomaryov: then it was about to complete, but complete has a bug which I am fixing now16:36
*** timcl has joined #openstack-manila16:36
gansovponomaryov: so, complete being invoked in not incorrect16:36
vponomaryovganso: really? )))16:37
*** akapil has quit IRC16:43
*** akapil has joined #openstack-manila16:48
*** gaurangt has left #openstack-manila16:54
vponomaryovganso: second trace in http://paste.openstack.org/show/553616/ is raised in any case16:54
vponomaryovganso: even when first phase was completed successfully16:55
*** dsariel has joined #openstack-manila16:56
gansovponomaryov: that is also correct... tempest test invokes complete after 1st phase16:57
*** dsariel has quit IRC16:58
vponomaryovganso: "TypeError: migration_complete() got an unexpected keyword argument 'new_share_instance_id'" is expected? =)16:58
vponomaryovganso: I am talking about improper args16:58
gansovponomaryov: it is not expected. That is the bug I am fixing, coming up in a few minutes16:59
gansovponomaryov: what is not a bug is migration-complete being invoked16:59
openstackgerritValeriy Ponomaryov proposed openstack/manila: [ZFSonLinux] Add share migration support  https://review.openstack.org/35341716:59
vponomaryovganso: I find it strange that you "clarify" it16:59
vponomaryovganso: above version of ZFSonLinux driver does use get_driver_info because of approach for making decision17:00
vponomaryovganso: so, no lets keep it as is for now17:01
*** baojg has quit IRC17:03
vponomaryovganso: so, if it passes, then left only to update unit tests. Hope, will be able to finish them tomorrow17:04
vponomaryovganso: concurrency of update access and get progress I leave for future17:05
*** tbarron is now known as tbarron|lunch17:09
openstackgerritTiago Pasqualini da Silva proposed openstack/manila: Add Hitachi HSP driver  https://review.openstack.org/32913417:18
*** akapil has quit IRC17:23
*** akapil has joined #openstack-manila17:25
*** harlowja has joined #openstack-manila17:27
*** akapil has quit IRC17:29
*** yangyapeng has quit IRC17:38
tpsilvagouthamr: ping17:41
openstackgerritAlexey Ovchinnikov proposed openstack/manila: Container driver  https://review.openstack.org/30893017:57
*** lpetrut has quit IRC18:07
openstackgerritAlexey Ovchinnikov proposed openstack/manila: Fix for LV mounting issue in docker containers  https://review.openstack.org/35346318:22
openstackgerritAlexey Ovchinnikov proposed openstack/manila: [DNM] Tests for scenario tests  https://review.openstack.org/34023618:23
cknighttpsilva: Looks pretty good.  I still think that with negligible effort you can improve your exception handling in the REST interface.18:28
*** catintheroof has quit IRC18:34
gouthamrtpsilva: pong18:34
*** xyang_ has quit IRC18:35
tpsilvagouthamr: hi, it was about that HNAS driver fix18:35
gouthamrtpsilva: which one, sorry was afk..18:35
tpsilvagouthamr: I commented on the patch, we had a similar issue with thin_provisioning on our driver...18:35
tpsilvagouthamr: https://review.openstack.org/#/c/352400/18:35
gouthamrtpsilva: yes.. in that case, you did not support thick provisioning18:36
gouthamrtpsilva: you do not*18:36
tpsilvagouthamr: yes18:36
tpsilvagouthamr: and dedupe is the same thing18:36
tpsilvagouthamr: dedupe might be enabled on the filesystem that manila is using, but the driver is always reporting it as False18:36
gouthamrtpsilva: so, if you weren't reporting thin_provisioning=True to the scheduler, the scheduler wouldn't be performing all that inflation and max_over_subscription_ratio calculation..18:37
gouthamrtpsilva: sigh, thin line... it feels like a feature.. because if the users were never using that share type, they were getting dedupe with or without this fix.18:38
openstackgerritRamana Raja proposed openstack/manila: cephfs_native: add read-only share support  https://review.openstack.org/33401018:38
gouthamrtpsilva: so if the users can begin using the extra_spec now... it's a feature.. :)18:39
tpsilvagouthamr: I'm failing to understand why the thin_provisioning issue was a bugfix and this is not, but if it's just a matter of closing the bug and issuing a new BP, I'm ok with changing that18:44
openstackgerritRodrigo Barbieri proposed openstack/manila: Share migration Newton improvements  https://review.openstack.org/32843118:46
openstackgerritRodrigo Barbieri proposed openstack/manila: Fix Share Migration improper behavior for drivers  https://review.openstack.org/33226718:46
*** xyang_ has joined #openstack-manila18:46
gansogouthamr: we already have customers that enabled dedupe in backend, they created a  share type to map to that backend through backend_name18:48
gansogouthamr: so the feature was already there, but improperly mapped to the extra_specs18:48
gansogouthamr: since the customers are already using the feature, we intend to backport18:49
gansotpsilva: it will not be possible to backport if it is implemented as feature18:50
gouthamrtpsilva: thin_provisioning is a special capability to the scheduler.. we do some math around it in the scheduler.. so not reporting thin_provisioning was definitely a bug18:50
gouthamrtpsilva ganso: lets think hard and see if this will break anything if backported.. bswartz says it couldn't, because it is an extra-spec and a capability unlike anything that the scheduler cares about.. new API calls to the backend or something, and this would have been a feature.. so, thin line. :)18:51
*** tbarron|lunch is now known as tbarron18:52
bswartzganso: it sounds to me like it lies in the gray area between feature and bug -- but I could get on board with treating this as a bug18:52
gansobswartz: have you taken a look at the code?18:53
gouthamrtpsilva: can you yank out the prelude18:53
gouthamrtpsilva: and change the release note.. to explain what was done rather than what was wrong?18:54
bswartzganso: no18:54
tpsilvagouthamr: ok... actually it's not my patch :) but I'll tell alyson_ that18:55
gansobswartz: https://review.openstack.org/#/c/35240018:56
*** lpetrut has joined #openstack-manila18:56
*** dsariel has joined #openstack-manila18:56
*** xyang_ has quit IRC18:57
*** akapil has joined #openstack-manila19:00
*** xyang_ has joined #openstack-manila19:08
*** xyang_ has quit IRC19:11
openstackgerritMerged openstack/manila-specs: Add spec for storing auth secrets in access map  https://review.openstack.org/32297119:18
openstackgerritRodrigo Barbieri proposed openstack/manila: Add share type change to Share Migration  https://review.openstack.org/34514219:21
*** xyang_ has joined #openstack-manila19:23
*** mtanino has quit IRC19:34
*** bill_az_ has quit IRC19:47
*** JoseMello has joined #openstack-manila19:49
*** xyang_ has quit IRC19:54
*** xyang_ has joined #openstack-manila19:55
*** bill_az_ has joined #openstack-manila19:58
*** xyang_ has quit IRC20:00
*** martyt has joined #openstack-manila20:05
*** akerr has quit IRC20:09
*** porrua has quit IRC20:10
*** merooney has quit IRC20:16
*** timcl has quit IRC20:24
openstackgerritAlex Meade proposed openstack/manila: Manila Share Groups  https://review.openstack.org/33509320:32
openstackgerritAlex Meade proposed openstack/manila: Manila Share Groups  https://review.openstack.org/33509320:32
*** yangyapeng has joined #openstack-manila20:39
*** akapil has quit IRC20:40
*** yangyapeng has quit IRC20:43
*** xyang_ has joined #openstack-manila20:52
*** cknight has quit IRC20:57
*** martyt has quit IRC21:02
*** martyt has joined #openstack-manila21:06
*** xyang_ has quit IRC21:17
*** tpsilva has quit IRC21:18
*** vbellur has quit IRC21:20
*** xyang_ has joined #openstack-manila21:20
*** akapil has joined #openstack-manila21:40
*** akapil has quit IRC21:45
*** gouthamr has quit IRC22:00
*** gouthamr has joined #openstack-manila22:02
*** xyang1 has quit IRC22:06
*** bill_az has quit IRC22:08
*** nherciu has quit IRC22:09
*** JoseMello has quit IRC22:20
*** dustins has quit IRC22:20
*** lpetrut has quit IRC22:21
*** eharney has quit IRC22:23
*** martyt has quit IRC22:41
*** vbellur has joined #openstack-manila22:47
*** xyang_ has quit IRC22:50
*** akapil has joined #openstack-manila22:55
*** akapil has quit IRC23:00
*** merooney has joined #openstack-manila23:04
*** vbellur has quit IRC23:06
openstackgerritAlex Meade proposed openstack/manila: Manila Share Groups  https://review.openstack.org/33509323:08
openstackgerritAlex Meade proposed openstack/python-manilaclient: Add share group support to Manila client  https://review.openstack.org/33512023:09
*** ganso has quit IRC23:21
*** chlong|mtg has quit IRC23:31
*** ksumit has joined #openstack-manila23:38
« Tuesday, 2016-08-09 Index Thursday, 2016-08-11 »

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