| *** absubram has joined #openstack-manila | 00:01 | |
| *** akerr has joined #openstack-manila | 00:07 | |
| *** akerr_ has joined #openstack-manila | 00:08 | |
| *** akerr has quit IRC | 00:11 | |
| *** rushil_ has quit IRC | 00:13 | |
| *** jasonsb_ has quit IRC | 00:14 | |
| *** rushil has joined #openstack-manila | 00:14 | |
| *** mtanino_ has quit IRC | 00:21 | |
| openstackgerrit | Matt Riedemann proposed openstack/python-manilaclient: Update path to subunit2html in post_test_hook https://review.openstack.org/219934 | 00:23 |
|---|---|---|
| *** bill_az_ has quit IRC | 00:36 | |
| *** ganso_ has quit IRC | 00:45 | |
| *** rushil has quit IRC | 01:02 | |
| *** cknight has joined #openstack-manila | 01:03 | |
| *** cknight1 has joined #openstack-manila | 01:06 | |
| *** cknight has quit IRC | 01:08 | |
| *** BharatK has joined #openstack-manila | 01:20 | |
| *** ociuhandu has joined #openstack-manila | 01:24 | |
| *** rbradbury has quit IRC | 01:26 | |
| *** ganso_ has joined #openstack-manila | 01:27 | |
| *** jasonsb has joined #openstack-manila | 01:46 | |
| *** akerr_ has quit IRC | 02:23 | |
| *** akerr has joined #openstack-manila | 02:27 | |
| *** akerr is now known as akerr_away | 02:27 | |
| *** akerr_away has quit IRC | 02:29 | |
| *** haomaiwang has quit IRC | 02:36 | |
| *** haomaiwang has joined #openstack-manila | 02:38 | |
| openstackgerrit | Csaba Henk proposed openstack/manila: glusterfs: directory mapped share layout https://review.openstack.org/215293 | 02:39 |
| openstackgerrit | Csaba Henk proposed openstack/manila: glusterfs/common: refactor GlusterManager https://review.openstack.org/214921 | 02:39 |
| openstackgerrit | Csaba Henk proposed openstack/manila: glusterfs*: factor out common parts https://review.openstack.org/214462 | 02:39 |
| openstackgerrit | Csaba Henk proposed openstack/manila: glusterfs/layout: add layout base classes https://review.openstack.org/215172 | 02:39 |
| openstackgerrit | Csaba Henk proposed openstack/manila: glusterfs: volume mapped share layout https://review.openstack.org/215173 | 02:39 |
| openstackgerrit | Csaba Henk proposed openstack/manila: glusterfs-native: cut back on redundancy https://review.openstack.org/215021 | 02:39 |
| *** haomaiwang has quit IRC | 02:57 | |
| *** haomaiwa_ has joined #openstack-manila | 03:00 | |
| *** cknight1 has quit IRC | 03:00 | |
| *** rushil has joined #openstack-manila | 03:07 | |
| *** rushil has quit IRC | 03:17 | |
| *** bswartz has quit IRC | 03:30 | |
| *** bswartz has joined #openstack-manila | 03:34 | |
| *** Zhongjun_ has joined #openstack-manila | 03:48 | |
| *** gouthamr has quit IRC | 03:59 | |
| *** BharatK has quit IRC | 04:04 | |
| *** cknight has joined #openstack-manila | 04:08 | |
| *** baojg has joined #openstack-manila | 04:11 | |
| *** cknight has quit IRC | 04:13 | |
| *** baojg has quit IRC | 04:31 | |
| *** ganso_ has quit IRC | 04:45 | |
| *** BharatK has joined #openstack-manila | 04:59 | |
| *** BharatK has quit IRC | 05:06 | |
| *** rraja has joined #openstack-manila | 05:07 | |
| *** BharatK has joined #openstack-manila | 05:13 | |
| *** shausy has joined #openstack-manila | 05:16 | |
| *** deepakcs has joined #openstack-manila | 05:18 | |
| *** shausy has quit IRC | 05:27 | |
| *** lpetrut has joined #openstack-manila | 05:29 | |
| *** nkrinner has joined #openstack-manila | 06:02 | |
| *** vponomaryov1 has joined #openstack-manila | 06:14 | |
| vponomaryov1 | bswartz: we require "tagging" of novaclient with recent fix. Without it we are still blocked | 06:15 |
| vponomaryov1 | bswartz: https://github.com/openstack/python-novaclient/releases/tag/2.27.0 is pulled and it does not contain latest fix | 06:16 |
| vponomaryov1 | to_all: ^ | 06:16 |
| vponomaryov1 | if someone knows who should be poked to make a new tag in Novaclient it is time to do it =) | 06:17 |
| *** sgotliv has joined #openstack-manila | 06:30 | |
| *** kaisers1 has quit IRC | 06:34 | |
| *** lpetrut has quit IRC | 06:37 | |
| *** anubhav has joined #openstack-manila | 07:00 | |
| *** vponomaryov1 has left #openstack-manila | 07:35 | |
| *** ociuhandu has quit IRC | 07:41 | |
| *** Zhongjun_ has quit IRC | 07:47 | |
| *** lpetrut has joined #openstack-manila | 07:47 | |
| *** ociuhandu has joined #openstack-manila | 07:52 | |
| mkoderer | vponomaryov: can we set "use_library_from_git" and pull the fix as a workaround | 07:57 |
| openstackgerrit | Marc Koderer proposed openstack/manila: Workaround for Manila gate failure https://review.openstack.org/220028 | 08:04 |
| mkoderer | vponomaryov: ^ | 08:05 |
| vponomaryov | mkoderer: we still need a tag =) | 08:51 |
| vponomaryov | mkoderer: also have you tried to reproduce Neutron unstable bug on your lab? | 08:52 |
| vponomaryov | mkoderer: that is https://bugs.launchpad.net/manila/+bug/1480870 ? | 08:52 |
| openstack | Launchpad bug 1480870 in Manila "Service instance does not get ssh key with Generic driver" [High,In progress] - Assigned to Valeriy Ponomaryov (vponomaryov) | 08:52 |
| mkoderer | vponomaryov: already ping someone in #openstack-nova | 08:53 |
| mkoderer | s/ping/pinged/ | 08:53 |
| vponomaryov | mkoderer: very strange that it was not done after merge of commit | 08:54 |
| mkoderer | vponomaryov: no, I didn't had the time to reproduce anything in our lab | 08:54 |
| vponomaryov | mkoderer: I tried, reproduced once, and after restart of Neutron services it changed behaviour | 08:57 |
| vponomaryov | mkoderer: that confused even more | 08:57 |
| vponomaryov | mkoderer: then after lab reinstall it was not reproduced | 08:58 |
| openstackgerrit | Igor Malinovskiy proposed openstack/manila: Return share_type UUID instead of name in Share API https://review.openstack.org/220046 | 09:23 |
| *** anubhav has quit IRC | 09:23 | |
| *** deepakcs has quit IRC | 10:54 | |
| *** ganso_ has joined #openstack-manila | 10:55 | |
| openstackgerrit | Rodrigo Barbieri proposed openstack/manila: Share Migration support in generic driver https://review.openstack.org/179791 | 11:15 |
| openstackgerrit | Rodrigo Barbieri proposed openstack/manila: Add Share Migration feature https://review.openstack.org/179790 | 11:15 |
| bswartz | vponomaryov: they said they'd tag yesterday | 11:18 |
| vponomaryov | bswartz: https://github.com/openstack/python-novaclient/releases | 11:18 |
| vponomaryov | bswartz: they did tag yesterday, but not with our bugfix | 11:19 |
| vponomaryov | bswartz: so, need to clarify that need to tag it one more time with latest changes | 11:19 |
| vponomaryov | bswartz: so yesterdays ninja bugfix was useless because tag of novaclient was not followed right after merge of fix | 11:21 |
| bswartz | vponomaryov: I'm starting to wish we'd done the other workaround we talked about -- where we go around novaclient and call rest API directly | 11:21 |
| *** eharney has joined #openstack-manila | 11:21 | |
| *** BharatK has quit IRC | 11:22 | |
| *** timcl has joined #openstack-manila | 11:27 | |
| *** alyson_ has joined #openstack-manila | 11:27 | |
| *** ociuhandu has quit IRC | 11:36 | |
| *** BharatK has joined #openstack-manila | 11:36 | |
| mkoderer | vponomaryov: strange thing is that my workaround with LIBS_FROM_GIT doesn't work | 11:37 |
| mkoderer | http://logs.openstack.org/28/220028/1/check/gate-manila-tempest-dsvm-neutron/83973b1/logs/devstack-gate-setup-workspace-new.txt.gz#_2015-09-03_08_08_19_294 | 11:37 |
| mkoderer | novaclient is cloned with the fix | 11:38 |
| vponomaryov | mkoderer: why strange? novaclietn was used from dep - from site packages | 11:39 |
| openstackgerrit | Igor Malinovskiy proposed openstack/manila: Return share_type UUID instead of name in Share API https://review.openstack.org/220046 | 11:41 |
| mkoderer | vponomaryov: mh but I would say the newer version should be active if LIBS_FROM_GIT is set | 11:42 |
| mkoderer | at least this is what the documenation says | 11:42 |
| vponomaryov | mkoderer: do you have link to mentioned documentation? | 11:44 |
| vponomaryov | mkoderer: here is pip packages http://logs.openstack.org/28/220028/1/check/gate-manila-tempest-dsvm-neutron/83973b1/logs/pip-freeze.txt.gz - python-novaclient==2.26.0 | 11:44 |
| vponomaryov | mkoderer: and it does not refer to custom commit | 11:44 |
| mkoderer | vponomaryov: https://github.com/openstack-dev/devstack/blob/1d0b0d363e23896e72d9c66028d59c6046eb6046/doc/source/configuration.rst#libraries-from-git | 11:45 |
| vponomaryov | mkoderer: as for tempest, swift, manilaclient, etc... | 11:45 |
| mkoderer | vponomaryov: ok seems the documentation is misleading | 11:46 |
| bswartz | rofl: https://bugs.launchpad.net/python-novaclient/+bug/1491579 | 11:46 |
| openstack | Launchpad bug 1491579 in python-novaclient "against all sanity, nova needs to work around broken public clouds" [Critical,Confirmed] | 11:46 |
| vponomaryov | mkoderer: maybe there is bug. You sure that it is enough to be in pre_test_hook? | 11:47 |
| bswartz | I hope we do not face similar issues to ^ | 11:47 |
| vponomaryov | mkoderer: of Manila? | 11:47 |
| mkoderer | vponomaryov: I will have a closer look to the code... but I guess we will get our version tag in the next hours | 11:48 |
| vponomaryov | mkoderer: hours, yeah | 11:49 |
| *** breitz has quit IRC | 11:51 | |
| *** breitz has joined #openstack-manila | 11:51 | |
| *** ociuhandu has joined #openstack-manila | 11:55 | |
| *** dzamboni has joined #openstack-manila | 12:11 | |
| *** JoseMello has joined #openstack-manila | 12:22 | |
| *** martyturner has joined #openstack-manila | 12:23 | |
| *** absubram has quit IRC | 12:24 | |
| *** xyang1 has joined #openstack-manila | 12:24 | |
| *** akerr has joined #openstack-manila | 12:27 | |
| *** akerr has quit IRC | 12:28 | |
| *** akerr has joined #openstack-manila | 12:29 | |
| openstackgerrit | Alyson proposed openstack/manila: Adds the random option to Manila retry function https://review.openstack.org/219851 | 12:33 |
| *** porrua has joined #openstack-manila | 12:36 | |
| *** rraja has quit IRC | 12:59 | |
| *** BharatK has quit IRC | 13:00 | |
| *** tpsilva has joined #openstack-manila | 13:01 | |
| bswartz | vponomaryov: ping | 13:07 |
| vponomaryov | bswartz: pong | 13:07 |
| bswartz | vponomaryov: why exactly does the tag matter for novaclient | 13:08 |
| *** timcl has quit IRC | 13:08 | |
| vponomaryov | bswartz: we need it in pip | 13:09 |
| vponomaryov | bswartz: novaclient is pulled as dependency | 13:09 |
| bswartz | we can't use the LIBS_FROM_GIT feature of devstack? | 13:09 |
| vponomaryov | bswartz: mkoderer already tried | 13:09 |
| vponomaryov | bswartz: did not help | 13:10 |
| bswartz | why didn't it work | 13:10 |
| bswartz | it should work | 13:10 |
| mkoderer | bswartz: https://review.openstack.org/#/c/220028/ | 13:10 |
| mkoderer | bswartz: it's a bug in devstack | 13:10 |
| mkoderer | LIBS_FROM_GIT seems not to work properly | 13:10 |
| bswartz | what bug in devstack? I've used that feature before and it works fine | 13:10 |
| vponomaryov | bswartz: we say the fact, it did not work for us | 13:11 |
| vponomaryov | bswartz: see https://review.openstack.org/#/c/220028/ | 13:11 |
| vponomaryov | bswartz: version 2.26 was used | 13:11 |
| bswartz | I'm looking over the logs | 13:11 |
| bswartz | can you see where 2.26 was used? | 13:12 |
| vponomaryov | bswartz: http://logs.openstack.org/28/220028/1/check/gate-manila-tempest-dsvm-neutron/83973b1/logs/pip-freeze.txt.gz | 13:12 |
| bswartz | that's pip though | 13:12 |
| bswartz | LIBS_FROM_GIT means don't install the pip version | 13:13 |
| *** rraja has joined #openstack-manila | 13:13 | |
| mkoderer | bswartz: but novaclient is installed | 13:13 |
| bswartz | https://bugs.launchpad.net/devstack/+bug/1434457 | 13:13 |
| openstack | Launchpad bug 1434457 in devstack "LIBS_FROM_GIT have no effect on OS project clients" [Undecided,Invalid] | 13:13 |
| bswartz | marked invalid.... | 13:14 |
| mkoderer | bswartz: I tried to ping several nova folks to get the tag issue fixed.. maybe you have more success | 13:15 |
| *** dustins has joined #openstack-manila | 13:15 | |
| mkoderer | bswartz: I guess fixing the devstack bug takes more time | 13:16 |
| *** BharatK has joined #openstack-manila | 13:16 | |
| vponomaryov | mkoderer: everyone from us tried to ping | 13:17 |
| vponomaryov | mkoderer: looks like they are concerned more about another bug - https://bugs.launchpad.net/python-novaclient/+bug/1491579 | 13:17 |
| openstack | Launchpad bug 1491579 in python-novaclient "against all sanity, nova needs to work around broken public clouds" [Critical,In progress] - Assigned to Sean Dague (sdague) | 13:17 |
| bswartz | I know | 13:17 |
| mkoderer | vponomaryov: yeah | 13:17 |
| *** gouthamr has joined #openstack-manila | 13:18 | |
| vponomaryov | mkoderer: but anyway, "tag" is one-sec-op | 13:18 |
| * mkoderer don't care about public clouds :) | 13:18 | |
| vponomaryov | broken public clouds & =) | 13:19 |
| bswartz | I thought the vendors who operated broken public clouds were involved with OpenStack development -- why aren't they speaking up about that issue? | 13:21 |
| vponomaryov | bswartz: what do you think about running simple manila job for Nova, Neutron and CInder? | 13:25 |
| vponomaryov | bswartz: where we could perform simple things like creations of shares and snapshots? | 13:25 |
| *** BharatK has quit IRC | 13:26 | |
| bswartz | vponomaryov: how is that different from what gate-manila-tempest-dsvm-neutron does? | 13:28 |
| akerr | bswartz: I think he means have the tests run in nova's gate | 13:29 |
| vponomaryov | bswartz: it is proposal, this one is full-test-run | 13:29 |
| vponomaryov | akerr: right | 13:29 |
| bswartz | oh you want to add manila tests to the integrated gate | 13:29 |
| bswartz | which would prevent them from breaking us in the future | 13:30 |
| vponomaryov | bswartz: yes =) | 13:30 |
| bswartz | that's an fantastic idea if they will go for it | 13:30 |
| akerr | may as well add cinder to the list too :) | 13:31 |
| bswartz | vponomaryov: the downside to that approach is that manila patches will have to go through the integrated gate unlike the situation today were we have our own gate | 13:31 |
| bswartz | s/were/where/ | 13:31 |
| vponomaryov | bswartz: what do you mean by "integrated gate"? | 13:34 |
| bswartz | the "integrated" change queue in the gate -- it contains all core projects (nova, cinder, keystone, neutron, etc) | 13:35 |
| bswartz | it means that manila patches would have to be tested in combination with everything else in the integrated change queue in order to merge | 13:36 |
| bswartz | because if there are manila tests in the integrated gate, then we can break them too | 13:36 |
| bswartz | the effect would be that when the gate backs up, our merges would go more slowly | 13:37 |
| bswartz | today we don't have that problem | 13:37 |
| bswartz | because manila has its own change queue | 13:37 |
| *** rraja has quit IRC | 13:38 | |
| *** rraja has joined #openstack-manila | 13:38 | |
| *** cknight has joined #openstack-manila | 13:44 | |
| dustins | bswartz: But isn't it better to know if one of their changes (or one of ours) breaks something in something else? | 13:44 |
| vponomaryov | bswartz: we can not break them until our project is not used there | 13:45 |
| bswartz | dustins: I didn't say it wasn't a worthwhile tradeoff, just that there is a tradeoff | 13:45 |
| vponomaryov | bswartz: I talked to owner of a bug https://bugs.launchpad.net/devstack/+bug/1434457 | 13:45 |
| openstack | Launchpad bug 1434457 in devstack "LIBS_FROM_GIT have no effect on OS project clients" [Undecided,Invalid] | 13:45 |
| bswartz | the price you pay for having your project gate other projects is that you have to follow the same rules they do | 13:46 |
| vponomaryov | bswartz: in simple words - some problem still exists | 13:46 |
| bswartz | vponomaryov: clearly some problem still exists | 13:46 |
| vponomaryov | bswartz: we can not just take latest master for any of python-*client | 13:47 |
| dustins | bswartz: Certainly upsides and downsides to both approaches | 13:47 |
| bswartz | so why did Nikolay abandon his fix? | 13:47 |
| vponomaryov | bswartz: read comment sin https://review.openstack.org/#/c/163335/ | 13:48 |
| vponomaryov | s/comment sin/comments in/ | 13:48 |
| vponomaryov | bswartz: answer similar to thing we already follow | 13:48 |
| vponomaryov | bswartz: wait for tag | 13:48 |
| vponomaryov | bswartz: =) | 13:48 |
| *** eharney has quit IRC | 13:49 | |
| *** rraja has quit IRC | 13:49 | |
| vponomaryov | bswartz: also https://review.openstack.org/#/c/166149/ | 13:51 |
| *** rraja has joined #openstack-manila | 13:54 | |
| *** lpabon has joined #openstack-manila | 13:58 | |
| *** timcl has joined #openstack-manila | 13:58 | |
| openstackgerrit | Igor Malinovskiy proposed openstack/manila: Allow service image download to be skipped https://review.openstack.org/205222 | 13:59 |
| *** rraja has quit IRC | 14:00 | |
| bswartz | vponomaryov: you following nova channel / nova meeting? | 14:02 |
| bswartz | looks like they may bring back 219696 | 14:02 |
| *** zhongjun has quit IRC | 14:02 | |
| vponomaryov | bswartz: yeah, I do | 14:03 |
| vponomaryov | bswartz: and I had facepalm | 14:03 |
| *** rushil has joined #openstack-manila | 14:04 | |
| u_glide | :) | 14:04 |
| bswartz | vponomaryov: btw this is why I advocated for not depending on any nova change at all yesterday | 14:04 |
| bswartz | I knew it would take more than 24 hours for anything useful to happen | 14:04 |
| bswartz | we still have the option to write custom nova client for attach volume call | 14:05 |
| bswartz | and unblock ourselves | 14:05 |
| vponomaryov | bswartz: hm, you want to do it in Liberty? | 14:05 |
| vponomaryov | bswartz: or Mitaka? | 14:05 |
| bswartz | I wanted to do it yesterday | 14:06 |
| bswartz | it makes less sense today though because if they bring back 219696 then we're back on track | 14:06 |
| bswartz | vponomaryov: it was a proposal for a temporary workaround anyways -- we would have backed it out after nova fixed the problem | 14:07 |
| vponomaryov | bswartz: I like minimizes dependencies | 14:07 |
| vponomaryov | minimizing | 14:07 |
| bswartz | vponomaryov: minimizing dependencies is nice but you end up reinventing the world and taking on more maintenance burden | 14:08 |
| bswartz | it's better to find a good balance | 14:08 |
| *** rushil has quit IRC | 14:09 | |
| vponomaryov | bswartz: in case of novaclient we use "several" things | 14:09 |
| vponomaryov | bswartz: not ALL features | 14:09 |
| bswartz | still having them maintain the client saves us significant work | 14:10 |
| *** rraja has joined #openstack-manila | 14:11 | |
| vponomaryov | bswartz: not really agree, because APi is expected to be unchanged | 14:12 |
| vponomaryov | bswartz: so, we will write it once | 14:12 |
| vponomaryov | bswartz: as, for example, Neutron-related stuff | 14:12 |
| vponomaryov | bswartz: like net interfaces | 14:12 |
| bswartz | still there are bugs, and refeactors that happen | 14:12 |
| bswartz | currently that work is done by other people | 14:13 |
| bswartz | if we write our own layer then all the maintenance falls on us | 14:13 |
| bswartz | you can argue it's low, but low is still more than zero | 14:13 |
| vponomaryov | bswartz: but it is under our control - it is huge benefit | 14:14 |
| bswartz | that's debatable | 14:14 |
| bswartz | it's a benefit, I'd argue not a huge one, and when you weigh it against the costs it's not a clear win | 14:14 |
| bswartz | developers make these tradeoffs all the time though -- consider all of the code we've deleted from manila because we now use the oslo libs | 14:16 |
| bswartz | one could argue it's better not to use oslo and to keep our own implementations for the same reasons you mention | 14:16 |
| bswartz | the logical extreme of your argument is to have no deps at all and to implement everything using standard python | 14:17 |
| vponomaryov | bswartz: we carry elephant for small things =) | 14:18 |
| *** eharney has joined #openstack-manila | 14:19 | |
| bswartz | that's how software development works though -- usually when you import a library you only use 5% of what that library provides and you ignore the rest | 14:19 |
| bswartz | we import dozens of libraries and use only what we need | 14:20 |
| bswartz | it's still a win to do an import instead of writing your own lib | 14:20 |
| bswartz | the only reason I brought up writing our own nova client is because (1) it would solve a critical problem quickly (2) waiting for nova would be slow (3) I thought it would be a small amount of work which we could throw away easily after nova is un-broken | 14:22 |
| vponomaryov | bswartz: (3) thought? not anymore? | 14:25 |
| bswartz | I still think that writing a client for 1 api would be lightweight | 14:26 |
| bswartz | but it's not needed | 14:26 |
| bswartz | vponomaryov: just have to wait 11 hours for gate now | 14:28 |
| bswartz | and change your depends-on back to that change | 14:28 |
| vponomaryov | bswartz: why 11? | 14:28 |
| bswartz | vponomaryov: that's how backed up the integrated gate is | 14:28 |
| *** rushil_ has joined #openstack-manila | 14:28 | |
| bswartz | yesterday it was 15 hours so this is an improvement | 14:29 |
| ganso_ | :( | 14:30 |
| vponomaryov | oh | 14:30 |
| vponomaryov | bswartz; I can implement client side fix faster and we can merge it faster | 14:31 |
| vponomaryov | bswartz: I mean workaround | 14:31 |
| vponomaryov | bswartz: into our repo | 14:31 |
| bswartz | vponomaryov: that was true yesterday as well | 14:33 |
| *** devlaps has quit IRC | 14:33 | |
| bswartz | with the server side fix though, we can simply add the depends-on and start workflowing | 14:33 |
| bswartz | everything has already missed L-3 so now it's just a matter of getting everything done over the next few days | 14:33 |
| *** eharney has quit IRC | 14:34 | |
| *** ayma has joined #openstack-manila | 14:34 | |
| cknight | ganso_: Are there functional tests for share migration? | 14:34 |
| *** mtanino has joined #openstack-manila | 14:35 | |
| cknight | ganso_: I tried to get migration working last night using host-only migrations with my driver. | 14:36 |
| cknight | ganso_: I stepped through your code in a debugger to follow the flow. | 14:36 |
| cknight | ganso_: I followed it through API and scheduler layers, but it kept blowing up in the share manager. | 14:37 |
| *** markstur has quit IRC | 14:38 | |
| cknight | ganso_: I'd expect that since migration works in the generic driver, we would have Tempest tests for that before merging it. Have you written tests? | 14:38 |
| cknight | ganso_: I'm looking for something to tell me if I'm doing something wrong. | 14:39 |
| *** lpabon_ has joined #openstack-manila | 14:40 | |
| *** lpabon has quit IRC | 14:40 | |
| *** markstur has joined #openstack-manila | 14:43 | |
| *** eharney has joined #openstack-manila | 14:48 | |
| ganso_ | cknight: I have not written functional tests for migration | 14:52 |
| ganso_ | cknight: could you please post you log with the errors you faced in Manager code? | 14:52 |
| ganso_ | cknight: *your log | 14:52 |
| cknight | ganso_: ok | 14:52 |
| *** mtanino has quit IRC | 14:53 | |
| csaba | bswartz: ping | 14:58 |
| bswartz | csaba: pong | 14:58 |
| csaba | bswartz: I'd like to add a meeting topic but can't as the OpenId login seems to have some issue | 14:58 |
| bswartz | lol | 14:58 |
| bswartz | what topic | 14:58 |
| csaba | bswartz: "questions about autommount" | 14:58 |
| bswartz | oh crud I can OpenID either | 14:59 |
| bswartz | 503 Service Unavailable No server is available to handle this request. | 14:59 |
| csaba | yep | 14:59 |
| *** Zhongjun_ has joined #openstack-manila | 15:00 | |
| *** rushil_ has quit IRC | 15:04 | |
| openstackgerrit | Alex Meade proposed openstack/manila: Add DB changes for consistency-groups https://review.openstack.org/215343 | 15:08 |
| openstackgerrit | Alex Meade proposed openstack/manila: Scheduler changes for consistency groups https://review.openstack.org/215344 | 15:08 |
| openstackgerrit | Alex Meade proposed openstack/manila: Add Consistency Groups API https://review.openstack.org/215345 | 15:08 |
| openstackgerrit | Alex Meade proposed openstack/manila: Consistency Group Support for the Generic Driver https://review.openstack.org/219891 | 15:09 |
| openstackgerrit | Alex Meade proposed openstack/manila: Add functional tests for Manila consistency groups https://review.openstack.org/215346 | 15:09 |
| openstackgerrit | Alex Meade proposed openstack/manila: Consistency groups in NetApp cDOT drivers https://review.openstack.org/215347 | 15:09 |
| *** esker has joined #openstack-manila | 15:10 | |
| openstackgerrit | Valeriy Ponomaryov proposed openstack/manila: Add possibility to setup password for generic driver https://review.openstack.org/219788 | 15:11 |
| *** esker has quit IRC | 15:14 | |
| *** crobertsrh has joined #openstack-manila | 15:17 | |
| ganso_ | vponomaryov: you changed your "depends-on" of https://review.openstack.org/#/c/219788/, if I just recheck, will it test with your updated "depends-on"? | 15:19 |
| vponomaryov | ganso_: I think yes | 15:20 |
| vponomaryov | ganso_: it should | 15:20 |
| ganso_ | great! | 15:20 |
| ganso_ | vponomaryov: thanks | 15:20 |
| *** nkrinner has quit IRC | 15:34 | |
| bswartz | vponomaryov: what about nova job? | 15:34 |
| vponomaryov | bswartz: some one failed, will need to perform recheck | 15:34 |
| vponomaryov | bswartz: queue is long | 15:35 |
| vponomaryov | bswartz: sooo | 15:35 |
| *** mtanino has joined #openstack-manila | 15:36 | |
| crobertsrh | Anyone in here seen an issue where running devstack in a vm with manila enabled causes horizon to bomb out with "too many redirects"? | 15:41 |
| openstackgerrit | Valeriy Ponomaryov proposed openstack/manila: Fix volume attach to Nova VM call https://review.openstack.org/220204 | 15:41 |
| bswartz | crobertsrh: yes others have seen that | 15:41 |
| bswartz | crobertsrh: there is a proposed fix for it | 15:41 |
| crobertsrh | Ooh, that was, of course, my next question | 15:41 |
| crobertsrh | Is it a manila fix or a horizon fix? | 15:42 |
| vponomaryov | bswartz: above is our local fix for gates ^ | 15:43 |
| ganso_ | vponomaryov: Should I add "depends-on" on 220204 as well? | 15:49 |
| *** breitz has quit IRC | 15:49 | |
| vponomaryov | ganso_: as you want | 15:49 |
| vponomaryov | ganso_: it workarounds Nova bug | 15:49 |
| vponomaryov | ganso_: and depends on us | 15:50 |
| vponomaryov | ganso_: not other projects | 15:50 |
| ganso_ | vponomaryov: great! thanks | 15:50 |
| *** absubram has joined #openstack-manila | 15:51 | |
| *** rushil has joined #openstack-manila | 15:56 | |
| bswartz | over here | 16:00 |
| * bswartz waves | 16:00 | |
| akerr | lunch time, bugs can wait | 16:00 |
| akerr | :P | 16:00 |
| bswartz | so regarding blueprints, I'll just reiterate that it's best to keep the discussion in gerrit | 16:01 |
| bswartz | regarding bugs | 16:01 |
| vponomaryov | akerr: lol | 16:01 |
| *** a_ta has joined #openstack-manila | 16:01 | |
| *** rraja has quit IRC | 16:02 | |
| bswartz | we want to merge any bugfix that solves gate problems obviously | 16:02 |
| bswartz | https://review.openstack.org/#/c/219788/ | 16:02 |
| vponomaryov | also local Nova bug workaround: https://review.openstack.org/220204 | 16:02 |
| bswartz | ^ this is aimed at fixing an instability caused by neutron | 16:02 |
| vponomaryov | Nova's bugfix here - https://review.openstack.org/#/c/219696/ | 16:02 |
| bswartz | https://review.openstack.org/#/c/219696/ this is something we need Depends-On: to get around the nova bug | 16:03 |
| *** zhongjun has joined #openstack-manila | 16:03 | |
| bswartz | https://review.openstack.org/#/c/220204/ is really interesting -- this is what I was hoping to see yesterday | 16:04 |
| bswartz | cknight: any bugs that need merging for microversions to unblock stuff | 16:04 |
| bswartz | cFouts: ^ | 16:04 |
| * bswartz should have known everyone would leave after the meeting ended | 16:05 | |
| vponomaryov | not everyone =) | 16:05 |
| cknight | bswartz: Not that I know of. The /v2 patch is still in our gate. akerr is nursing that one. | 16:06 |
| bswartz | everyone here in eastern time zone, where it's lunch time | 16:06 |
| bswartz | cknight: not even on client side? | 16:06 |
| cknight | bswartz: cfouts' first patch is up, so I think we're OK. | 16:06 |
| bswartz | merged? | 16:06 |
| cknight | bswartz: Of course not, but it's there. | 16:07 |
| cknight | bswartz: It couldn't get through the gate yesterday. | 16:07 |
| bswartz | how are people able to test new features? | 16:07 |
| vponomaryov | bswartz: сруккн-зшслштп, | 16:07 |
| bswartz | they must download cfout's client patch | 16:07 |
| bswartz | ? | 16:07 |
| vponomaryov | bswartz: cherry-picking changes | 16:07 |
| cknight | bswartz: yes | 16:07 |
| bswartz | okay | 16:07 |
| bswartz | that's not a great situation but I suppose we'll take it, given how bad everything else is | 16:08 |
| bswartz | I'll be wathcing for jobs to start passing jenkins | 16:09 |
| bswartz | in the mean time I'll be preparing to start merging anything that doesn't have -1s / -2s | 16:10 |
| bswartz | and looks for an ML post from me about the technical FFE exception deadline | 16:10 |
| openstackgerrit | Andrew Kerr proposed openstack/manila: Add v2 Manila API path as base for microversions https://review.openstack.org/220221 | 16:30 |
| *** esker has joined #openstack-manila | 16:38 | |
| ganso_ | cknight: please send me your migration error logs when you have some spare time | 16:47 |
| *** lpabon_ has quit IRC | 16:48 | |
| *** lpetrut has quit IRC | 16:49 | |
| markstur | Success! | 16:52 |
| markstur | bswartz, xyang1, toabctl Previously in gate: https://review.openstack.org/#/c/219333/ passed Jenkins, but needs reapproval. | 16:54 |
| markstur | bswartz, ^ bug was not targetted. https://bugs.launchpad.net/manila/+bug/1488662 | 16:54 |
| openstack | Launchpad bug 1488662 in Manila "HP 3PAR extra-spec prefix needs to be hp3par" [Low,In progress] - Assigned to Mark Sturdevant (mark-sturdevant) | 16:54 |
| *** martyturner has quit IRC | 16:57 | |
| *** haomaiw__ has joined #openstack-manila | 17:02 | |
| *** timcl has quit IRC | 17:02 | |
| *** haomaiwa_ has quit IRC | 17:03 | |
| vponomaryov | bswartz: ping | 17:04 |
| bswartz | okay stuff is start to pass check jobs | 17:04 |
| * bswartz cheers | 17:04 | |
| *** martyturner has joined #openstack-manila | 17:05 | |
| vponomaryov | bswartz: what you think about we abandon https://review.openstack.org/#/c/205222/ and just set default image existing Cirros? | 17:05 |
| vponomaryov | for those CIs tha tdo not need image | 17:05 |
| vponomaryov | so, right now, everyone can avoid download of our image without any change to source code | 17:05 |
| vponomaryov | only update of one env var | 17:06 |
| bswartz | vponomaryov: wouldn't that cause cirros to download twice? | 17:06 |
| *** johnthetubaguy has joined #openstack-manila | 17:07 | |
| vponomaryov | bswartz: good question, it depends on what does devstack func "upload_image" do in case of duplication | 17:08 |
| bswartz | vponomaryov: I like how Ramy's patch is clear about its purpose | 17:08 |
| johnthetubaguy | bswartz: so we think this change should fix your gate, without the need for the new python-novaclient release: https://review.openstack.org/#/c/219696/ | 17:08 |
| johnthetubaguy | bswartz: its not merged yet though | 17:08 |
| vponomaryov | johnthetubaguy: we already depend on it | 17:08 |
| bswartz | johnthetubaguy: it already allowed check jobs to start passing | 17:09 |
| vponomaryov | johnthetubaguy: in our commits | 17:09 |
| johnthetubaguy | awesome | 17:09 |
| bswartz | johnthetubaguy: when it gets through the gate then our changes will be able to get through too | 17:09 |
| johnthetubaguy | sorry about the break there, thanks for keeping pushing on us! | 17:09 |
| johnthetubaguy | its great you folks found this rather than post release (eek!) | 17:09 |
| bswartz | vponomaryov: Ramy's patch just adds 1 boolean option which allows skipping of any image download | 17:11 |
| bswartz | since 90% of CI systems don't need to download any image, it seems like a good idea | 17:11 |
| vponomaryov | bswartz: which patch? | 17:11 |
| bswartz | the one you linked | 17:11 |
| vponomaryov | its Igor's one | 17:11 |
| vponomaryov | oh, wait | 17:12 |
| vponomaryov | yes, you are right, need to rest | 17:12 |
| openstackgerrit | Csaba Henk proposed openstack/manila: glusterfs: directory mapped share layout https://review.openstack.org/215293 | 17:14 |
| openstackgerrit | Csaba Henk proposed openstack/manila: glusterfs/common: refactor GlusterManager https://review.openstack.org/214921 | 17:14 |
| openstackgerrit | Csaba Henk proposed openstack/manila: glusterfs*: factor out common parts https://review.openstack.org/214462 | 17:14 |
| openstackgerrit | Csaba Henk proposed openstack/manila: glusterfs/layout: add layout base classes https://review.openstack.org/215172 | 17:14 |
| openstackgerrit | Csaba Henk proposed openstack/manila: glusterfs: volume mapped share layout https://review.openstack.org/215173 | 17:14 |
| openstackgerrit | Csaba Henk proposed openstack/manila: glusterfs-native: cut back on redundancy https://review.openstack.org/215021 | 17:14 |
| vponomaryov | bswartz: about image | 17:14 |
| vponomaryov | bswartz: it will be registered twice, but not downloaded twice | 17:14 |
| *** timcl has joined #openstack-manila | 17:14 | |
| vponomaryov | bswartz: as it is kept in special dir and is searched by names | 17:15 |
| bswartz | vponomaryov: still not a good reason to abandon ramy's change | 17:18 |
| bswartz | skipping the download/glance logic is clearly better when it's not needed, and adding 1 boolean option is minimal | 17:18 |
| *** a_ta has quit IRC | 17:18 | |
| vponomaryov | bswartz: there is way to achieve result without changes | 17:19 |
| bswartz | that's always been true | 17:19 |
| vponomaryov | bswartz: but I do not insist. For the moment it has broken logic where image is tried to be registered anyway | 17:19 |
| vponomaryov | bswartz: just without registration | 17:19 |
| bswartz | I can set my manila image to be http://example.com/one-byte-file | 17:19 |
| bswartz | that's a hack though | 17:20 |
| vponomaryov | bswartz: our local workaround passed - https://review.openstack.org/#/c/220204/ | 17:20 |
| bswartz | vponomaryov: good to know we have that | 17:21 |
| vponomaryov | bswartz; or wait for Nova's bugfix successful merge? | 17:21 |
| bswartz | I want to wait for nova's fix | 17:21 |
| bswartz | because if we merge your workaround we'll have to unmerge it later | 17:21 |
| vponomaryov | bswartz: not really | 17:21 |
| bswartz | yes really | 17:21 |
| bswartz | we don't want to start diverging from what nova clients does | 17:22 |
| bswartz | I expect nova will continue to enhance their client and we get those enhancements for free | 17:22 |
| *** esker has quit IRC | 17:23 | |
| bswartz | vponomaryov: if only we had 220204 yesterday | 17:25 |
| bswartz | okay about 90 minutes late for lunch | 17:26 |
| bswartz | time to go fix that | 17:26 |
| *** martyturner has quit IRC | 17:30 | |
| openstackgerrit | Valeriy Ponomaryov proposed openstack/manila: Add possibility to setup password for generic driver https://review.openstack.org/219788 | 17:32 |
| *** lpabon has joined #openstack-manila | 17:35 | |
| *** timcl has quit IRC | 17:36 | |
| *** lpabon has quit IRC | 17:36 | |
| openstackgerrit | Valeriy Ponomaryov proposed openstack/manila: Add possibility to setup password for generic driver https://review.openstack.org/219788 | 17:37 |
| *** timcl has joined #openstack-manila | 17:37 | |
| vponomaryov | ganso_: I updated https://review.openstack.org/#/c/219788/ because it was fixing api tempest tests but not scenario | 17:38 |
| *** timcl1 has joined #openstack-manila | 17:39 | |
| vponomaryov | ganso_: now should fix both | 17:39 |
| ganso_ | vponomaryov: thanks! I just issued a recheck :) | 17:39 |
| *** timcl has quit IRC | 17:41 | |
| *** lpetrut has joined #openstack-manila | 17:43 | |
| *** absubram has quit IRC | 17:43 | |
| openstackgerrit | Valeriy Ponomaryov proposed openstack/manila: Add share hooks https://review.openstack.org/201669 | 17:43 |
| openstackgerrit | Valeriy Ponomaryov proposed openstack/manila: Add POC notification hook https://review.openstack.org/201697 | 17:46 |
| openstackgerrit | Valeriy Ponomaryov proposed openstack/manila: Add doc for share driver hooks https://review.openstack.org/202101 | 17:47 |
| *** Zhongjun_ has quit IRC | 17:57 | |
| *** lpetrut has quit IRC | 18:06 | |
| *** martyturner has joined #openstack-manila | 18:10 | |
| *** jasonsb has quit IRC | 18:14 | |
| *** jasonsb has joined #openstack-manila | 18:14 | |
| *** esker has joined #openstack-manila | 18:17 | |
| *** esker has quit IRC | 18:19 | |
| *** jasonsb has quit IRC | 18:19 | |
| *** pgbridge has quit IRC | 18:20 | |
| *** esker has joined #openstack-manila | 18:21 | |
| bswartz | vponomaryov: why patchset 7? https://review.openstack.org/#/c/219788/ | 18:22 |
| bswartz | did it pass jenkins only because it fell back to SSH and we got lucky not triggering neutron bug? | 18:22 |
| *** pgbridge has joined #openstack-manila | 18:28 | |
| *** rushil has quit IRC | 18:29 | |
| *** martyturner has quit IRC | 18:34 | |
| *** esker has quit IRC | 18:38 | |
| *** esker has joined #openstack-manila | 18:39 | |
| *** esker has quit IRC | 18:39 | |
| *** esker has joined #openstack-manila | 18:40 | |
| openstackgerrit | Clinton Knight proposed openstack/manila: Add v2 Manila API path as base for microversions https://review.openstack.org/220221 | 18:40 |
| *** timcl1 has quit IRC | 18:41 | |
| ganso_ | bswartz: pin | 18:44 |
| ganso_ | bswartz: g | 18:44 |
| bswartz | ganso_: pon | 18:49 |
| bswartz | ganso_: g | 18:49 |
| ganso_ | bswartz: lol | 18:50 |
| ganso_ | bswartz: I created a functional test | 18:51 |
| ganso_ | bswartz: I just do not know if I created in the correct folder, etc, but I will submit because I want to test it on Jenkins | 18:51 |
| ganso_ | bswartz: it is simple for now, but if it works, I will create the more elaborate ones and optimizations | 18:51 |
| ganso_ | bswartz: the thing is, I have 2 patches, the core Share Migration, and Generic driver support | 18:51 |
| ganso_ | bswartz: the test requires at least 1 driver | 18:52 |
| ganso_ | bswartz: so I cannot include it in core Share Migration patch | 18:52 |
| ganso_ | bswartz: In case of generic driver I need to include it in generic driver patch, or in another patch that depends on it | 18:52 |
| *** eharney has quit IRC | 18:52 | |
| ganso_ | bswartz: what do you think? may I include in generic driver's test? | 18:53 |
| bswartz | ganso_: it's fine for it to be a different patch as long as it's in gerrit and mergeable | 18:53 |
| ganso_ | bswartz: /s/test/patch | 18:53 |
| bswartz | that's how CG's tests were done | 18:53 |
| *** lpetrut has joined #openstack-manila | 18:53 | |
| ganso_ | bswartz: ok, thanks! | 18:54 |
| *** jasonsb has joined #openstack-manila | 18:57 | |
| bswartz | vponomaryov: ping | 18:59 |
| *** martyturner has joined #openstack-manila | 18:59 | |
| openstackgerrit | Rodrigo Barbieri proposed openstack/manila: Share Migration support in generic driver https://review.openstack.org/179791 | 19:00 |
| openstackgerrit | Rodrigo Barbieri proposed openstack/manila: Add Share Migration feature https://review.openstack.org/179790 | 19:00 |
| openstackgerrit | Rodrigo Barbieri proposed openstack/manila: [WIP] Add Share Migration tempest functional tests https://review.openstack.org/220278 | 19:00 |
| *** dzamboni has quit IRC | 19:02 | |
| *** eharney has joined #openstack-manila | 19:07 | |
| *** timcl has joined #openstack-manila | 19:08 | |
| ganso_ | cknight: ping | 19:16 |
| cknight | ganso_: pong | 19:16 |
| ganso_ | cknight: https://review.openstack.org/#/c/220278/ | 19:16 |
| ganso_ | cknight: it has not run yet | 19:16 |
| ganso_ | cknight: but locally it works | 19:17 |
| cknight | ganso_: Thanks, I was just looking at your latest migration patch. I'll look over this next. Appreciate the quick turnaround. | 19:17 |
| cknight | ganso_: I'm still trying to get migration (host-only) working with my driver. It's failing in the access rule update code, but I don't know why yet. I'll send you something when I have it. | 19:18 |
| ganso_ | cknight: ok, thanks! | 19:18 |
| bswartz | xyang1: https://review.openstack.org/#/c/219333/ this isn't a gate related bug and not a BP so it should be lower priority to merge | 19:20 |
| bswartz | I don't have a problem with it merging but please prioritize BPs and bugs related to gate issues | 19:20 |
| xyang1 | bswartz: You want me to remove +A? | 19:21 |
| bswartz | it would help to hold off until next week | 19:22 |
| bswartz | bugs we have more time to get in | 19:22 |
| bswartz | I doubt it would cause a serious problem but since we can merge it next week it's better to wait | 19:22 |
| xyang1 | bswartz: Removed +A | 19:23 |
| *** lpetrut has quit IRC | 19:27 | |
| openstackgerrit | Andrew Kerr proposed openstack/manila: Add functional tests for Manila consistency groups https://review.openstack.org/215346 | 19:28 |
| *** lpetrut has joined #openstack-manila | 19:30 | |
| openstackgerrit | Clinton Knight proposed openstack/manila: Consistency groups in NetApp cDOT drivers https://review.openstack.org/215347 | 19:30 |
| *** akerr has quit IRC | 19:31 | |
| *** lpetrut has quit IRC | 19:36 | |
| bswartz | anyone seen lpetrut? | 19:42 |
| *** eharney has quit IRC | 19:55 | |
| *** ChanServ changes topic to "The gate is broken! Fix is https://review.openstack.org/#/c/219788/" | 19:56 | |
| *** eharney has joined #openstack-manila | 20:00 | |
| *** lpetrut has joined #openstack-manila | 20:21 | |
| *** akerr has joined #openstack-manila | 20:21 | |
| *** akerr has quit IRC | 20:26 | |
| *** eharney has quit IRC | 20:27 | |
| *** tpsilva has quit IRC | 20:27 | |
| *** akerr has joined #openstack-manila | 20:33 | |
| *** lpetrut has quit IRC | 20:38 | |
| *** eharney has joined #openstack-manila | 20:42 | |
| *** pgbridge has quit IRC | 20:49 | |
| *** akerr is now known as akerr_away | 20:59 | |
| *** dustins has quit IRC | 21:05 | |
| *** crobertsrh is now known as _crobertsrh | 21:05 | |
| ameade | xyang1: hey, talked to Ben and I have an idea | 21:08 |
| xyang1 | ameade: Hi | 21:08 |
| ameade | xyang1: so i think it's too late to do teh full fledged protected shares idea atm, so I am thinking i can fall back to what cknight had suggested with the --consistency-group flag on the share delete | 21:08 |
| ameade | then maybe remove that after we have protected shares later on | 21:09 |
| xyang1 | That is fine | 21:09 |
| ameade | xyang1: sorry for the miscommunication, i thought it was clear I was thinking to do the protected shares separate | 21:09 |
| cknight | xyang1: Thanks. Making that switch later is OK because of the experimental API status. | 21:09 |
| xyang1 | Ok, no problem | 21:10 |
| ameade | xyang1: what do you think would work better? having the DELETE call with a body or just having a parameter in the url? | 21:10 |
| ameade | the api-wg is against delete bodies it seems | 21:11 |
| ameade | but they only say it's discouraged | 21:11 |
| xyang1 | ameade: I like it in the body better | 21:12 |
| xyang1 | ameade: If API WG says the other way, I won't be against it either | 21:13 |
| ameade | it says 'Because of this, they may not be supported properly by some client frameworks and we would discourage API methods from accepting request bodies for GET, DELETE, TRACE, OPTIONS and HEAD methods.' | 21:13 |
| ameade | do you think /shares?consistency_group_id=blah is good? | 21:13 |
| ameade | easier to implement too 0:) | 21:14 |
| xyang1 | ameade: So you provide an cg id? | 21:15 |
| xyang1 | ameade: At delete share time | 21:15 |
| ameade | xyang1: yeah exactly | 21:15 |
| xyang1 | Rather than a true false flag | 21:15 |
| ameade | oh i see | 21:16 |
| *** akerr_away has quit IRC | 21:16 | |
| ameade | we could get by with just true/false but the id seems better to me atm for some reason | 21:17 |
| xyang1 | ameade: So that will be more like create share where you need to specify a cg id | 21:17 |
| xyang1 | ameade: Probably fine | 21:17 |
| ameade | kk i'll code that up | 21:18 |
| xyang1 | ameade: Ok, thanks | 21:18 |
| *** gouthamr has quit IRC | 21:18 | |
| *** eharney has quit IRC | 21:26 | |
| *** porrua has quit IRC | 21:27 | |
| bswartz | wow the gate is looking strange | 21:46 |
| bswartz | I've got my eye on 219696 and the integrated gate queue is rendering all strange | 21:46 |
| bswartz | looks like we're still about 6 hours from the nova patch landing | 21:47 |
| * bswartz considers staying up past midnight to watch it | 21:48 | |
| *** haomaiwa_ has joined #openstack-manila | 21:54 | |
| *** haomaiw__ has quit IRC | 21:56 | |
| *** JoseMello has quit IRC | 22:09 | |
| *** alyson_ has quit IRC | 22:10 | |
| *** martyturner has quit IRC | 22:15 | |
| cknight | ganso_: ping | 22:20 |
| ganso_ | cknight: pong | 22:21 |
| cknight | ganso_: So I got migration to work, albeit on an empty share. | 22:22 |
| cknight | ganso_: This is with the NetApp driver. | 22:22 |
| ganso_ | cknight: that's good news :) | 22:22 |
| cknight | ganso_: There are definitely corner cases to address. For example, if the feature isn't correctly configured (such as by omitting migration_data_copy_node_ip), then access_allow on the driver is called with None for the address. | 22:23 |
| cknight | ganso_: That obviously shouldn't happen, and should be caught somewhere above the driver. | 22:23 |
| cknight | ganso_: If that happens, or any of the other issues I hit, and the migration fails, then I'm left in a bad state. | 22:24 |
| cknight | ganso_: In the failure case, I'm left with 2 instances, one of them inactive, and there isn't any way to delete it. | 22:24 |
| cknight | ganso_: The only way to clean up is in mysql. I've deleted the manila DB at least a dozen times today after a failed migration. | 22:25 |
| ganso_ | cknight: that is expected | 22:25 |
| cknight | ganso_: Well, there has to be a way to clean up without resorting to hacking the DB. What is the right way to solve that? | 22:26 |
| ganso_ | cknight: as we discussed during meetings, we agreed to 1) let it crash wherever it crashes and 2) leave it in a bad state, the admin is responsible for cleaning it | 22:26 |
| cknight | ganso_: Sure, an admin must clean it, but not in the DB. Will the instance APIs let me delete the inactive instance? And how do I reset the share task state? | 22:27 |
| ganso_ | cknight: this is why I did not check for "None" etc in many situation, if I did, I would mostly end up throwing an exception and leaving as it was... or, if we agree otherwise, I can cleanup the destination instance... but we agreed not to | 22:27 |
| ganso_ | cknight: I am not familiar with Share Instances Admin API because it was not implemented at the time... I usually end up deleting the share, which deletes the source and destination instance along with it, so I do not have to update DB manually | 22:28 |
| ganso_ | cknight: but Share Instances Admin API should allow you to delete a specific instance | 22:28 |
| ganso_ | cknight: although I am not sure if the "inactive" state blocks it, I have not checked | 22:29 |
| cknight | ganso_: More often than not, that didn't work for me | 22:29 |
| cknight | manila delete b4683ac5-2dd8-4867-99b5-254d41973d24 | 22:29 |
| cknight | Delete for share b4683ac5-2dd8-4867-99b5-254d41973d24 failed: Invalid share: Share b4683ac5-2dd8-4867-99b5-254d41973d24 is busy as part of an active task. (HTTP 403) (Request-ID: req-bc65aaaa-d6f2-4aae-9c1e-f71fcdb73905) | 22:29 |
| cknight | ERROR: Unable to delete any of the specified shares. | 22:29 |
| ganso_ | cknight: that is interesting, I could not reproduce a case where task_state was different than success or error | 22:29 |
| ganso_ | cknight: I am curious to what error you faced that left you in that state | 22:30 |
| cknight | ganso_: Every time I got wedged like this, the only quick recourse was to delete the DB and start clean. | 22:30 |
| cknight | ganso_: It happened repeatedly, if the Copy job failed, or if the controller driver threw an exception because of the None IP. | 22:31 |
| ganso_ | cknight: when it was still WIP, I used to just update the task_state field to get rid of that... but then later I fixed and could not get stuck in that state again | 22:31 |
| ganso_ | cknight: Whenever an exception.ShareMigrationException is thrown, it i captured in manager | 22:31 |
| ganso_ | cknight: *it should be captured in Manager and task_state changed to migration_error | 22:31 |
| cknight | ganso_: OK, I can try to reproduce and provide the steps to you. | 22:32 |
| ganso_ | cknight: great! that will help a lot | 22:32 |
| ganso_ | cknight: I am about to push a new tempest patch | 22:32 |
| ganso_ | cknight: the one I previously pushed was missing microversions | 22:33 |
| ganso_ | cknight: so it did not work | 22:33 |
| cknight | ganso_: OK. Because of all the foreign keys in the DB, it's too much to ask an admin to clean it up manually. The APIs must always provide a way out. | 22:33 |
| ganso_ | cknight: also, I implemented task_state for manila show <id> in api/views... so no change to manila client is nneded | 22:34 |
| cknight | ganso_: Great, thanks. | 22:34 |
| ganso_ | cknight: I agree... but I implemented it this way because we agreed to that... maybe we need to have another meeting or just agree to cleanup everything now, this specifically is a lot of work for tomorrow, it will break several unit tests | 22:35 |
| cknight | ganso_: I'm not suggesting you do a lot of work tomorrow. After all, this is an experimental feature. But we should discuss how to make the feature as resilient as possible, as well as how to minimize the work for an admin having to clean up after a failure. | 22:37 |
| ganso_ | bswartz: I think it is important for you to be aware of this ^ | 22:38 |
| ganso_ | cknight: btw, I was coding a more robust migration tempest test, such as mounting, adding files, etc, but I noticed it is very complicated to create a test that is supposed to run as a main API test, for all drivers, while being so dependent on network and driver configuration | 22:42 |
| ganso_ | cknight: it can be done, but I would need to change the code to expose more information, such as the IP, or have several additional parameters in tempest.conf, just like in manila.conf | 22:43 |
| cknight | ganso_: I don't doubt that! And it will get more so with share servers. Since tempest just uses the Manila API, what code would have to change? | 22:45 |
| openstackgerrit | Rodrigo Barbieri proposed openstack/manila: Share Migration support in generic driver https://review.openstack.org/179791 | 22:45 |
| openstackgerrit | Rodrigo Barbieri proposed openstack/manila: Add Share Migration tempest functional tests https://review.openstack.org/220278 | 22:45 |
| openstackgerrit | Rodrigo Barbieri proposed openstack/manila: Add Share Migration feature https://review.openstack.org/179790 | 22:45 |
| ganso_ | cknight: I would need to have an IP that allows the node running tempest to mount the shares, that could be obtained through an API call, DB, or as config parameter in tempest.conf | 22:47 |
| *** adrianofr has quit IRC | 22:47 | |
| ganso_ | cknight: then it would be possible to have drivers testing migration in all their CIs | 22:47 |
| ganso_ | cknight: in a more robust and complete way | 22:48 |
| ganso_ | cknight: instead of just migrating an empty share | 22:48 |
| cknight | ganso_: That's something you will need anyway with dhss=True migration, right? | 22:48 |
| ganso_ | cknight: I am already doing dhss=true migration, the patch I just submitted for tempest supports only that so far | 22:49 |
| ganso_ | cknight: it may support dhss=false, I have not tested yet | 22:49 |
| cknight | ganso_: I'm testing with dhss=False. With dhss=True, is every share server supposed to make shares accessible on migration_mounting_backend_ip? | 22:51 |
| cknight | ganso_: How does that work with per-tenant Neutron networks? | 22:52 |
| ganso_ | cknight: in dhss=true mode, it is expected that the IP you would normally fill in migration_mounting_backend_ip or migration_data_copy_node_ip to be dynamic... so the best approach for dhss=true drivers is overriding some methods in driver.py base class to obtain those IPs... supposing you have saved them in share_server entity during share_server creation | 22:55 |
| ganso_ | cknight: It was a very simple implementation for the generic driver | 22:55 |
| ganso_ | cknight: #link https://review.openstack.org/#/c/179791/ | 22:55 |
| ganso_ | cknight: save the "service_ip" when creating the share_server and obtain it in _get_access_rule_for_data_copy, by overriding it | 22:56 |
| cknight | ganso_: I see. So I just request an extra IP in get_network_allocations_number, so I have one to use for the service IP. Is that right? | 22:57 |
| ganso_ | cknight: possibly yes, I do not know exactly how NetApp drivers handle their IPs | 22:58 |
| cknight | ganso_: OK, that makes sense to me. NetApp clustered Data ONTAP can have anywhere from 2 - 24 nodes in a cluster, and a share should be accessible from any node. So I request one unique IP for each node. It's no big deal to request one more for the service IP. That sounds simple and elegant, thanks! | 23:00 |
| ganso_ | cknight: :) | 23:00 |
| cknight | ganso_: OK, I gotta run. I will play with this more later and provide more feedback. Thanks for your Tempest work today. | 23:01 |
| ganso_ | cknight: I will be online tonight, ping me if you need anything, I hope we can stay in touch to have feedback on the patches ASAP. | 23:02 |
| ganso_ | cknight: it is important that we fix the stuck task_state scenario you mentioned | 23:05 |
| *** ayma has quit IRC | 23:05 | |
| *** cknight has quit IRC | 23:08 | |
| *** akerr has joined #openstack-manila | 23:17 | |
| *** tbarron is now known as tbarron_afk | 23:19 | |
| ganso_ | bswartz: something is wrong with gate: http://logs.openstack.org/90/179790/36/check/gate-manila-tempest-dsvm-neutron/ae37886/logs/devstacklog.txt.gz | 23:25 |
| ganso_ | bswartz: sorry, I meant our Jenkins | 23:25 |
| *** akerr is now known as akerr_away | 23:27 | |
| *** cknight has joined #openstack-manila | 23:33 | |
| *** akerr_away has quit IRC | 23:39 | |
| bswartz | ganso_: I see that | 23:48 |
| bswartz | is it new failure? is it reproducable? | 23:48 |
| bswartz | it would not be at all surprising if something else changed that broke our gate | 23:49 |
| bswartz | that's what FF week is all about | 23:49 |
| ganso_ | bswartz: I set recheck on https://review.openstack.org/#/c/220204/ to test that | 23:51 |
| ganso_ | bswartz: not sure if it was only my change breaking | 23:52 |
| ganso_ | bswartz: the job is still on devstack | 23:52 |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!