14th December 2010

People Present

  1. iansmith
  2. zyga
  3. plars
  4. mwhudson
  5. asac
  6. salgado
  7. james_w

Agenda

Action Items from last Meeting

Action

Due

Assigned to

Status

Action Items this Meeting

Action

Due

Assigned to

Status

Minutes IRC (Transcript)

Started logging meeting in #linaro-meeting
[20:02:06] <iansmith> OK, let's start!
[20:02:15] <zyga> ok
[20:02:18] <iansmith> [TOPIC] Actions
[20:02:30] <iansmith> -none - alldone!
[20:02:57] <iansmith> [TOPIC] Progress
[20:03:11] <iansmith> zyga: Did you want to go?
[20:03:15] <zyga> yes thanks
[20:03:18] <zyga> okay I'll be bried
[20:03:21] <zyga> brief
[20:03:56] <zyga> I'm working with deepti to finish her first contribution to l-c, she is on good progress to submit a merge request today, the only missing bit is unit tests that we were working on today
[20:04:10] <zyga> this week is devoted to getting gcc performance data into d.l.o
[20:04:36] <zyga> I'm mirroring the data using wget, I read all the scripts that make it and I'm confident I can convert that to dashboard bundles this week
[20:04:52] <plars> zyga: this is the performance data pulled directly from what michaelh has already run? or are these new runs with a script to have it feed into d.l.o from now on?
[20:04:54] <zyga> I hope to integrate this into michael's workflow so that it's a constant stream of incoming data
[20:05:08] <zyga> plars, the former
[20:05:17] <plars> any plans to migrate towards the latter?
[20:05:22] <zyga> I contacted speed.pypy.org upstream for collaboration
[20:05:34] <zyga> plars, ah sorry it's actually both
[20:05:37] <zyga> I'm working on old data
[20:05:46] <plars> cool
[20:05:47] <zyga> and I have a script that just converts the log files into bundles
[20:05:53] <zyga> so data == raw data
[20:06:03] <zyga> as soon as it's ready michael agreed to use it
[20:06:06] <plars> that's good for an interim step, we certainly want to capture what has already been done there if possible
[20:06:10] <zyga> it should be packaged and ready in a ppa
[20:06:14] <plars> sweet
[20:06:31] <zyga> I talked with pypy guys, they have some nice benchmark infrastructure
[20:06:33] <iansmith> zyga: Is that everything? If so, did you have Any Other Business - so you can put the children to sleep?
[20:06:48] <zyga> they want to collaborate but it's not a finished topic, I'll do more this week
[20:06:52] <zyga> that's all from me right now
[20:06:56] * zyga rushes downstairs
[20:07:28] <iansmith> mwhudson: Sorry - I've just seen you are around too! Hi!
[20:07:42] <mwhudson> hi!
[20:07:55] <iansmith> mwhudson: Did you want to go next?
[20:07:59] <mwhudson> ok
[20:08:14] <mwhudson> i've mostly been working on improving the way hwpacks get built and installed
[20:08:32] <mwhudson> to record which hwpacks are installed in the image
[20:08:37] <mwhudson> make the upgrading story nicer
[20:08:48] <mwhudson> and allow more than one pack to be installed at once
[20:09:06] <iansmith> mwhudson: That sounds like very useful additions.
[20:09:20] <mwhudson> all the code is done and landed for this
[20:09:40] <mwhudson> so i guess i need to look at making use of it
[20:09:50] <iansmith> mwhudson: Excellent! Nice to have that complete before the holidays :-)
[20:09:51] <asac> what are the improvements to hwpack building? just code cleanup etc.? or something else?
[20:10:07] <mwhudson> asac: various, i can talk to you about in #linaro in a bit?
[20:10:14] <asac> sure
[20:10:18] <asac> thx
[20:10:28] <mwhudson> i only have three days left of work (including today)
[20:10:31] <mwhudson> this year
[20:11:09] <iansmith> mwhudson: If it isn't documented on e.g. LaunchPad, it may be worth putting something in writing, so you don't keep having to repeat the work you have done!
[20:11:30] <asac> its most likely in the spec
[20:11:31] <mwhudson> iansmith: also so i don't forget :-) good idea
[20:11:35] <asac> just wanted quick idea
[20:12:09] <iansmith> asac: Sorry - didn't want to deny you - more thinking of longer term :-)
[20:12:15] <asac> sure
[20:12:50] <iansmith> salgado: Did you want to tell us what you have been doing - before your nice holiday? :-)
[20:12:58] <salgado> sure!
[20:13:27] <salgado> I basically ported lots more stuff from linaro-media-create to python. it's unlikely that it will be completely ported before I leave on holidays (Friday afternoon), though
[20:14:03] <iansmith> salgado: So will this work be completed shortly after we start back in the New Year?
[20:14:17] <salgado> iansmith, yes
[20:15:06] <iansmith> salgado: Would you estimate the first or the second week back (when we are in warmer places - for zyga, james_w and me)?
[20:15:48] <salgado> iansmith, the porting should be completed on the first week, but there are some other cleanups that might take parts of the second week as well
[20:16:48] <iansmith> salgado: So we can confidently tell the stakeholders that by the end of the rally, this project should be complete - or are there little additions to be made afterwards?
[20:17:40] <salgado> I think it's fair to say that
[20:18:06] <iansmith> salgado: Many thanks!
[20:18:10] <iansmith> plars: Did you want to go next (I trust you have recovered from your travels)?
[20:18:23] <plars> yep
[20:18:35] <plars> Spent the first few days last week at a qa related sprint, then the rest of the week at Linaro working on setting up hardware for automated testing
[20:18:52] <plars> Hardware was a little late, so I didn't get as much done as I would have liked, but it is mostly setup and configured now, with a BeagleXM and a Vexpress board for the moment (more to come soon I hope)
[20:19:04] <plars> Some more details about this can be found in the spec here:
[20:19:14] <plars> https://wiki.linaro.org/Platform/Validation/Specs/HardwareSetup which is still being developed
[20:19:22] <plars> I have hudson running, but not doing anything just yet
[20:19:37] <plars> I should have it at least replicating what I had at home (before my server died) very soon
[20:19:50] <plars> Which is simply pulling the latest image each day and running it under qemu
[20:20:06] <plars> followed soon by trying to get some regular testing running under that and on the hardware there
[20:20:16] <plars> Also working on getting some other specs written for new work related to QA/Validation
[20:20:47] <plars> that's about it for now
[20:20:55] <iansmith> plars: Is the hardware running in the Linaro offices? If so, do we have access to the boards there for general access?
[20:21:00] <james_w> plars, could we use that hudson instance for other tasks (like the originally planned WG tests?)
[20:21:18] <mwhudson> plars: do the board have nfs roots or something?
[20:21:23] <plars> iansmith: I can get access for those who need it on the team, on an as needed basis for now
[20:21:34] <mwhudson> or are we envisioning robots that move sd cards around?
[20:21:34] <plars> james_w: maybe, that's TBD
[20:21:43] <james_w> plars, ok
[20:21:51] <iansmith> plars: Thanks - I think I may speak to you offline about this!
[20:22:13] <plars> mwhudson: no, what I'm currently looking at doing is having it boot from a stable partition, writing the boot/rootfs to test partitions, and booting from those by way of serial control of uboot
[20:22:28] <plars> there are issues with this approach, mainly that vexpress is having some trouble
[20:22:32] <plars> dmart is helping with that though
[20:22:47] <plars> mwhudson: this doesn't require any manual interaction though
[20:22:52] <plars> and no robots
[20:22:54] <plars> sadly :(
[20:22:55] <plars> :)
[20:22:58] <mwhudson> ok cool
[20:23:00] <mwhudson> heh heh
[20:23:09] <iansmith> plars: Many thanks. It sounds like it was all work and no play at Linaro :-)
[20:23:15] <plars> remote control of robots in the linaro office would be really fun though
[20:23:49] <plars> we do have remote power control over all the boards as well
[20:23:58] <iansmith> plars: I used to design mechanical robots - so I have this image of a physical robot - not a software robot :-)
[20:24:11] <plars> so we should also be able to automate the recovery process if a board becomes hung or panics during a test
[20:24:21] <iansmith> plars: Excellent!
[20:25:02] <iansmith> james_w: What have you been doing this week?
[20:25:35] <james_w> I have been working on the blueprint tracker, trying to get it up and running again after we changed the way we used blueprints this cycle
[20:25:50] <james_w> that seems to be going well, with the only known issues being with the blueprints themselves
[20:26:15] <asac> thanks for getting this going!!
[20:26:24] <iansmith> james_w: And the burndown charts are also working now, aren't they?
[20:26:34] <asac> next is to get milestones working, right?
[20:26:42] <james_w> however, it's currently running in my homedir, and not sharing the instance or code with platform, as I haven't been able to keep the behaviour the same for Ubuntu.
[20:27:00] <james_w> plus, there are still many things wrong with the code that mean we can't get all the information that we want
[20:27:08] <james_w> but we are almost back to the starting point
[20:27:41] <james_w> iansmith, I hesitate to say that they are working, but you can see them at http://people.canonical.com/~jamesw/workitems/natty/
[20:27:52] <iansmith> james_w: Does this mean we have to do hacks each cycle to get the blueprint tracking working - or is the solution generic across cycles?
[20:27:53] <james_w> e.g. http://people.canonical.com/~jamesw/workitems/natty/linaro.html
[20:28:08] <james_w> asac, that is now done, except that it won't play nice if we share a db with platform currently
[20:28:28] <asac> cool ... i see milestones now
[20:28:36] <james_w> iansmith, it is generic across cycles where we don't completely change the way that we organise blueprints
[20:29:18] <james_w> iansmith, plus there is a known issue that currently we don't consistently use series in linaro projects, so there will be no way to distinguish a 11.05 blueprint from an 11.11 one when we start creating those
[20:29:19] <iansmith> james_w: That's good - if there are changes to the way blueprints are organised, we (Linaro) have to accept development to sort that out!
[20:29:43] <james_w> iansmith, yes, but maybe we could get the two a little closer together next time
[20:30:06] <james_w> the next thing I am working on is tracking TRs, as currently we have no visibility on their progress
[20:30:12] <asac> james_w: but we will get series support at somet point, right?
[20:30:13] <iansmith> james_w: Is that work part of the current project - or will we want a new project to cover that?
[20:30:49] <james_w> I have half the solution to this, but I'm unsure how to present the grouped data well
[20:31:01] <james_w> series> I have no current plans to work on it
[20:31:17] <james_w> it first requires us deciding how we want to separate the work anyway
[20:31:23] <iansmith> asac: Is the series work important?
[20:31:52] <james_w> iansmith, it's likely very important, but not for at least 3 months
[20:31:54] <asac> we can live without it for now ... moving forward we should get series support though
[20:32:01] <asac> right
[20:32:30] <iansmith> asac, james_w: So it could be another project under the Stakeholder process to be added into our list?
[20:32:34] <james_w> iansmith, and I won't be very popular if I tell people that I'm working on something we don't currently need when everyone is saying that this is ultra-high-number-0 priority for linaro
[20:32:51] <asac> if people have blueprints from last cycle still ongoing its ok to assume they are 11.05 ones until we get closer to end of cycle
[20:33:08] <james_w> iansmith, it should certainly be discussed. It could be an hours work, depending on how we decide to handle it
[20:33:30] <iansmith> james_w: OK, thanks.
[20:33:52] <iansmith> Have I missed anyone out?
[20:34:09] <james_w> asac, yeah. I think at this point though they shouldn't be claiming they are blueprints aiming for a release that has happened and just treat them as 11.05 blueprints
[20:34:45] <iansmith> [TOPIC] Any Other Business
[20:35:00] <iansmith> Any thing anyone wants to raise?
[20:35:17] <james_w> yes
[20:35:32] <iansmith> james_w: OK...
[20:35:32] <james_w> now that we have a work item tracker, we need to pay more attention to workitems
[20:35:54] <james_w> please ensure that any blueprint that is assigned to you has accurate workitems
[20:36:04] <asac> ++
[20:36:34] <iansmith> I agree - work items need not only to be accurate, but also updated as to the status - so the tracking makes sense.
[20:36:39] <james_w> if there is a blueprint assigned to you and you don't think you'll be able to get to any of the workitems as you have too much on your plate, please let me know and we will see what to do
[20:37:32] <asac> my topic: who is our hudson expert that I could grab for some discussion in the next few days? its like two 1 hour calls until end of week i think
[20:37:50] <iansmith> If you can all keep me copied if you are talking about postponing work items - or any major change - so that it doesn't take me by surprise!
[20:37:55] <james_w> asac, plars or I probably
[20:38:12] <asac> james_w: hmm ... maybe anyone who is usually less busy ;)?
[20:38:53] <james_w> salgado, would you be able to help asac with hudson?
[20:39:14] <salgado> I don't have much experience; have never deployed a single instance or played much with it
[20:39:20] <asac> ok
[20:39:22] <salgado> I can try, though
[20:39:35] <james_w> asac, what help are you looking for?
[20:40:31] <asac> james_w: doing some talks to better understand what parts of build service hudson can do and cannot do ...
[20:40:46] <james_w> asac, OBS?
[20:41:04] <james_w> asac, let's take this out of the meeting
[20:41:31] <iansmith> Any other "Any Other Business"?
[20:42:01] <asac> james_w: this is concretely about android and chrome ... but also i want to make up my mind to understand better to what degree hudson could be a strategic asset for linaro platform infrastructure in future
[20:43:28] <iansmith> Is that everything? If so, "speak" to you all again next week!
[20:43:39] <iansmith> Many thanks!
[20:43:53] <salgado> thanks iansmith
[20:44:04] <mwhudson> iansmith: what's the meeting schedule for after xmas?
[20:44:17] <iansmith> salgado: Have a good holiday!
[20:44:18] <mwhudson> i guess there's a meeting on the 5th?
[20:44:37] <mwhudson> ah yes, but not one for me
[20:44:56] <iansmith> mwhudson: There should be a meeting on the 4th?
[20:45:07] <mwhudson> 5th for me :)
[20:45:21] <mwhudson> but yeah, it's at 3am my time according to the calendar
[20:46:06] <iansmith> mwhudson: Yup - so I wouldn't have expected you then - and the following week, we can chat face to face :-)
[20:46:33] <mwhudson> indeed!
[20:47:21] <iansmith> OK! Speak to some of you next week! And those who won't, have good holidays!
[20:47:27] <iansmith> #endmeeting
Meeting ended.

internal/archive/Platform/Infrastructure/Meetings/2010-12-14 (last modified 2013-08-23 01:58:41)