| *** baojg has quit IRC | 00:01 | |
| *** yangyapeng has quit IRC | 00:03 | |
| *** gouthamr has joined #openstack-manila | 00:22 | |
| *** openstack has joined #openstack-manila | 00:26 | |
| *** gouthamr has quit IRC | 00:27 | |
| *** mtanino has quit IRC | 00:28 | |
| openstackgerrit | daiki kato proposed openstack/python-manilaclient: Fix typos https://review.openstack.org/293182 | 00:29 |
|---|---|---|
| openstackgerrit | Goutham Pacha Ravi proposed openstack/manila: Add tests to ensure snapshots across replicas https://review.openstack.org/293113 | 00:31 |
| *** erlon has quit IRC | 00:37 | |
| *** eharney has quit IRC | 00:37 | |
| *** docaedo has quit IRC | 00:55 | |
| *** docaedo has joined #openstack-manila | 01:01 | |
| *** david-lyle has joined #openstack-manila | 01:04 | |
| *** yangyapeng has joined #openstack-manila | 01:10 | |
| *** Yogi1_ has joined #openstack-manila | 01:12 | |
| *** yangyapeng has quit IRC | 01:22 | |
| *** gouthamr_ is now known as gouthamr | 01:23 | |
| *** yangyapeng has joined #openstack-manila | 01:24 | |
| *** docaedo has quit IRC | 01:24 | |
| *** cknight has joined #openstack-manila | 01:29 | |
| *** docaedo has joined #openstack-manila | 01:31 | |
| *** akshai has quit IRC | 01:33 | |
| *** esker has joined #openstack-manila | 01:37 | |
| *** kaisers1 has joined #openstack-manila | 01:44 | |
| *** kaisers has quit IRC | 01:45 | |
| *** baojg has joined #openstack-manila | 01:59 | |
| *** akshai has joined #openstack-manila | 02:05 | |
| *** akshai_ has joined #openstack-manila | 02:07 | |
| *** Yogi1_ has quit IRC | 02:09 | |
| *** Yogi1 has joined #openstack-manila | 02:10 | |
| *** akshai has quit IRC | 02:10 | |
| *** baojg has quit IRC | 02:17 | |
| *** ganso has quit IRC | 02:19 | |
| *** baojg has joined #openstack-manila | 02:20 | |
| *** amrith is now known as _amrith_ | 02:24 | |
| openstackgerrit | Merged openstack/manila: HDS-HNAS: fixed exception when export not found https://review.openstack.org/291366 | 02:31 |
| *** houming has joined #openstack-manila | 02:36 | |
| *** Yogi1 has quit IRC | 02:54 | |
| *** gouthamr has quit IRC | 03:06 | |
| *** baojg has quit IRC | 03:08 | |
| *** baojg has joined #openstack-manila | 03:10 | |
| *** esker has quit IRC | 03:11 | |
| *** houming has quit IRC | 03:14 | |
| *** houming_ has joined #openstack-manila | 03:14 | |
| *** baojg has quit IRC | 03:42 | |
| *** baojg has joined #openstack-manila | 03:43 | |
| *** akshai_ has quit IRC | 03:52 | |
| openstackgerrit | Goutham Pacha Ravi proposed openstack/manila: NetApp cDOT: Handle replicated snapshots https://review.openstack.org/293243 | 03:59 |
| *** esker has joined #openstack-manila | 04:01 | |
| *** cknight1 has joined #openstack-manila | 04:03 | |
| *** cknight has quit IRC | 04:04 | |
| *** esker has quit IRC | 04:06 | |
| *** akshai has joined #openstack-manila | 04:36 | |
| *** cknight1 has quit IRC | 04:38 | |
| *** baojg has quit IRC | 04:45 | |
| openstackgerrit | zhongjun proposed openstack/manila: Fix the scheduler choose a disable share service https://review.openstack.org/293246 | 04:54 |
| *** sgotliv has joined #openstack-manila | 04:57 | |
| *** sgotliv has quit IRC | 05:03 | |
| zhongjun | bswartz: ping, Could you please tag this bug(#1557854) to RC1? Thanks | 05:04 |
| *** akshai_ has joined #openstack-manila | 05:04 | |
| *** akshai has quit IRC | 05:07 | |
| openstackgerrit | Merged openstack/manila: Fix delete when share not found in update_access https://review.openstack.org/291279 | 05:29 |
| openstackgerrit | Merged openstack/manila: glusterfs volume layout: take care of deletion of DOA shares https://review.openstack.org/292177 | 05:29 |
| *** sgotliv has joined #openstack-manila | 05:30 | |
| *** esker has joined #openstack-manila | 05:33 | |
| *** esker has quit IRC | 05:38 | |
| openstackgerrit | Merged openstack/manila: update quota of origin user on share extend/shrink https://review.openstack.org/291725 | 05:38 |
| *** sgotliv has quit IRC | 05:40 | |
| *** akshai_ has quit IRC | 05:44 | |
| *** david-lyle has quit IRC | 05:45 | |
| *** david-lyle has joined #openstack-manila | 05:45 | |
| *** yangyape_ has joined #openstack-manila | 05:56 | |
| *** yangyapeng has quit IRC | 06:00 | |
| *** yangyapeng has joined #openstack-manila | 06:07 | |
| *** yangyape_ has quit IRC | 06:10 | |
| openstackgerrit | zhongjun proposed openstack/manila: Fix the scheduler choose a disable share service https://review.openstack.org/293246 | 06:18 |
| *** sgotliv has joined #openstack-manila | 06:18 | |
| *** esker has joined #openstack-manila | 06:49 | |
| *** esker has quit IRC | 06:54 | |
| openstackgerrit | Thomas Bechtold proposed openstack/manila: Use sslutils from oslo_service https://review.openstack.org/289382 | 07:01 |
| *** lpetrut has joined #openstack-manila | 07:15 | |
| *** nkrinner has joined #openstack-manila | 07:17 | |
| *** rraja has joined #openstack-manila | 07:21 | |
| *** esker has joined #openstack-manila | 07:23 | |
| *** esker has quit IRC | 07:28 | |
| *** darkwsh has joined #openstack-manila | 07:33 | |
| *** a_ta has joined #openstack-manila | 07:33 | |
| *** a_ta has quit IRC | 07:38 | |
| *** sgotliv has quit IRC | 07:38 | |
| *** sgotliv has joined #openstack-manila | 07:42 | |
| *** chlong has quit IRC | 07:42 | |
| openstackgerrit | Valeriy Ponomaryov proposed openstack/manila: Fix update of access rules in ZFSonLinux driver https://review.openstack.org/293289 | 07:44 |
| openstackgerrit | Valeriy Ponomaryov proposed openstack/manila: Fix update of access rules in ZFSonLinux driver https://review.openstack.org/293289 | 07:46 |
| *** sgotliv has quit IRC | 07:55 | |
| *** lpetrut has quit IRC | 08:04 | |
| openstackgerrit | daiki kato proposed openstack/manila: Fix typos https://review.openstack.org/293302 | 08:17 |
| *** yangyapeng has quit IRC | 08:23 | |
| *** yangyapeng has joined #openstack-manila | 08:26 | |
| *** ig0r_ has quit IRC | 08:39 | |
| *** dalgaaf has joined #openstack-manila | 08:55 | |
| *** yangyapeng has quit IRC | 08:56 | |
| *** esker has joined #openstack-manila | 09:12 | |
| *** esker has quit IRC | 09:16 | |
| *** sgotliv has joined #openstack-manila | 09:19 | |
| *** esker has joined #openstack-manila | 09:32 | |
| *** esker has quit IRC | 09:39 | |
| openstackgerrit | Tom Patzig proposed openstack/manila: service instance: also recognize instance name https://review.openstack.org/269777 | 09:43 |
| *** openstackgerrit has quit IRC | 09:53 | |
| *** openstackgerrit_ is now known as openstackgerrit | 09:53 | |
| *** openstackgerrit has quit IRC | 09:53 | |
| *** openstackgerrit_ has joined #openstack-manila | 09:54 | |
| *** openstackgerrit_ is now known as openstackgerrit | 09:54 | |
| *** openstackgerrit_ has joined #openstack-manila | 09:55 | |
| *** openstackgerrit has quit IRC | 09:55 | |
| *** openstackgerrit has joined #openstack-manila | 09:56 | |
| *** tpsilva has joined #openstack-manila | 10:07 | |
| openstackgerrit | Marc Koderer proposed openstack/manila: Fix keystone v3 issues for all clients https://review.openstack.org/290511 | 10:09 |
| *** lpetrut has joined #openstack-manila | 10:09 | |
| *** yangyapeng has joined #openstack-manila | 10:23 | |
| mkoderer___ | tpatzig: dgonzalez: ready for review/test https://review.openstack.org/#/c/290511/10 | 10:23 |
| *** yangyapeng has quit IRC | 10:23 | |
| openstackgerrit | Marc Koderer proposed openstack/manila: Fix keystone v3 issues for all clients https://review.openstack.org/290511 | 10:31 |
| *** a_ta has joined #openstack-manila | 10:35 | |
| *** a_ta has quit IRC | 10:40 | |
| *** ociuhandu has quit IRC | 10:55 | |
| tpsilva | vponomaryov: sorry about that :) | 10:56 |
| openstackgerrit | Tiago Pasqualini da Silva proposed openstack/manila: Fix Hitachi HNAS driver version https://review.openstack.org/293175 | 11:03 |
| *** esker has joined #openstack-manila | 11:04 | |
| vponomaryov | tpsilva: ? | 11:06 |
| tpsilva | vponomaryov: breaking the zfs driver | 11:06 |
| vponomaryov | tpsilva: I +2'ed it personally )) | 11:06 |
| vponomaryov | tpsilva: >< | 11:06 |
| tpsilva | vponomaryov: it happens :) | 11:06 |
| *** esker has quit IRC | 11:08 | |
| *** merooney has joined #openstack-manila | 11:12 | |
| openstackgerrit | Valeriy Ponomaryov proposed openstack/manila: Make ZFSonLinux driver handle snapshots of replicated shares properly https://review.openstack.org/293377 | 11:15 |
| *** _amrith_ is now known as amrith | 11:16 | |
| *** merooney has quit IRC | 11:17 | |
| *** openstackgerrit has quit IRC | 11:18 | |
| *** openstackgerrit has joined #openstack-manila | 11:18 | |
| openstackgerrit | Valeriy Ponomaryov proposed openstack/manila: Make ZFSonLinux driver handle snapshots of replicated shares properly https://review.openstack.org/293377 | 11:21 |
| *** merooney has joined #openstack-manila | 11:22 | |
| *** ganso has joined #openstack-manila | 11:31 | |
| *** amrith has left #openstack-manila | 11:32 | |
| *** erlon has joined #openstack-manila | 11:44 | |
| openstackgerrit | Tom Patzig proposed openstack/manila: service instance: also recognize instance name https://review.openstack.org/269777 | 11:48 |
| openstackgerrit | Alexey Ovchinnikov proposed openstack/manila-image-elements: Fix for image name https://review.openstack.org/293385 | 11:49 |
| openstackgerrit | Marc Koderer proposed openstack/manila: Fix keystone v3 issues for all clients https://review.openstack.org/290511 | 11:52 |
| *** yangyapeng has joined #openstack-manila | 11:56 | |
| *** merooney has quit IRC | 12:00 | |
| openstackgerrit | Alexey Ovchinnikov proposed openstack/manila: LXC/LXD: Fix for incorrect user handling bug https://review.openstack.org/292871 | 12:03 |
| *** merooney has joined #openstack-manila | 12:06 | |
| openstackgerrit | Tiago Pasqualini da Silva proposed openstack/python-manilaclient: Reverted fix access rules functional tests https://review.openstack.org/293392 | 12:11 |
| openstackgerrit | Tiago Pasqualini da Silva proposed openstack/python-manilaclient: Revert fix access rules functional tests https://review.openstack.org/293392 | 12:12 |
| *** aovchinnikov has joined #openstack-manila | 12:13 | |
| *** a_ta has joined #openstack-manila | 12:15 | |
| openstackgerrit | Tiago Pasqualini da Silva proposed openstack/manila: Fix Hitachi HNAS driver version https://review.openstack.org/293175 | 12:19 |
| openstackgerrit | Csaba Henk proposed openstack/manila: glusterfs: Implement update_access() method https://review.openstack.org/291151 | 12:20 |
| openstackgerrit | Csaba Henk proposed openstack/manila: ganesha: implement update_access https://review.openstack.org/282602 | 12:20 |
| openstackgerrit | Tiago Pasqualini da Silva proposed openstack/manila: Fix Hitachi HNAS driver version https://review.openstack.org/293175 | 12:21 |
| *** a_ta has quit IRC | 12:22 | |
| *** ociuhandu has joined #openstack-manila | 12:25 | |
| *** JoseMello has joined #openstack-manila | 12:27 | |
| *** pcaruana has quit IRC | 12:28 | |
| *** akshai has joined #openstack-manila | 12:29 | |
| *** gouthamr has joined #openstack-manila | 12:33 | |
| *** timcl has joined #openstack-manila | 12:39 | |
| *** merooney has quit IRC | 12:45 | |
| *** cknight has joined #openstack-manila | 12:49 | |
| *** merooney has joined #openstack-manila | 12:51 | |
| *** csaba has quit IRC | 12:52 | |
| gouthamr | vponomaryov: ping | 12:56 |
| vponomaryov | gouthamr: pong | 12:56 |
| *** csaba has joined #openstack-manila | 12:56 | |
| gouthamr | hi vponomaryov! the cDOT fix for the snapshot instances bug is here: https://review.openstack.org/#/c/293243/1 | 12:56 |
| *** eharney has joined #openstack-manila | 12:57 | |
| gouthamr | vponomaryov: i feel like the driver i/fs should be as open as possible for all possible implementations.. given that we don't want the driver to ask the database for these objects.. | 12:58 |
| vponomaryov | i/fs? | 12:58 |
| gouthamr | vponomaryov: interfaces | 12:58 |
| gouthamr | vponomaryov: would appreciate your review on that when you have time. | 12:58 |
| vponomaryov | gouthamr: when ZFS part is finished, and now I have there problem | 12:59 |
| vponomaryov | gouthamr: would be great to have list of snapshot instances for "create_replica" method | 12:59 |
| gouthamr | vponomaryov: so that you would be able to send the snapshots immediately? | 13:00 |
| vponomaryov | gouthamr: kind of | 13:01 |
| gouthamr | vponomaryov: we could do that.. what do you think of the amount of information being handed down? :P between ameade and me, the joke is that all of these are becoming sqldumps. | 13:01 |
| gouthamr | vponomaryov: i'm not sure what the best approach is here. | 13:02 |
| ameade | +1 | 13:03 |
| *** kaisers1 has quit IRC | 13:09 | |
| *** kro_sinus_hell is now known as krot_sinus_purga | 13:15 | |
| *** krot_sinus_purga is now known as kro_sniffles | 13:15 | |
| *** smcginnis_away is now known as smcginnis | 13:15 | |
| *** merooney has quit IRC | 13:17 | |
| *** kaisers has joined #openstack-manila | 13:19 | |
| *** xyang1 has joined #openstack-manila | 13:19 | |
| mkoderer___ | gouthamr: vponomaryov: hi, the keystone v3 patch should be fine now https://review.openstack.org/#/c/290511/ | 13:20 |
| mkoderer___ | feedback is welcome :) | 13:20 |
| gouthamr | thanks mkoderer___.. will take a looj | 13:22 |
| gouthamr | look* | 13:22 |
| mkoderer___ | gouthamr: btw I tried to run netapp-ci but it seems to ignore me | 13:22 |
| gouthamr | mkoderer___: i saw the job running on the netapp-ci.. it's running :) | 13:23 |
| mkoderer___ | gouthamr: :) ok fine | 13:23 |
| gouthamr | mkoderer___: will vote in roughly 39 minutes | 13:23 |
| *** ig0r_ has joined #openstack-manila | 13:24 | |
| mkoderer___ | gouthamr: ah it already commented to the patch .. fine fine | 13:25 |
| *** merooney has joined #openstack-manila | 13:25 | |
| gouthamr | vponomaryov: create replica only takes model update for the replica.. if we pass snapshot instances through, it would need to take snapshot instance updates too? | 13:25 |
| *** kaisers has quit IRC | 13:26 | |
| vponomaryov | gouthamr: creating replica, I create all source snapshots too | 13:27 |
| vponomaryov | gouthamr: so, I would say yes, we should handles snapshot instances too | 13:27 |
| gouthamr | vponomaryov: in the netapp implementation ameade used the provider location to name snapshots so that all snapshots used the same "name" across replicas. that way, if a new unknown snapshot is being queried in the update_replicated_snapshot call, we just correlate the provider location from any of the other snapshot instances and query for its availability, or initiate an update of the replica so that teh snapshot makes its way and is updated | 13:28 |
| gouthamr | on the next update_replicated_snapshot call | 13:28 |
| *** kaisers has joined #openstack-manila | 13:28 | |
| *** martyturner has joined #openstack-manila | 13:28 | |
| vponomaryov | gouthamr: ok, I will think more and try to workaround | 13:28 |
| ameade | this could be something we change in newton | 13:29 |
| gouthamr | vponomaryov: sure .. thank you. | 13:29 |
| gouthamr | ameade: yes, cknight wanted to know how we can improve the interfaces for these things.. because its a similar problem we may have when we have the generic groups he was proposing | 13:30 |
| ameade | true story | 13:30 |
| gouthamr | haha yes, so lets do it in (fig) newton | 13:30 |
| * gouthamr thinks this would need wider participation and discussion among other driver developers.. we should speak about it in Austin | 13:31 | |
| *** dustins has joined #openstack-manila | 13:37 | |
| *** porrua has joined #openstack-manila | 13:46 | |
| openstackgerrit | Merged openstack/manila: Fix update of access rules in ZFSonLinux driver https://review.openstack.org/293289 | 13:48 |
| openstackgerrit | Merged openstack/manila: service instance: also recognize instance name https://review.openstack.org/269777 | 13:49 |
| *** akerr has joined #openstack-manila | 13:52 | |
| *** merooney has quit IRC | 13:52 | |
| *** mtanino has joined #openstack-manila | 13:54 | |
| *** houming_ has quit IRC | 13:56 | |
| *** martyturner has quit IRC | 13:57 | |
| *** merooney has joined #openstack-manila | 13:58 | |
| *** merooney has quit IRC | 13:58 | |
| *** merooney has joined #openstack-manila | 13:58 | |
| *** a_ta has joined #openstack-manila | 14:00 | |
| *** Yogi1 has joined #openstack-manila | 14:01 | |
| *** martyturner has joined #openstack-manila | 14:01 | |
| openstackgerrit | Dustin Schoenbrun proposed openstack/manila: Add common capabilities matrix to devref https://review.openstack.org/288504 | 14:02 |
| openstackgerrit | Alex Meade proposed openstack/manila: NetApp cDOT: Fix status updates for replicas https://review.openstack.org/291803 | 14:03 |
| openstackgerrit | Yogesh proposed openstack/manila: Add tests to ensure snapshots across replicas https://review.openstack.org/293113 | 14:03 |
| *** timcl has quit IRC | 14:04 | |
| *** timcl has joined #openstack-manila | 14:08 | |
| *** absubram has joined #openstack-manila | 14:11 | |
| *** absubram_ has joined #openstack-manila | 14:12 | |
| *** absubram has quit IRC | 14:15 | |
| *** absubram_ is now known as absubram | 14:15 | |
| *** baojg has joined #openstack-manila | 14:31 | |
| openstackgerrit | Merged openstack/manila: Fix HNAS snapshot creation on deleted shares https://review.openstack.org/291806 | 14:38 |
| *** absubram has quit IRC | 14:40 | |
| *** absubram has joined #openstack-manila | 14:43 | |
| *** merooney has quit IRC | 14:51 | |
| openstackgerrit | Rodrigo Barbieri proposed openstack/manila: HDS-HNAS: Fix exception in update_access not found https://review.openstack.org/291358 | 14:51 |
| mkoderer___ | bswartz: https://bugs.launchpad.net/bugs/1555554 I have a fix for it.. and I would keep the milestone on it | 14:56 |
| openstack | Launchpad bug 1555554 in Manila "LXC/LXD driver add neutron port binding without checking port state" [Medium,New] - Assigned to Marc Koderer (m-koderer) | 14:56 |
| *** merooney has joined #openstack-manila | 14:56 | |
| mkoderer___ | for me it's a critical issue that causes race conditions | 14:56 |
| bswartz | mkoderer___: http://lists.openstack.org/pipermail/openstack-dev/2016-March/089522.html | 14:58 |
| *** timcl has quit IRC | 15:02 | |
| *** mlima has quit IRC | 15:02 | |
| bswartz | oh mkoderer___ I see your bugfix doesn't actually touch the driver | 15:03 |
| bswartz | sorry I untargeted all bugs with LXD in the name | 15:03 |
| bswartz | mkoderer___ can you rename the bug please? | 15:03 |
| *** mlima has joined #openstack-manila | 15:05 | |
| vponomaryov | gouthamr: ping | 15:07 |
| gouthamr | vponomaryov: pong | 15:08 |
| vponomaryov | gouthamr: my local testing shows that "share manager" should call "update_replicated_snapshots" right after each creation of replica | 15:08 |
| openstackgerrit | Merged openstack/manila: Check share-network in 'share create' API https://review.openstack.org/256353 | 15:08 |
| vponomaryov | gouthamr: if we do not do so, then exists one case when we can get race | 15:08 |
| gouthamr | vponomaryov: it is called on the same interval as the replicas.. | 15:08 |
| vponomaryov | gouthamr: exactly, race appears when it is not called yet | 15:09 |
| gouthamr | vponomaryov: what's the race you're seeing? | 15:09 |
| *** timcl has joined #openstack-manila | 15:09 | |
| vponomaryov | gouthamr: here scenario: | 15:09 |
| gouthamr | vponomaryov: i've been testing with setting it as low as possible, 60 seconds. | 15:09 |
| vponomaryov | gouthamr: 1) create share; 2) create snapshot; 3) create replica; 4) promote replica not waiting periodic update; 5) delete old active replica; 6) try delete snapshot | 15:10 |
| vponomaryov | gouthamr: this is the problem, we should call it after each creation of share | 15:10 |
| vponomaryov | gouthamr: that's all | 15:10 |
| vponomaryov | gouthamr: on step 6 we will not have "provider_location" | 15:11 |
| vponomaryov | gouthamr: because it was not populated | 15:11 |
| vponomaryov | gouthamr: OR | 15:11 |
| vponomaryov | gouthamr: you can copy "provider location" always for all replicas | 15:12 |
| vponomaryov | gouthamr: then it will be the best | 15:12 |
| vponomaryov | gouthamr: solution, I think | 15:12 |
| gouthamr | vponomaryov: we could do that.. its not exactly a race.. its a known situation.. provider_location is a solution we tried since we name the snapshots the same across replicas on the NetApp backend. what if drivers want the provider location to be different across replicas? | 15:13 |
| vponomaryov | gouthamr: I find it useful too to use exactly "same" value | 15:13 |
| vponomaryov | gouthamr: it is "replicas" | 15:13 |
| vponomaryov | gouthamr: are you aware about cases where replicated snapshots have different names? | 15:14 |
| vponomaryov | gouthamr: completely different? | 15:14 |
| vponomaryov | gouthamr: without any common part? | 15:14 |
| gouthamr | vponomaryov: makes sense, ameade and I did think really hard about this... but then we realized that it is called 'provider' location for a reason.. i.e, drivers would use them how they want.. isn't that correct? | 15:15 |
| *** david-lyle has quit IRC | 15:15 | |
| *** david-lyle has joined #openstack-manila | 15:15 | |
| gouthamr | vponomaryov: not aware of such a situation. | 15:15 |
| vponomaryov | gouthamr: it is called so, because ofcompatibility with Cinder )) | 15:15 |
| gouthamr | vponomaryov: ah | 15:15 |
| vponomaryov | gouthamr: or, at least, you can make it conditional | 15:16 |
| gouthamr | vponomaryov: so we had this workaround in the netapp driver, where we correlate the provider location from the other snapshot instances that are passed in the update call.. | 15:16 |
| vponomaryov | gouthamr: when we return some special key, manager should consider it and copy providr locations | 15:16 |
| *** ociuhandu has quit IRC | 15:16 | |
| gouthamr | ameade: ^^ | 15:16 |
| vponomaryov | gouthamr: see, each driver has such problem, share manager should be updated to solve it for them | 15:16 |
| gouthamr | vponomaryov: i think that's a good idea.. | 15:17 |
| gouthamr | vponomaryov: will think about this and ping you in a bit.. | 15:18 |
| vponomaryov | gouthamr: I have about 30 min, then I will be available tomorrow | 15:18 |
| vponomaryov | gouthamr: if you agree, it is enough for me | 15:18 |
| gouthamr | vponomaryov: what do you think about the workaround? the driver's getting all the snapshot instances.. | 15:19 |
| vponomaryov | gouthamr: "provider location" was the only reason for it in my case | 15:19 |
| openstackgerrit | Alexey Ovchinnikov proposed openstack/manila: Revert "LXC/LXD driver" https://review.openstack.org/293498 | 15:19 |
| vponomaryov | gouthamr: if we have at least one proper snapshot instance then it is not required and manager can do its stuff | 15:20 |
| vponomaryov | gouthamr: I suppose same is true for NetApp driver | 15:21 |
| gouthamr | vponomaryov: if you promote a replica where the snapshot instance was in 'creating' state, it will be set to 'error' | 15:22 |
| vponomaryov | gouthamr: why error? gap between "ok" snapshot creation and next update is HUGE | 15:22 |
| *** ociuhandu has joined #openstack-manila | 15:23 | |
| gouthamr | vponomaryov: yes. you're correct | 15:23 |
| *** Yogi1 has quit IRC | 15:23 | |
| vponomaryov | gouthamr: most snapshot creations should be fast | 15:23 |
| vponomaryov | gouthamr: too fast for 60 sec | 15:23 |
| gouthamr | vponomaryov: at this point, if the snapshot is in 'error', the admin would need to be involved to reset the state | 15:24 |
| vponomaryov | gouthamr: it is different story | 15:24 |
| gouthamr | vponomaryov: if the admin can confirm that the snapshot is really there on the share that got promoted | 15:24 |
| openstackgerrit | Alexey Ovchinnikov proposed openstack/manila: Revert "Fix pylxd hard dependencies" https://review.openstack.org/293503 | 15:24 |
| vponomaryov | gouthamr: we need set provide location with each new registration of snapshot instance | 15:24 |
| *** Yogi1 has joined #openstack-manila | 15:25 | |
| vponomaryov | s/provide/provider/ | 15:25 |
| gouthamr | vponomaryov: okay, ill copy it from the active replica's snapshot instance | 15:25 |
| vponomaryov | gouthamr: what about condition? | 15:25 |
| gouthamr | vponomaryov: but allow drivers to update it to whatever they want in the update_replicated_snapshot call | 15:25 |
| vponomaryov | gouthamr: or leave it for future when driver with unique provider locations appear? | 15:25 |
| bswartz | aovchinnikov: your revert patch wasn't based on master | 15:25 |
| vponomaryov | gouthamr: I am ok with possibility to update | 15:26 |
| gouthamr | vponomaryov: awesome. it will be in the next patch :) | 15:26 |
| vponomaryov | gouthamr: ty | 15:26 |
| gouthamr | vponomaryov: no, thank you! | 15:26 |
| aovchinnikov | bswartz: hmm, it seems there is another patch in the way. does that mean that it should be done manually or it is ok to revert them one by one? | 15:29 |
| *** david-lyle_ has joined #openstack-manila | 15:34 | |
| *** david-lyle has quit IRC | 15:34 | |
| *** jasonsb has joined #openstack-manila | 15:37 | |
| openstackgerrit | Alex O'Rourke proposed openstack/manila: 3PAR remove file tree on delete when using nested shares https://review.openstack.org/290209 | 15:44 |
| *** rraja has quit IRC | 15:48 | |
| openstackgerrit | Alexey Ovchinnikov proposed openstack/manila: Revert "LXC/LXD driver" https://review.openstack.org/293523 | 15:48 |
| bswartz | aovchinnikov: we should block any further changes that affect LXD but the removal patch should be based master | 15:51 |
| bswartz | based on master | 15:51 |
| mkoderer___ | bswartz: cknight: can I ask for a review on https://review.openstack.org/#/c/290511/ | 15:52 |
| mkoderer___ | I would really like to get keystone v3 working :) | 15:52 |
| cknight | mkoderer: Sure. I'm not a Keystone expert, so you might want to find one of those, too. | 15:53 |
| mkoderer___ | cknight: good idea... I will ask someone from the keystone team | 15:53 |
| *** martyturner has quit IRC | 16:01 | |
| *** merooney has quit IRC | 16:01 | |
| *** cknight has quit IRC | 16:02 | |
| *** rraja has joined #openstack-manila | 16:03 | |
| *** baojg has quit IRC | 16:03 | |
| *** martyturner has joined #openstack-manila | 16:04 | |
| *** dustins has quit IRC | 16:07 | |
| *** a_ta has quit IRC | 16:11 | |
| *** a_ta has joined #openstack-manila | 16:12 | |
| openstackgerrit | Merged openstack/manila: glusterfs/common: don't suppress vol set errors https://review.openstack.org/284683 | 16:12 |
| *** a_ta has quit IRC | 16:12 | |
| *** a_ta has joined #openstack-manila | 16:12 | |
| *** adrianofr_ has joined #openstack-manila | 16:18 | |
| *** david-lyle_ is now known as david-lyle | 16:20 | |
| *** cknight has joined #openstack-manila | 16:30 | |
| *** timcl has quit IRC | 16:31 | |
| *** timcl has joined #openstack-manila | 16:49 | |
| *** merooney has joined #openstack-manila | 16:52 | |
| *** nkrinner has quit IRC | 16:54 | |
| *** sgotliv has quit IRC | 16:57 | |
| *** merooney has quit IRC | 17:05 | |
| *** lpetrut has quit IRC | 17:07 | |
| cknight | ganso: ping | 17:09 |
| ganso | cknight: pong | 17:11 |
| cknight | ganso: It seems I should also submit a patch with ShareResourceNotFound. https://github.com/openstack/manila/blob/master/manila/share/drivers/netapp/dataontap/cluster_mode/lib_base.py#L1050 | 17:13 |
| *** merooney has joined #openstack-manila | 17:13 | |
| ganso | cknight: ideally yes | 17:16 |
| *** vbellur has quit IRC | 17:17 | |
| ganso | cknight: share is deletable with force-delete now | 17:17 |
| *** a_ta has quit IRC | 17:18 | |
| *** a_ta has joined #openstack-manila | 17:19 | |
| ganso | cknight: so you can either throw ShareResourceNotFound | 17:19 |
| ganso | cknight: or allow error if update_access does not find share in backend | 17:20 |
| cknight | ganso: I see, so if the backend resource is gone, I have to issue a force delete. The way I've done it in my driver, a non-admin can delete a share in that case. Maybe I'll leave it as-is. | 17:20 |
| ganso | cknight: yes, but then your calls to update_access may succeed while your share is gone from backend | 17:22 |
| *** merooney has quit IRC | 17:23 | |
| *** a_ta has quit IRC | 17:23 | |
| ganso | cknight: and user will never know because he will not see logs | 17:24 |
| cknight | ganso: it's definitely a trade-off. I was surprised to discover update_access being called in the delete path. It'd be nice to know in update_access that a delete is in progress, but that might not be simple. | 17:25 |
| cknight | bswartz: ^^ How important is this to you? | 17:26 |
| *** sage has quit IRC | 17:27 | |
| *** lpetrut has joined #openstack-manila | 17:28 | |
| *** sage has joined #openstack-manila | 17:29 | |
| ganso | cknight: that's a different solution, it is simple, access helper would swallow exception if status is deleting. But that would also change current behavior of an error being thrown if rules cannot be removed properly when deleting a share (if share is indeed in the backend) | 17:29 |
| *** jasonsb has quit IRC | 17:29 | |
| cknight | ganso: I only want to swallow the exception in the driver's update_access if the share is gone and it is about to be deleted. I don't have enough info do that now. | 17:31 |
| ganso | cknight: share['status']? | 17:32 |
| cknight | ganso: oh, interesting, perhaps I could look for a 'deleting' status. | 17:33 |
| cknight | ganso: thanks, why didn't I think of that? | 17:33 |
| ganso | cknight: then again, if you are going to make this change, why not throw ShareResourceNotFound? | 17:34 |
| cknight | ganso: The design point in our driver is to not blow up if a user deletes something that is already gone. It's a corner case, to be sure, but I don't like frustrating users trying to delete stuff. | 17:35 |
| cknight | ganso: And I can throw the exception in a non-delete path. | 17:36 |
| rraja | bswartz: hi! what's the deadline for driver doc fixes to make it for mitaka release? | 17:38 |
| *** johnthetubaguy_ has joined #openstack-manila | 17:39 | |
| *** johnthetubaguy has quit IRC | 17:41 | |
| *** johnthetubaguy_ is now known as johnthetubaguy | 17:41 | |
| *** lpetrut has quit IRC | 17:43 | |
| *** lpetrut has joined #openstack-manila | 18:05 | |
| openstackgerrit | Alex Meade proposed openstack/manila: NetApp cDOT: Fix status updates for replicas https://review.openstack.org/291803 | 18:07 |
| openstackgerrit | Alex O'Rourke proposed openstack/manila: 3PAR remove file tree on delete when using nested shares https://review.openstack.org/290209 | 18:14 |
| *** martyturner has quit IRC | 18:14 | |
| *** a_ta has joined #openstack-manila | 18:18 | |
| openstackgerrit | Alex O'Rourke proposed openstack/manila: 3PAR remove file tree on delete when using nested shares https://review.openstack.org/290209 | 18:18 |
| *** dustins has joined #openstack-manila | 18:23 | |
| *** akshai has quit IRC | 18:25 | |
| *** akshai has joined #openstack-manila | 18:33 | |
| *** vbellur has joined #openstack-manila | 18:36 | |
| *** amit2131 has quit IRC | 18:55 | |
| *** amit213 has joined #openstack-manila | 18:55 | |
| openstackgerrit | Dustin Schoenbrun proposed openstack/manila: Add common capabilities matrix to devref https://review.openstack.org/288504 | 19:00 |
| dustins | tpsilva: ^^ See if that's a little more to your liking :) | 19:01 |
| bswartz | tpsilva: ping | 19:08 |
| tpsilva | bswartz: pong | 19:13 |
| bswartz | tpsilva: I wanted to talk to you about https://review.openstack.org/#/c/287758/ | 19:14 |
| bswartz | tpsilva: was going to suggest google hangout but it's still busted | 19:14 |
| tpsilva | bswartz: I'm in a meeting right know, but it should finish in a few minutes... I'll ping you after it finishes, ok? | 19:16 |
| tpsilva | dustins: looks great, thank you! | 19:16 |
| bswartz | k | 19:17 |
| dustins | tpsilva: you're welcome! | 19:17 |
| *** gouthamr has quit IRC | 19:23 | |
| *** merooney has joined #openstack-manila | 19:24 | |
| *** kro_sniffles is now known as kro_mnckn_doc | 19:27 | |
| tpsilva | bswartz: I'm at my desk now | 19:28 |
| tpsilva | bswartz: hangouts is fine for me | 19:29 |
| *** akshai has quit IRC | 19:30 | |
| bswartz | hangouts isn't working | 19:30 |
| bswartz | and I have no idea why | 19:30 |
| * bswartz is technologically challenged | 19:30 | |
| tpsilva | oh... weird | 19:30 |
| tpsilva | skype? | 19:30 |
| bswartz | skype seems to be unavailable on fedora | 19:31 |
| tpsilva | bswartz: hm... any ideas? I'm ok with irc as well | 19:34 |
| bswartz | yeah IRC is fine it's just slower | 19:34 |
| tpsilva | yep | 19:34 |
| *** akerr has quit IRC | 19:34 | |
| bswartz | cknight and I were reviewing your patch, and it seems to make thing better than they are now, but it's not immune to race conditions | 19:35 |
| *** ociuhandu has quit IRC | 19:35 | |
| bswartz | I'm trying to decide if we should push for a perfect solution in Mitaka, or if we should take this good solution and continue to iterate in Newton | 19:36 |
| tpsilva | I agree that it would not be completely immune to race conditions, I just can't think of one right now, but I'm pretty sure there is | 19:37 |
| bswartz | well my main regret with the current design is that we should not have removed the per-rule status | 19:37 |
| tpsilva | probably these 2 statuses for all the rules is not the best approach | 19:38 |
| tpsilva | exactly | 19:38 |
| bswartz | I think that we need to add that column back in and use it to track for each rule, whether it's been applied or not | 19:38 |
| cknight | tpsilva: The race is this: if the backend sets the rules to ACTIVE after the API layer reads the rule status, the API layer could think they are still out of sync and drop the request. | 19:38 |
| *** akshai has joined #openstack-manila | 19:39 | |
| bswartz | tpsilva: I don't think it's a good idea to make DB schema changes this late in the release, and I don't see another way to make this completely reliable | 19:39 |
| *** gouthamr has joined #openstack-manila | 19:40 | |
| bswartz | we could introduce more states and move the checking from the API to the manager, which would solve some problems | 19:40 |
| tpsilva | we could block the requests if access_rules_status is out of sync, but this would change the behavior | 19:41 |
| bswartz | no we should never block requests -- that breaks legacy behavior | 19:41 |
| tpsilva | I mean, disallow any requests to rules if they are not active | 19:41 |
| tpsilva | why? | 19:41 |
| bswartz | in the past you could send 100 access allow requests and they would all succeed | 19:42 |
| tpsilva | oh I see | 19:42 |
| bswartz | we should preserve that user experience | 19:42 |
| tpsilva | what would you suggest? could we somehow lock the update_access_rules method and make sure all the calls come as they were queued? | 19:44 |
| *** gouthamr_ has joined #openstack-manila | 19:44 | |
| bswartz | if the api service changed the rule state to needs_update and the manager changed the rule state to updating, then the manager could simply ignore any RPC calls for shares which were updating | 19:45 |
| tpsilva | but would that solve the race condition that cknight described above? | 19:45 |
| bswartz | in the design the API service would always send the RPC | 19:45 |
| bswartz | let me think | 19:45 |
| bswartz | yes because the API service would send the RPC no matter what | 19:46 |
| cknight | it would move any potential race to the manager, but you's still have multiple calls to update_access_rules in parallel | 19:46 |
| tpsilva | you're right, the manager would make this check to update or not, not api | 19:46 |
| bswartz | so no matter where the manager is in its processing, it will get another RPC telling it to do more work | 19:46 |
| bswartz | the manager could decide to drop the RPC if it already has pending working to do the update | 19:47 |
| cknight | if you make that change *and* use a lock in the manager, it could be done safely. | 19:47 |
| bswartz | the key is to use 2 different states | 19:47 |
| cknight | and when we adopt tooz, that would also solve the HA issue | 19:47 |
| *** gouthamr has quit IRC | 19:48 | |
| cknight | a lock is effectively a state, so would you really need 2 states? | 19:48 |
| bswartz | so if the manager starts doing an update, and the API service gets a new request, it changes the state back to needs_update, and the manager sees after it's done that it needs another update | 19:48 |
| bswartz | I'm suggesting a method which would work without tooz, but only with 1 instance of the manager running | 19:48 |
| bswartz | the moment you have 2 managers handling the same share a different class of issues pops up | 19:49 |
| cknight | Right. But can we design for both just as easily? | 19:49 |
| cknight | Using a lock (which later is backed by tooz) would do that, right? | 19:49 |
| bswartz | many things need to be redesigned to deal with multiple managers | 19:49 |
| bswartz | I don't think it's simple as relying on distributed locking | 19:50 |
| bswartz | distributed locked in necessary but not sufficient to solve the problem when you have 2 managers | 19:51 |
| bswartz | you need to handle recover paths when a manager crashes while holding a lock | 19:51 |
| cknight | ok, I agree your suggestion will work well with just 1 manager. | 19:51 |
| cknight | tiago, did you understand Ben's solution? | 19:51 |
| tpsilva | I did, I just got lost when you started talking about multiple managers... didn't know manila allowed that | 19:52 |
| bswartz | tpsilva: also, assuming you understood it, is it something that can be done in <24 hours? | 19:52 |
| cknight | tpsilva: you can do it, but there is another class of concurrency issues that can occur | 19:52 |
| bswartz | we're going to need to start punting bugs and this bug is bad enough that I'll take a good fix over a perfect fix | 19:52 |
| cknight | tpsilva: multiple processes on separate systems can pull from the event queue, so distributed locks would be needed to make it reliable. | 19:53 |
| bswartz | tpsilva: yes it's often referred to as Active/Active HA (high availability) and it's something that was intended to work across openstack but it's not tested and there are numerous known issues | 19:53 |
| cknight | tpsilva: people do it anyway, though | 19:53 |
| tpsilva | bswartz: looks like that would not require that much work.. I intent to start working on that early tomorrow | 19:54 |
| cknight | tpsilva: we need it merged tomorrow. I'll review it as soon as it's available (and I'm awake) | 19:54 |
| bswartz | okay let's leave cknight -1 on it and plan for another patchset | 19:54 |
| tpsilva | ok... it should be up before the meeting | 19:54 |
| cknight | tpsilva: thanks! | 19:55 |
| *** ig0r_ has quit IRC | 19:55 | |
| tpsilva | cknight, bswartz: no problem, thank you for finding these edge cases | 19:55 |
| *** ociuhandu has joined #openstack-manila | 19:56 | |
| *** merooney has quit IRC | 20:00 | |
| *** mlima has quit IRC | 20:01 | |
| *** timcl has quit IRC | 20:02 | |
| *** sgotliv has joined #openstack-manila | 20:10 | |
| *** martyturner has joined #openstack-manila | 20:15 | |
| bswartz | rraja: for docs, talk to the docs team about their deadlines -- for manila dev docs, you can push changes any time, but we WILL NOT backport dev docs changes ever | 20:17 |
| bswartz | markstur_: ping re https://bugs.launchpad.net/manila/+bug/1525010 | 20:18 |
| openstack | Launchpad bug 1525010 in Manila "manila 3par mediator message missing proper 3parclient pip package name" [Medium,Confirmed] - Assigned to Mark Sturdevant (mark-sturdevant) | 20:18 |
| rraja | bswartz: thanks! | 20:25 |
| *** dencaval has quit IRC | 20:27 | |
| *** Yogi1 has quit IRC | 20:29 | |
| *** lpetrut has quit IRC | 20:44 | |
| *** sgotliv has quit IRC | 20:44 | |
| *** cknight has quit IRC | 20:48 | |
| *** abhi has quit IRC | 20:48 | |
| *** yangyapeng has quit IRC | 20:58 | |
| *** yangyapeng has joined #openstack-manila | 20:59 | |
| *** dustins has quit IRC | 21:04 | |
| *** yangyapeng has quit IRC | 21:04 | |
| openstackgerrit | Clinton Knight proposed openstack/manila: NetApp cDOT: Raise ShareResourceNotFound in update_access https://review.openstack.org/293708 | 21:04 |
| *** yangyapeng has joined #openstack-manila | 21:06 | |
| *** ganso has quit IRC | 21:09 | |
| *** a_ta has quit IRC | 21:12 | |
| *** DericHorn-HP has joined #openstack-manila | 21:12 | |
| *** DericHorn-HP has quit IRC | 21:12 | |
| *** a_ta has joined #openstack-manila | 21:12 | |
| *** a_ta has quit IRC | 21:17 | |
| *** kro_mnckn_doc has quit IRC | 21:19 | |
| *** porrua has quit IRC | 21:29 | |
| *** eharney has quit IRC | 21:30 | |
| *** abhi has joined #openstack-manila | 21:30 | |
| *** darrenc is now known as darren_afk | 21:31 | |
| *** abhi has quit IRC | 21:31 | |
| *** abhi has joined #openstack-manila | 21:32 | |
| *** csaba has quit IRC | 21:33 | |
| *** rraja has quit IRC | 21:42 | |
| *** csaba has joined #openstack-manila | 21:46 | |
| *** darren_afk is now known as darrenc | 21:46 | |
| *** absubram has quit IRC | 21:49 | |
| *** absubram has joined #openstack-manila | 21:50 | |
| *** darkwsh has quit IRC | 21:50 | |
| *** tpsilva has quit IRC | 22:17 | |
| *** erlon has quit IRC | 22:17 | |
| *** martyturner has quit IRC | 22:26 | |
| *** chlong has joined #openstack-manila | 22:32 | |
| *** dalgaaf has quit IRC | 22:38 | |
| *** JoseMello has quit IRC | 22:39 | |
| *** vbellur has quit IRC | 22:42 | |
| *** xyang1 has quit IRC | 22:52 | |
| *** cknight has joined #openstack-manila | 22:54 | |
| *** merooney has joined #openstack-manila | 23:00 | |
| *** krotscheck has joined #openstack-manila | 23:11 | |
| *** akshai has quit IRC | 23:24 | |
| *** vbellur has joined #openstack-manila | 23:40 | |
| *** chlong has quit IRC | 23:41 | |
| *** cknight has quit IRC | 23:55 | |
| openstackgerrit | Mark Sturdevant proposed openstack/manila: Fix error logged for wrong HPE 3par client https://review.openstack.org/293763 | 23:57 |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!