13:00:31 #startmeeting hyper-v 13:00:32 Meeting started Wed Oct 5 13:00:31 2016 UTC and is due to finish in 60 minutes. The chair is claudiub. Information about MeetBot at http://wiki.debian.org/MeetBot. 13:00:33 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 13:00:36 The meeting name has been set to 'hyper_v' 13:00:47 hellou 13:01:02 gonna wait a bit for people to join 13:01:50 Hi 13:02:39 hello :) 13:02:57 sagar_nikam: anyone else joining? 13:03:49 hi all 13:03:54 hello :) 13:03:56 no 13:04:00 we can start 13:04:17 cool 13:04:30 #topic nova patches 13:04:56 sooo, just fyi, we already merged 2 blueprints in ocata in nova 13:05:05 nice !!! 13:05:07 which ones ? 13:05:20 those are the Hyper-V UEFI one, and the storage QoS one. 13:05:27 #link Hyper-V storage QoS support patch: https://review.openstack.org/#/c/175479/ 13:05:36 #link Hyper-V UEFI Secure Boot https://review.openstack.org/#/c/209581/ 13:05:55 there have been a couple of python3 related patches that merged too 13:06:06 hey guys 13:06:10 any progress on FC ? 13:06:15 in ocata 13:06:15 and I can say that nova-compute on Hyper-V works well with python 3 13:06:41 any perf improvement after moving to python 3 13:06:41 ? 13:07:06 the os-brick in nova patch is still pending. 13:07:21 #link os-brick in nova patch: https://review.openstack.org/#/c/273504/ 13:07:33 the blueprint is approved, and it got some core review on it 13:07:38 soo, yeah 13:08:02 ok 13:08:15 also, for the Hyper-V OVS vif driver, the blueprint is approved, and the patch no longer has a -2. still waiting reviews 13:08:39 as for the python 3 performance impact: we didn't do rally tests against it yet 13:08:52 but python 3 is typically slower than python 2.7 13:09:17 on the other hand, WS 2016 seems to bring some considerable performance improvements 13:09:37 lpetrut: that is irrespective of python 3 ? 13:09:45 yep 13:09:48 yeah, so it can potentially balance it out. :) 13:10:50 as for blueprints, we still have the shielded vms spec up for review. 13:10:59 didn't get much attention 13:11:18 ok 13:11:28 questions? 13:11:35 no 13:11:41 not much as of now 13:11:53 moving on 13:11:58 #topic PyMI release 13:12:07 soo, PyMI 1.0.4 was released. 13:12:15 #topic PyMI release: https://pypi.python.org/pypi/PyMI 13:12:23 feel free to grab it 13:12:37 as mentioned in the previous meetings, it boasts better performance. :) 13:13:12 ok.. we will check it 13:13:57 #topic monasca status 13:14:08 ehm, I have to rebase pretty much everything again :) 13:14:24 also got some comments. will have to check them out. 13:14:58 why ? the last time they were almost fine and just wanted some documentation 13:15:21 don't know if I've mentioned this previously, but I've also sent a patch adding documentaion 13:16:06 yeah, will have to check the comments. it probably won't be anyting big anyways 13:16:10 the last time we discussed this... you attended a meeting with monasca team and they were fine with the code except docs 13:16:23 yep 13:17:13 so new code comments again ? 13:17:25 yep. anyways, will sort it as soon as i can 13:17:44 ok 13:17:54 thanks 13:17:58 np :) 13:18:41 #topic open discussion 13:19:08 hm, so there are a couple of things that I want to mention here 13:20:01 first of all, on your hyper-v compute nodes, you should change the neutron_hyperv_agent.conf's firewall_driver to hyperv.neutron.security_groups_driver.HyperVSecurityGroupsDriver 13:20:57 starting with ocata, the old value will no longer work 13:21:09 ok ... 13:21:21 so this is for ocata only ? 13:22:18 #info The neutron_hyperv_agent.conf firewall_driver config value must be changed to hyperv.neutron.security_groups_driver.HyperVSecurityGroupsDriver when upgrading to Ocata 13:22:41 the old value has been deprecated since mitaka 13:22:51 still works, but in ocata it won't work anymore 13:23:03 for more info, see patch: 13:23:11 ok got it. will mention it to sonu 13:23:20 # Removes remaining Hyper-V plugin in neutron: https://review.openstack.org/#/c/378299/1 13:23:27 #link Removes remaining Hyper-V plugin in neutron: https://review.openstack.org/#/c/378299/1 13:23:45 the docs have been updated as well. 13:24:59 secondly, there has been a bug affecting python 3 on hyper-v, on python-swiftclient, which has been fixed and merged 13:25:08 #link https://bugs.launchpad.net/python-swiftclient/+bug/1628107 13:25:09 Launchpad bug 1628107 in python-swiftclient "Futures package should not be installed for python 3" [Undecided,Fix released] - Assigned to Claudiu Belu (cbelu) 13:26:04 basically, python-swiftclient was installing the futures library in python3, and it shouldn't have, as, ironically, futures is not python 3 compatible 13:26:13 python 3 comes with a native futures library anyways 13:26:26 why is this affecting hyperv ? nova's requirements.txt has it and hence it gets installed on nova compute for hyperv ? 13:27:11 it does, but if you check the requirement, it says that it will only be installed on python 2.6 or python 2.7 13:28:10 futures>=3.0;python_version=='2.7' or python_version=='2.6' # BSD 13:28:29 that line basically means that futures>=3.0 will be installed if the python version is 2.7 or 2.6 13:28:58 ok 13:30:15 also, currently working on solving some of the tests that are failing on the ci 13:30:39 seems i'll have to send a patch up, most probably oslo.messaging 13:31:06 anyways, that's about it from me 13:31:15 sagar_nikam: any news from your side? 13:31:19 one point from me 13:31:31 i am planning to start off on kubernetus 13:31:39 how is it going ? 13:31:46 anything that i can try 13:32:05 atuvenie: hi 13:32:24 heya, what are your plans there sagar? 13:32:58 the aprenda folks have a WIP pull request that is on the K8 repo 13:33:14 I'm helping in the review process with that 13:33:26 first of all.... kubernetus is new to me 13:33:43 i will start some initial learning and trying out basic things 13:33:56 and then if something is ready for windows, i can try that 13:34:01 and provide you feedback 13:34:02 on the other side, I'm still playing with OVN to setup a overlay network like k8 needs, but it's kinda slow 13:34:56 so as of now, basic k8s tests with windows also not a good option ? 13:35:25 i am starting on k8s may be in a week or 10 days.. and will try for few weeks 13:35:32 and then i can try with windows 13:35:33 well, you can use the pull request that is currently under review 13:35:55 so may be a month from now, i can try k8s for windows 13:35:59 using that, I can create pods but networking is still a issue for me 13:36:04 any docs available 13:36:10 on how to use it ? 13:36:44 and containers are windows containers ? 13:38:06 containers are indeed windows containers 13:38:13 have not tried with hyper-v containers 13:38:17 ok 13:38:28 so let me try k8s on linux first 13:38:36 just basic hands on 13:38:41 to get familar 13:38:54 and then may be i will reach out to you to try on windows 13:39:06 is that fine ? 13:39:32 sure 13:39:52 ok thanks 13:40:17 next question 13:40:36 claudiub: any plans on monasca-log ? 13:40:55 sagar_nikam: didn't look at it yet 13:41:08 ok 13:41:13 is that planned ? 13:41:27 did the monasca team mention anything sbout it 13:41:31 about it 13:41:48 no they haven't. it is a bit low on my priority list, to be honest 13:42:33 my main concern is nova and work on it. when I don't have anything to do there, and i have some free time, I can take a look 13:42:53 sagar_nikam: any news from the testing lab? 13:43:00 ok sure got it 13:43:11 no... getting slot has been difficult 13:43:28 may be pushed by some more time 13:43:38 ack 13:43:38 will let you know as and when we start it 13:43:58 we plan to test it on mitaka 13:44:20 sounds good 13:44:27 i was told by sonu that some neutron-agent code may not scale 13:44:32 we are not sure though 13:44:57 the last time, sonu's team picked some code from mitaka and put it in liberty 13:45:04 and found sclability issues 13:45:11 we reverted it back to liberty 13:45:24 and we achieved good performance 13:45:31 may be it just a backport issue 13:45:35 cant confirm 13:45:40 there shouldn't have been any performance regressions from liberty to mitaka 13:46:02 from our tests, mitaka was performing significantly better than liberty 13:46:19 i think we would have taken only some pieces of code 13:46:20 we presented those results a few months ago 13:46:35 we will try with full mitaka 13:46:44 rather than just some code backported 13:47:21 ok 13:47:31 well, that's about it from me 13:47:42 any other questions from you, sagar_nikam? 13:47:56 no not as of now 13:48:23 well then, we can end the meeting soon. :) 13:48:23 how is the summit plans going ? 13:48:50 well, we're currently working on demo's and other cool booth setups 13:49:16 hopefully it will be an enjoyable summit. :) 13:49:30 because after this, the summits won't be the same 13:49:37 as the summit will split into 2 13:49:41 why ? 13:49:50 oh ... what is that 13:49:53 dont know about it 13:49:55 into the general public / marketplace summit, and the design summit 13:50:02 ok 13:50:12 which will be in 2 separate locations, in 2 separate dates 13:50:17 i don't really know the reason 13:50:47 ok 13:51:03 i am done with all topics... thanks 13:51:11 well then, thanks folks for joining, see you next week! 13:51:33 #endmeeting