15:00:38 <anteaya> #startmeeting third-party
15:00:39 <openstack> Meeting started Mon Feb  2 15:00:38 2015 UTC and is due to finish in 60 minutes.  The chair is anteaya. Information about MeetBot at http://wiki.debian.org/MeetBot.
15:00:41 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
15:00:44 <openstack> The meeting name has been set to 'third_party'
15:01:16 <anteaya> if you are here for the third-party meeting do say hello
15:02:05 <wznoinsk_> hello
15:02:30 <wznoinsk_> hi anteaya how are you?
15:02:33 <riwinters> hi
15:02:43 <anteaya> wznoinsk: hello
15:02:50 <anteaya> I'm well thank you
15:02:54 <anteaya> and yourself?
15:04:21 <wznoinsk> I'm very good, progressing with 3rd party ci in lxc containers with hugepages here
15:04:30 <anteaya> congratulations
15:04:40 <anteaya> glad progress is a root verb here
15:08:10 <johnma> Hello anteaya, not sure whats the format of the meeting here but I do have some basic questions about setting up a third party CI. I am starting to look at setting up a third party CI for Trove and I need to understand the hardware requirements for the different CI components
15:08:10 <wznoinsk> anteaya: would you know whether any part of jenkins can/runs on a baremetal? my understanding is that everything runs on vm's using nodepool
15:08:27 <anteaya> johnma: hello
15:08:43 <johnma> hello anteaya :)
15:08:43 <anteaya> johnma: the format is whatever we want it to be
15:09:03 <anteaya> johnma: asking questions works so yes go ahead and ask
15:09:18 <johnma> ok, thanks anteaya
15:09:20 <anteaya> johnma: I don't know all the answers but asking them is a very good place to start
15:09:23 <anteaya> johnma: sure
15:09:52 <anteaya> wznoinsk: I don't know, the openstack infrastructure runs on vms
15:10:09 <johnma> right I understand. Plus there are lots of folks who have done this or are in the process of doing this, so it definitely helps to get feedback.
15:10:21 <anteaya> wznoinsk: adam_g is the contact for third party ci for ironic, which is the bare metal driver
15:10:35 <anteaya> wznoinsk: I don't know if adam_g uses nodepool or not
15:10:44 <anteaya> johnma: that it does
15:11:17 <anteaya> wznoinsk: so in short your understanding is correct, and I don't know if it runs on baremetal but adam_g might know
15:11:43 <wznoinsk> ok, thanks, I'll get in contact with him
15:11:59 <anteaya> wznoinsk: sure, then if you find out, do let me know
15:16:27 <anteaya> johnma: did you have some questions?
15:17:06 <johnma> sorry anteaya, got pulled into another call. yes I did
15:17:18 <anteaya> johnma: just wondering if I had lost you
15:17:20 <anteaya> :)
15:19:17 <johnma> so far I am waiting for our PTL to give us the requirements for setting up the third part CI but in the mean time I have been doing my own research and have been getting alot of help. Is there a way to estimate how big of an openstack cloud we will need for nodepool. I mean as far as how many VMs nodepool will need to run the jobs
15:19:52 <anteaya> ahhhh
15:19:53 <johnma> are there any standard numbers for that or does it just depend on the tests and patches for the project
15:20:05 <anteaya> so your PTL is slicknik?
15:20:11 <johnma> yes
15:20:21 <anteaya> it is a per project number
15:20:35 * anteaya goes looking
15:21:03 <johnma> cool, that makes sense
15:21:09 <patrickeast> depends on the project, for example with cinder i have ~6 VM’s available and it very rarely gets backed up in normal usage
15:21:29 <patrickeast> it spikes quite a bit at the end of milestones
15:21:30 <anteaya> so looking here: https://github.com/openstack/trove/graphs/commit-activity
15:21:47 <patrickeast> i saw at k-1 we were getting like 50+ more per day than normal
15:21:47 <anteaya> trove's peak activity is 21 commits per week
15:21:53 <anteaya> not very high
15:22:08 <anteaya> you may not even need a nodepool to start
15:22:20 <johnma> and for each project, does it depend on the number of patches and tests we have to run? Just trying to understand how we determine this
15:22:29 <anteaya> johnma: yes
15:22:52 <johnma> oh ok, nice. good to know how we get to that number
15:22:58 <anteaya> hey patrickeast
15:23:09 <patrickeast> keep in mind though that you will be running more than one test pass per commit, each review usually has a bunch of updates
15:23:20 <anteaya> for comparison here is cinder's graph: https://github.com/openstack/cinder/graphs/commit-activity
15:23:21 <patrickeast> anteaya: heya
15:23:23 <johnma> thanks patrickeast & anteaya
15:24:18 <anteaya> here are the jobs that run on trove: http://git.openstack.org/cgit/openstack-infra/project-config/tree/zuul/layout.yaml#n1482
15:24:45 <anteaya> here are the jobs that run on cinder: http://git.openstack.org/cgit/openstack-infra/project-config/tree/zuul/layout.yaml#n988
15:24:58 <anteaya> these are the jobs infra runs on these projects
15:25:12 <anteaya> but it gives you an idea of what we are running on them
15:26:35 <johnma> greats, thanks anteaya. This sure helps.
15:26:38 <johnma> lets say we start without a nodepool and later the commit rate goes high and we need to include a nodepool, how easy/difficult is it going to be to add this later on. Or is it safe to just go ahead and set it up with a nodepool to begin with
15:26:44 <anteaya> johnma: awesome
15:27:02 <anteaya> johnma: that is the question isn't it
15:27:11 <anteaya> and for that question I have no answere
15:27:20 <patrickeast> johnma: imo just start with nodepool, you can always run it with a single build slave
15:27:35 <anteaya> thank you patrickeast
15:27:36 <patrickeast> johnma: but you get the benefits of being able to scale out and have the single use slaves
15:27:44 <johnma> right, that would be a safe option.
15:27:46 <anteaya> he speaks with the voice of experience
15:28:24 <johnma> and I do appreciate it. Thank you patrickeast. This sure does help
15:28:58 <patrickeast> np
15:30:20 <anteaya> johnma: have you read all of http://ci.openstack.org/third_party.html
15:30:30 <anteaya> to ensure you have the account set up as required?
15:31:32 <johnma> I have read most of it. Yes anteaya. Plus I have also read IBM blog on powerkvm CI and Jay Pipe's blog. All have been very helpful
15:31:42 <anteaya> johnma: awesome
15:32:09 <anteaya> johnma: make sure you read the requirements section carefully and ask any questions if you need clarification
15:32:53 <johnma> I am sure I will have alot more questions when its time to actually setup the system but I am hoping to document the process
15:33:26 <anteaya> johnma: that would be very helpful thank you
15:33:36 <johnma> I will do that anteaya. Thank you so much. I think thats all I have for now. Really appreciate all the insight.
15:33:53 <anteaya> johnma: thanks for taking the time to show up and ask questions
15:34:02 <johnma> sure :)
15:34:07 <anteaya> :)
15:34:13 <anteaya> patrickeast: how are you doing today?
15:34:36 <patrickeast> anteaya: pretty good, yourself?
15:34:53 <anteaya> patrickeast: great, good thanks
15:34:59 <anteaya> at my parents in Tucson
15:35:07 <anteaya> hit the gem show on saturday
15:35:15 <anteaya> between meetings we are going to visit a cave
15:35:38 <patrickeast> ahh nice sounds like a good time
15:35:44 <anteaya> feels good
15:36:02 <anteaya> patrickeast: is there anything I can do to support any of your contribution goals?
15:36:16 <anteaya> can I point you to files or introduce you to anyone?
15:36:22 <anteaya> very glad to have your help
15:36:52 <patrickeast> mmm nothing right now, just focusing on finishing up some outstanding tasks for my cinder driver
15:37:01 <anteaya> patrickeast: cool
15:37:04 <anteaya> let me know
15:37:09 <patrickeast> thanks, will do
15:37:19 <anteaya> want to ensure you are kept fulling entertained and challenged
15:37:31 <anteaya> wouldn't want you to become bored at all
15:37:36 <patrickeast> hehe no problems there
15:37:44 <anteaya> :)
15:38:17 <wznoinsk> johnma: if you want to have a rough feeling on how many patchsets created are per particular project you could subscribe to gerrit stream event using either ssh, jenkins, other CI system and capture it, then run statistics off it after a few weeks
15:41:39 <johnma> wznoinsk, thats good to know. and I guess this helps determine the project load
15:50:03 <anteaya> looks like we are winding down
15:50:15 <anteaya> I'm feeling it might be an idea to tie this off this week
15:50:25 <anteaya> I did so too quickly the other week
15:50:41 <anteaya> so I'll announce my intentions and give it a minute this time
15:51:06 <anteaya> and my breakfast is ready
15:51:12 <anteaya> so I will eat it
15:51:22 <anteaya> thanks all for your attendance and participation today
15:51:29 <anteaya> grateful to have your here
15:51:33 <anteaya> see you all next week
15:51:37 <anteaya> #endmeeting