Thursday, 2014-11-20

« Wednesday, 2014-11-19 Index Friday, 2014-11-21 »
SlickNikHey guys: Quick reminder — please fill in the poll at http://doodle.com/ahrvmneuddzi92kn to figure out the dates for the Trove Mid Cycle Sprint.00:06
*** sushilkm has left #openstack-trove00:47
*** sushilkm has joined #openstack-trove00:50
*** cweid_ has quit IRC00:51
*** annashen has quit IRC00:54
*** IanGovett has quit IRC00:55
*** mattgriffin has joined #openstack-trove01:06
*** grapex has quit IRC01:12
*** mattgriffin has quit IRC01:24
*** Riddhi has joined #openstack-trove01:45
*** boblebauce has quit IRC01:46
*** haomaiwang has quit IRC01:47
*** radez_g0n3 is now known as radez01:56
*** vkmc has quit IRC01:57
*** nosnos has joined #openstack-trove01:59
*** haomaiwang has joined #openstack-trove02:07
*** Longgeek has joined #openstack-trove02:33
*** X019 has quit IRC02:33
*** erkules_ has joined #openstack-trove02:35
*** erkules has quit IRC02:38
*** sushilkm has quit IRC02:38
*** Longgeek has quit IRC02:40
*** X019 has joined #openstack-trove02:46
*** ViswaV_ has quit IRC03:10
*** nosnos has quit IRC03:24
*** rushiagr_away is now known as rushiagr03:27
*** Longgeek has joined #openstack-trove03:36
*** Longgeek has quit IRC03:41
*** mattgriffin has joined #openstack-trove03:47
*** nosnos has joined #openstack-trove03:49
*** Riddhi has quit IRC03:53
*** rushiagr is now known as rushiagr_away03:53
*** mattgrif_ has joined #openstack-trove04:04
*** mattgriffin has quit IRC04:05
*** radez is now known as radez_g0n304:12
*** mattgrif_ has quit IRC04:17
*** mattgriffin has joined #openstack-trove04:22
*** annashen has joined #openstack-trove04:30
*** rushiagr_away is now known as rushiagr04:33
*** erkules_ is now known as erkules04:35
*** Longgeek has joined #openstack-trove04:37
*** Longgeek has quit IRC04:42
*** mattgriffin has quit IRC04:49
*** mattgriffin has joined #openstack-trove05:09
*** annashen has quit IRC05:31
*** annashen has joined #openstack-trove05:32
*** annashen has quit IRC05:35
*** ajayaa has joined #openstack-trove05:57
*** Longgeek has joined #openstack-trove06:01
*** Longgeek has quit IRC06:02
*** Longgeek has joined #openstack-trove06:03
*** aj9 has joined #openstack-trove06:12
*** exploreshaifali has joined #openstack-trove06:13
*** haomaiwang has quit IRC06:17
*** amcrn has quit IRC06:24
*** harlowja is now known as harlowja_away06:31
*** k4n0 has joined #openstack-trove06:33
*** vigneshvar has joined #openstack-trove06:55
*** mattgriffin has quit IRC06:57
*** mattgriffin has joined #openstack-trove07:03
*** dloi has quit IRC07:12
*** mattgriffin has quit IRC07:13
*** dloi has joined #openstack-trove07:13
*** romainh has joined #openstack-trove07:15
*** pboros has joined #openstack-trove08:20
*** aj9 has quit IRC08:22
*** newb has quit IRC08:24
*** boblebauce has joined #openstack-trove09:29
*** Longgeek has quit IRC09:32
*** Longgeek has joined #openstack-trove09:41
*** exploreshaifali has quit IRC09:58
*** boblebauce has quit IRC10:57
*** boblebauce has joined #openstack-trove11:06
*** boblebauce has quit IRC11:15
*** newb has joined #openstack-trove11:33
*** Longgeek has quit IRC11:39
*** radez_g0n3 is now known as radez11:59
*** rushiagr is now known as rushiagr_away12:15
*** Longgeek has joined #openstack-trove12:19
*** IanGovett has joined #openstack-trove12:41
*** tomblank has quit IRC12:44
*** rushiagr_away is now known as rushiagr12:46
*** haomaiwa_ has joined #openstack-trove13:00
*** newb has quit IRC13:01
*** nosnos has quit IRC13:05
*** nosnos has joined #openstack-trove13:06
*** X019 has quit IRC13:14
*** boblebauce has joined #openstack-trove13:15
*** vigneshvar has quit IRC13:21
openstackgerritDenis M. proposed openstack/trove-specs: Added Cassandra cluster spec  https://review.openstack.org/12273613:30
*** newb has joined #openstack-trove13:31
*** newb has quit IRC13:35
*** k4n0 has quit IRC13:38
*** miqui has joined #openstack-trove13:46
*** miqui_ has joined #openstack-trove13:50
*** miqui__ has joined #openstack-trove13:52
*** ajayaa has quit IRC13:53
*** miqui___ has joined #openstack-trove13:54
*** miqui____ has joined #openstack-trove13:55
*** miqui_____ has joined #openstack-trove13:55
*** miqui___ has quit IRC13:56
*** miqui___ has joined #openstack-trove13:57
*** miqui__ has quit IRC13:58
*** miqui has quit IRC13:58
*** miqui____ has quit IRC13:58
*** miqui_ has quit IRC13:58
*** miqui___ has quit IRC13:58
*** rushiagr is now known as rushiagr_away13:59
*** miqui has joined #openstack-trove13:59
*** amrith is now known as _amrith_14:00
*** tomblank has joined #openstack-trove14:00
openstackgerritDenis M. proposed openstack/trove-specs: Added Cassandra cluster spec  https://review.openstack.org/12273614:02
*** ajayaa has joined #openstack-trove14:06
*** rushiagr_away is now known as rushiagr14:07
*** miqui has quit IRC14:14
*** miqui has joined #openstack-trove14:14
*** miqui_ has joined #openstack-trove14:15
*** jcru has joined #openstack-trove14:16
*** miqui_ has quit IRC14:17
*** nosnos has quit IRC14:34
*** newb has joined #openstack-trove14:41
*** _amrith_ is now known as amrith14:51
*** miqui has quit IRC15:00
*** isviridov is now known as isviridov_away15:05
*** miqui has joined #openstack-trove15:05
*** robertmyers has joined #openstack-trove15:06
*** Barker has joined #openstack-trove15:17
*** miqui_____ has quit IRC15:18
*** miqui_ has joined #openstack-trove15:22
*** miqui has quit IRC15:25
*** mattgriffin has joined #openstack-trove15:27
*** miqui has joined #openstack-trove15:31
*** miqui has quit IRC15:32
*** miqui has joined #openstack-trove15:34
*** grapex has joined #openstack-trove15:35
*** miqui has quit IRC15:35
*** X019 has joined #openstack-trove15:37
*** Riddhi has joined #openstack-trove15:39
*** Riddhi has quit IRC15:42
*** Riddhi has joined #openstack-trove15:43
*** miqui has joined #openstack-trove15:44
*** jmontemayor has joined #openstack-trove15:47
*** Riddhi has quit IRC15:57
*** tomblank_ has joined #openstack-trove16:04
*** tomblank has quit IRC16:05
*** tomblank_ is now known as tomblank16:05
*** amrith is now known as _amrith_16:07
*** thedodd has joined #openstack-trove16:10
*** __TheDodd__ has joined #openstack-trove16:18
*** thedodd has quit IRC16:19
*** _amrith_ is now known as amrith16:21
*** vigneshvar has joined #openstack-trove16:24
*** __TheDodd__ has quit IRC16:34
grapexamcrn SlickNik: Can take a gander at this at your leisure? https://review.openstack.org/#/c/132367/16:34
*** tomblank has quit IRC16:38
*** Longgeek has quit IRC16:45
*** clarkb has joined #openstack-trove16:45
*** clarkb has left #openstack-trove16:45
*** Riddhi has joined #openstack-trove16:51
*** amrith is now known as _amrith_16:54
denis_makogongrapex, SlickNik, guys, we have critical issue, https://bugs.launchpad.net/trove/+bug/139465416:55
grapexdenis_makogon: Hmmm16:56
*** anna_ has joined #openstack-trove16:57
denis_makogongrapex, there are lots of issues in taskmanager around volumes, i file like 4 of them, in last 3 hours16:57
denis_makogongrapex, and it's a shame, because we don't have integration tests that are examining code with disabled volumes =/16:58
*** X019 has quit IRC16:59
*** tomblank has joined #openstack-trove17:00
denis_makogongrapex, same for user data and config drive =)17:02
denis_makogonomg =))17:02
grapexAlso robertmyers just pointed out that codepath also has -- in it17:02
grapexdenis_makogon: You're right,that code path is bad17:03
grapexThough my guess is this isn't affecting anyone and we should look into deprecating it.17:03
denis_makogongrapex, but it would take lot more time to deprecate it17:03
denis_makogongrapex, in Mirantis, we need to make this codepath working17:04
*** tomblank has quit IRC17:04
denis_makogongrapex, robertmyers, you might take a look https://bugs.launchpad.net/trove/+bug/1394650 https://bugs.launchpad.net/trove/+bug/1394625 https://bugs.launchpad.net/trove/+bug/139459817:05
denis_makogonall about volumes17:05
denis_makogonthose bugs are affecting replication, clustering too17:07
denis_makogonactually i faced with them when was playing around mongo sharding17:07
*** annashen has joined #openstack-trove17:08
denis_makogoni'd say that we need to fix codepath first to make it working, and then deprecate17:08
denis_makogonand to test, lot more tests =)17:08
openstackgerritDenis M. proposed openstack/trove: Fix broken instance provisioning with disabled volume support  https://review.openstack.org/13606217:13
*** exploreshaifali has joined #openstack-trove17:14
denis_makogongrapex, hey, Tim, could you please review given patch (please do to -2), https://review.openstack.org/136062 - is there a way to get it in upstream during today/tomorrow ?17:16
grapexdenis_makogon: Have you run that yet?17:18
denis_makogongrapex, trying it right now17:21
denis_makogongrapex, will need to talk to you later about floating IPs design in Trove, is it ok for you ?17:22
denis_makogongrapex, or, even better, could you give me your email?17:23
denis_makogongrapex, my patch works fine as hell!17:24
*** anna_ is now known as X01917:25
robertmyersdenis_makogon: is hell really fine?17:27
*** romainh has quit IRC17:27
denis_makogonrobertmyers, pretty warm, i guess =)17:27
grapex+2'd the first bug17:29
denis_makogongrapex, there's another one on its way17:29
denis_makogonjust finishing description17:30
*** tomblank has joined #openstack-trove17:30
grapexdenis_makogon: I changed this to incomplete from confirmed: https://bugs.launchpad.net/trove/+bug/139459817:34
grapexI think you might be mistaken17:34
grapex"use_nova_server_volume" is only necessary if you're using an ancient version of Nova that allows you to pass a volume size in through the create call17:34
grapexit should not be necessary to set it to True if you're not using a volume17:34
denis_makogongrapex, do we still have to maintain this option?17:35
grapexdenis_makogon: I'd like to. Unfortunately, we still have a few places we're using it here. :-/17:35
*** mattgriffin has quit IRC17:36
denis_makogonok17:36
openstackgerritPavlo Shchelokovskyy proposed openstack/trove: Use native Heat template format and resources  https://review.openstack.org/11203517:37
*** boblebauce has quit IRC17:37
denis_makogongrapex, could you please check this one https://bugs.launchpad.net/trove/+bug/139468217:38
grapexdenis_makogon: My first question  on that bug is how do companies not using volumes run this code?17:41
grapexSlickNik vipul: Hp runs Reffy, but doesn't use volumes right? So how can this be a bug?  https://bugs.launchpad.net/trove/+bug/139468217:41
denis_makogongrapex, that's the question i've been asking myself during whole day17:41
*** mattgriffin has joined #openstack-trove17:42
denis_makogongrapex, i've been hitting this bugs with Trove deployed over Nova + ESXI/VMWare17:42
grapexdenis_makogon: Have you tried setting device_path to an empty string?17:43
denis_makogongrapex, i was trying to do everything, guest will fail because of incorrect arguments for blockdev CLI util17:44
grapexI don't understand17:45
denis_makogoni'm looking now how can we fix it fast with less code17:48
grapexI get you want to fix a problem of some kind. I don't understand the problem you're addressing.17:50
grapexAre you using volumes?17:50
denis_makogonno17:50
denis_makogoni'm not using volumes in this type of deployment17:51
denis_makogondue to some customer needs17:51
denis_makogonand of course due to certain problem in Trove itself, like 'virtio' against iSCSI, so each volume on iSCSI should have block mapping device that starts with 'sdb' but not 'vdb'17:52
grapexOk- so won't setting "device_path" to an empty string solve it?17:52
denis_makogonit doesn't17:53
grapexBecause you're mentioning ISCSI but then saying you're not using volumes. I'm confused17:53
denis_makogongrapex, so, we're not using volumes for Trove17:53
grapexOk- then how is "device_path" being used if not for volumes?17:54
vipulgrapex: we run without volumes, but our flavors have an ephemeral drive, so it is available as /dev/vdb17:54
grapexvipul: I see, so you're using device_path?17:54
vipulyes,17:54
denis_makogonhm17:55
denis_makogonneed to speak with deployers17:55
vipulour device path is vdb.. not sure if that makes it work17:55
grapexvipul: So its not a volumne, but a drive17:55
grapexI mean its a "Device" still17:55
grapexOk17:55
grapexmaybe the terms could be better inside the guest agent17:55
vipulright, if your flavor has ephemeral specified, nova will map a device17:56
*** tomblank has quit IRC17:56
vipulwe haven't tested flavors without ephemeral .. so maybe that's where the bug is17:56
grapexvipul: Got it- thanks17:56
denis_makogonTrove API doesn't allow to use flavors without ephemeral storage iirc17:57
grapexI think the "VolumeDevice" class should really be created by a factory function of some kind17:57
openstackgerritPeter Stachowski proposed openstack/trove-integration: Support new devstack LIBS_FROM_GIT env variable  https://review.openstack.org/13607817:57
grapexhmm, this seems like a huge can of works17:57
grapexdenis_makogon: I think you should write a bigger spec about this. We need to make sure we all understand the problem first17:57
grapexbrb17:57
*** harlowja_away is now known as harlowja18:02
*** Barker has quit IRC18:03
*** Barker has joined #openstack-trove18:06
*** jmontemayor has quit IRC18:09
denis_makogonsure, i'll do it once i'd have working installation18:09
denis_makogongrapex, setting device_path to '' works18:12
*** amcrn has joined #openstack-trove18:16
*** ViswaV has joined #openstack-trove18:22
*** ViswaV_ has joined #openstack-trove18:26
*** ViswaV has quit IRC18:26
*** radez is now known as radez_g0n318:29
*** sushilkm has joined #openstack-trove18:34
*** _amrith_ is now known as amrith18:39
*** ajayaa has quit IRC18:41
*** rushiagr is now known as rushiagr_away18:45
*** ViswaV_ has quit IRC18:45
*** jdaggett_ is now known as jdaggett18:47
*** Viswanath has joined #openstack-trove19:00
*** boblebauce has joined #openstack-trove19:01
*** eghobo has joined #openstack-trove19:06
*** Viswanath has quit IRC19:07
*** boblebauce has quit IRC19:07
amrithSlickNik, amcrn when's a good time to pick up the conversation on add-shard and deprecating it in favor of a new abstraction for clustering and replication etc?19:09
amrithSlickNik, amcrn ^^19:09
amcrni've got 20 minutes right now19:10
amrithok, if SlickNik comes along he can join.19:10
amrithI dug up some old notes from discussions at mid-cycle19:10
amriththe thought then, and it appears still valid is this19:10
amrithfor all datastores we could think of19:11
amrithand this included mysql and mongodb for sure19:11
amriththe notion of adding a replica and adding to a cluster were sufficient abstractions19:11
amrithin the mongodb case19:11
amrithadding a shard is equivalent to growing the cluster19:11
amrithand adding a node is only meaningful in the context of adding a replica19:11
amriththerefore the options are still replica or cluster19:12
amrithwe could choose another name for cluster19:12
amrithif it is not a good one19:12
amrithbut the two abstractions still hold19:12
amrithadding replicas (and in mongo you can have what, 50?)19:12
amrithand adding shards would be to add to the 'cluster'19:12
amrithhence the 'add-shard' which is mongo-specific can be eliminated.19:13
*** eghobo has quit IRC19:13
amcrnit's difficult to clearly articulate why this won't work well, especially without req/resp examples. if this proposal can be mocked with how to handle (1) adding an arbiter [mongo] (2) adding a slaveDelayed member [mongo] (3) adding a non-voting secondary [mongo] (4) adding a shard  [mongo] (5) adding a node [cassandra] (6) replacing a config-server [mongo] (7) replacing a node [cassandra] (8) repairing a node [cassandra] (9) etc. etc.19:16
amcrnit'll be easier to point to distinct problems vs. abstracts.19:16
amrithok19:18
amrithhow then does one add a new query router today?19:18
amrithor a config server?19:18
amrithin the mongodb context?19:18
amcrnit's not supported in the first iteration, however the design did anticipate how it *would* be done in the future19:18
amrithand would you expect those non-data nodes that are part of a cluster to also be handled in this abstraction?19:19
amrith(add-shard abstraction)19:19
amrithor would there be a new API for that?19:19
amcrnnot necessarily. my concern is that the original specification anticipated how to handle alll of the aforementioned examples, whereas this new proposal has not19:20
amcrnif it can be documented how it's possible, and it looks good, then we can consider and proceed19:20
amrithok, do you have thoughts or input on how you would like this common abstraction should look?19:21
amcrnwe're stuck with this api for a very long time, so i don't think it's wise to approve a piece of the clustering api in a vacuum, only to realize later that other abstractions or node-types or whatever don't play nicely with it19:21
amcrnfor the new proposal? no, because i'm fairly happy with the existing direction19:22
amrithexactly the reason why I don't want to be stuck with "add-shard".19:22
amrithand the proliferation of datastore specific things19:22
amcrnunderstand your concerns, but in short, a competiting proposal for me needs to illustrate how "all things clustering" will be modeled19:22
amcrncompeting*19:22
amritheven if we resign ourself to a per-datastore API, we should think sooner (rather than later) of how we will organize them in such a way that they won't all be "top level".19:23
*** vigneshvar has quit IRC19:23
amrithsuch as a naming convntion that would make it clear that it was mongodb-add-shard19:23
amrithrather than just having add-shard and later requiring a different one for MySQL shards (if we went that route).19:23
amcrnyou might be right, but without a fully written out spec, it's hard to say.19:23
amcrnif you feel strongly about an option, write a spec, and i'll definitely set aside some time to digest it and provide feedback.19:25
amcrni think that's the way forward. otherwise it'll be a lot like the original clustering api discussions: it gets mired in abstractions.19:26
*** eghobo has joined #openstack-trove19:26
amrithok, sounds good.19:27
*** exploreshaifali has quit IRC19:27
amcrncool, looking forward to getting this all ironed out19:28
*** rwsu has quit IRC19:32
*** rwsu has joined #openstack-trove19:33
*** vigneshvar has joined #openstack-trove19:35
*** edmund has joined #openstack-trove19:38
*** radez_g0n3 is now known as radez19:44
grapexamcrn: Can have +2? https://review.openstack.org/#/c/132367/19:57
*** Barker has quit IRC19:59
*** jmontemayor has joined #openstack-trove20:09
*** jmontemayor_ has joined #openstack-trove20:11
*** Barker has joined #openstack-trove20:14
*** jmontemayor has quit IRC20:14
*** grapex has quit IRC20:20
*** exploreshaifali has joined #openstack-trove20:33
*** vigneshvar has quit IRC20:38
*** grapex has joined #openstack-trove20:44
*** david-lyle has joined #openstack-trove21:16
amrithSlickNik, when you get a chance would you please ping me re: https://review.openstack.org/9162921:18
*** exploreshaifali has quit IRC21:20
*** sriram_tesora has joined #openstack-trove21:23
*** fifieldt__ has joined #openstack-trove21:31
*** Riddhi has quit IRC21:33
*** fifieldt_ has quit IRC21:34
*** pboros has quit IRC21:41
*** newb has quit IRC21:42
*** jmontemayor_ has quit IRC21:43
*** Barker has quit IRC21:45
*** eghobo has quit IRC21:46
*** amrith is now known as _amrith_21:50
*** eghobo has joined #openstack-trove21:52
*** grapex has quit IRC21:58
*** grapex has joined #openstack-trove21:59
*** mattgriffin has quit IRC22:02
*** dkehnm has quit IRC22:03
*** dkehn has quit IRC22:03
*** dkehnm has joined #openstack-trove22:03
*** grapex has quit IRC22:03
*** dkehn has joined #openstack-trove22:05
*** robertmyers has quit IRC22:05
*** denis_makogon_ has joined #openstack-trove22:06
*** denis_makogon has quit IRC22:08
*** denis_makogon_ is now known as denis_makogon22:09
*** dmakogon_ has joined #openstack-trove22:09
denis_makogonamcrn, ping22:11
amcrndenis_makogon: hi22:11
denis_makogonamcrn, got question to you22:11
denis_makogonamcrn, what kind of hypervisor do you use for Trove ?22:12
*** radez is now known as radez_g0n322:12
denis_makogonkvm or something more exotic ?22:12
*** rwsu has quit IRC22:12
amcrni'll abstain from diving into arch details on a publicly logged irc channel :)22:12
amcrnpm me22:13
*** Barker has joined #openstack-trove22:13
*** radez_g0n3 is now known as radez22:14
*** rwsu_ has joined #openstack-trove22:15
*** rwsu_ has quit IRC22:19
denis_makogonSlickNik, ping22:20
openstackgerritBob Thyne proposed openstack/python-troveclient: Adds support for Keystone V3 API  https://review.openstack.org/13412622:30
*** tomblank has joined #openstack-trove22:32
*** Barker has quit IRC22:44
*** dkehn has quit IRC22:48
*** sriram_ has joined #openstack-trove22:49
*** dkehn has joined #openstack-trove22:50
*** sriram_tesora has quit IRC22:53
*** IanGovett has quit IRC22:57
*** eghobo has quit IRC23:00
*** IanGovett has joined #openstack-trove23:04
*** sriram_ has quit IRC23:07
*** denis_makogon has quit IRC23:11
*** dkehn has quit IRC23:17
*** dkehn has joined #openstack-trove23:18
*** vkmc has joined #openstack-trove23:21
*** vkmc has joined #openstack-trove23:21
*** edmund has quit IRC23:26
*** jcru has quit IRC23:46
*** haomaiwa_ has quit IRC23:46
*** IanGovett has quit IRC23:51
« Wednesday, 2014-11-19 Index Friday, 2014-11-21 »

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