Include: Nothing found for "^## meetings$"!
Include: Nothing found for "^## end-meetings$"!
This page is being re-factored. Most of the old content has been moved to LAVA
|/11.11SchedulerBlueprints /LAVA /LDS-O /PrebuiltImageReliability /Specs /Status /Strategy /WeeklyReport|
- Actions from last meeting - Tech Lead chasing
- Review activity reports - All team members
- Engineer progress reports - All team members
- Work Item progress and team issues discussion - Tech Lead and Project Manager questions to the team
- Any Other Business (AOB)
Past Action items
- Last weeks action items to discuss.
- Action items that came out of this weeks meeting.
Dave Pigott <davepigott>
- Talking to leased line suppliers and gathering quotes
- Looking at NAS solutions
- Looking at USB hub solutions
- Dealing with power supply and cable issues
- Picked up Bug 771182 - Identified fix - awaiting re-org to check in
- Face to face with leased line suppliers
- Deploy more boards into validation lab
- Look at data views and reporting
- Half day Friday
Spring Zhang (springz)
- Merged mp serial log attaching.
- Summary user stories and work items for dispatcher from UDS discussion
- Report a bug and fixed in abrek: [Bug 776268] [NEW] stream report error and no bundle file generate
Tracing an error "UnicodeDecodeError: 'ascii' codec can't decode byte 0xc2 in position 941: ordinal not in range(128)", found in self.context.client.get_seriallog().
- Submit a bug on MX53loco: [Bug 787722] [NEW] [imx53loco] system stop after DA9052 detection when boot up sometimes 'reboot'
- A small mp about adding pkg dependency in abrek: merged
- New blueprint and work items.
Issues & Absence
Le Chi Thu <ChiThu>
- Started to get the snowball hardware up. Still no received a working hardware pack from landing team.
- Updated the Improvement of Abrek blueprint.
- Started to merge outstanding approved merge proposal. Problem with the launchpad. The issue is fired to launchpad team.
- Fix the reported bugs in Abrek
Zygmunt Krynicki <zyga>
- Wrapped up previous cycle work items
- Added some scenarios to dashboard improvements blueprint
- Created lava-server with michael hudson (lp:lava-server)
- Made lava-server capable of hosting "extensions" such as scheduler or the dashboard
- Made the dashboard work with lava-server (in lp:lava-dashboard)
- Work on token auth and lava-dashboard
It's too hot in Spain already
Paul Larson <plars>
- Did some testing on the android lava branch
- Made some fixes and merged basic support for android, more work to be done here still
- Did some minor adjustments and merged Springs serial console log attachments branch to LAVA
- Merged caching support, and several other fixes and catch-ups with what was needed to run in the validation farm
- Redeployed lava to the validation farm using fully what is in trunk, created a separate branch for scripts to generate test jobs
- Modified the Jenkins jobs so that they test all 4 LEBs on all of our boards, ran some tests and they seem to be going pretty well
- Lots of work on Blueprints
- Figure out what is still needed to get Android fully running in the lab
- More blueprint and public plan review work
- I will be out, Monday, May 30 for Memorial Day
Mirsad Vojnikovic <mirsad>
- Working on adding some basic test cases to scheduler daemon implementation
- Discussion about my scheduler blueprints proposal
- Work on 11.11 scheduler blueprints proposal, finalize it and prepare work items for implementation
Michael Hudson-Doyle <mwhudson>
- Collaborated with Zygmunt on splitting lava-server out of launch-control
- Some work on authenticated API stuff (both server and client side)
- Make a skeleton lava-scheduler as a lava-server extension and be able to make an authenticated call to it from lava-tool
- Monday June 6th is a national holiday here.
Your Name <irc nick>
- Short bullet points of work you've done that week which convey progress and highlights which can be used to report on how the team is progressing as a whole.
- Your individual plans for the coming week(s).
- Your individual plans for the coming week(s).
Meeting opened by plars at 01:02
<mwhudson> so who can kick zyga_ out of the channel?
<plars> fabo: is here too I think
mwhudson has set a reminder on his phone for the meeting now
<plars> I can call his wife and tell her to drag him away from the computer
<zyga> plars, she's right next to me
<zyga> plars, we're all in the attic tonight
<plars> [TOPIC] Actions from last meeting
Actions from last meeting
<zyga> guests at home
<plars> hmm, it's not listed on the wiki page
<plars> but the action from last week was for everyone to update the blueprints
<plars> most people have done that
<plars> I've done some editing on it as well
<plars> if you have any updates, please get them in asap
<zyga> plars, did you get a chance to look at dashboard improvements use cases?
<plars> zyga: I did, I think they looked pretty good
<plars> zyga: I will take a second look in a bit and at least approve the direction
<plars> zyga: the idea is, that on these higher level blueprints, or theme blueprints, or whatever you call them, we will have the user stories (sometimes epic stories)
<plars> those user stories can be in a sub-blueprint, which would contain actual work items
<zyga> I knew 'use cases' was not the right term
<mwhudson> i have been a bit slack at updating blueprints
<zyga> just like stories + tasks on the backlog
<plars> and we will add the detail to those as we go, not try to do it all at the beginning
<springz> plars, do we keep the user stories in the blueprint
<plars> so the idea, for now, is to try to sort the priority as best we can
<plars> springz: yes!
<plars> springz: and it's ok to add some bullets under them for clarification
<springz> plars, seems we lost them in previous bp 11.05
<springz> plars, ok
<plars> when a new blueprint is created, please add the link to the whiteboard, and also make it a dependency
<plars> springz: we weren't doing it like this in 11.05
<plars> springz: we are trying to make some changes to the process, and it's all a bit experimental at the moment
<plars> [TOPIC] Engineer progress reports
Engineer progress reports
<plars> yes I skipped activity reports
<plars> I see no need for both
<plars> I'd like to make this short since zyga refuses to sleep
<plars> zyga: I think you have some good news to share
<zyga> well there are some nice things showing up
<zyga> mwhudson and me have split l-c into lava-server and lava-dashboard
<zyga> (the latter is still somewhat broken if your run unit tests, works fine otherwise)
<zyga> and just now I pushed token auth support for linaro-django-xmlrpc
<plars> so the client pieces are going to go into lava-tool?
<springz> zyga, is it easy to rename lp:xxxx branch name when bzr
<zyga> lava-server is also (and this is actually interesting) a host for extensions, so you can hack your custom django appHH lava-server extensions like the scheduler and more easily
<mwhudson> plars: yes
<zyga> springz, lp:foo actually refers to more than just a branch, mwhudson knows if it's easy (probably not unless you have rights to do it)
<zyga> plars, yes, I'll hack them early next cycle
<mwhudson> springz: i can rename launchpad projects
<mwhudson> springz: what's currently called lava will be renamed to lava-dispatcher shortly
<plars> mwhudson: are you basically just excluding everything but the /lava/dispatcher tree for that piece?
<springz> mwhudson, in lp:lava now there is scheduler code, will they also put into lava-scheduler?
<plars> mwhudson: or is there more to it than that?
<mwhudson> plars: no, that's basically it i think
<mwhudson> springz: yeah, i'll copy that across
<mwhudson> springz: there's not a lot there really
<plars> mwhudson: if that's the case, then having a dev environment and deployment should still be very simple, good
zyga will hack on scheduler next cycle
<mwhudson> plars: i'm not sure it's simple now tbh, but no harder than today
<zyga> I'll probably implement the core logic unless someone wants to do it more eagerly than me
<mwhudson> zyga: well, i'm trying to get to it but keep getting rabbitholed
<plars> mwhudson: right, it's a pain to do the first time, more documentation should be written at a minimum
<zyga> plars, the actual difference wrt server side apps is that you run lava-server in once place and hack in another
<plars> but that goes for everything
<zyga> plars, and that things like setup-dev-env are currently impossible to do in one shot, I'll make that possible next cycle
<mwhudson> zyga: i have no problem with you actually implementing it so long as you do it a sensible way
<plars> zyga: next cycle as in 11.11?
<zyga> plars, 11.06
<mwhudson> (i.e. we should talk before too much hacking happens)
<zyga> mwhudson, absolutely
<zyga> mwhudson, I'm mostly interested in the core logic for both scheduler and driver
<zyga> mwhudson, the other 80% of the code will be much harder to write
<plars> zyga: would be nice if you could write up an email about your plans with this, or better yet, a blueprint which corresponds to a user story
<zyga> plars, I will design this and document the design before starting
<plars> awesome, thanks
<plars> mwhudson: any highlights other than what you just mentioned?
<plars> mwhudson: also, if you could add your status to https://wiki.linaro.org/Platform/Validation/Meetings/2011-05-26 it will keep fabo from nagging you
<plars> perhaps we lost him
<plars> springz: ?
<mwhudson> ah soory\
<mwhudson> plars: no no real highlights, sort of slow progress on multiple fronts
<springz> yes, I fixed the no bundle generated bug
<mwhudson> mostly to do with being able to start on the scheduler from the command line end
<plars> I saw, did I merge that yet?
<springz> and look into the UnicodeEncodeError issue, actually, I'm tracing it these days
<plars> springz: I think there was a comment or two I had on it, but I just glanced quickly. Will try to review that asap
<springz> plars, yes, I proposed a mp
<plars> springz: I think I checked in a workaround for that
<mwhudson> plars: cool, i'll subscribe to updates to such pages, that should help me remember to do it
<zyga> plars, springz: the unicode decode bug is actually more complex, you cannot treat serial data as unicode
<zyga> it will never work properly
<zyga> treat it as binary data and allow others to make heads and tails of it
<springz> plars, I met it in one of my old attaching-seriallog branch, and in that situation, it is caused by pexpect.TIMEOUT
<springz> and pexpect quit I think that time
<plars> it's because StringIO can't handle unicode and 8-bit strings combined
<plars> I ran into it today with the android stuff I was working on
<zyga> mixing unicode and binary data is a mess, don't do it
<plars> logcat seems to spew unprintable characters at the beginning
<plars> zyga: right
<plars> I cheated a bit, but using cStringIO makes it a non-issue
<springz> plars, yes, cStringIO can help?
<plars> and hopefully should be slightly faster
<zyga> plars, I saw that, it might still break in the future
<plars> zyga: if they add unicode support?
<zyga> plars, I'd need to check but it seems that one coerces to unicode while the other does not
<zyga> plars, if we don't need to have unicode let's use binary for now
<springz> plars, I'll try it, but I'm confused that I didn't meet it when I do test in attaching-seriallog mp, I met it just after I put seriallog to the big bundle
<zyga> I'd love to be py3k friendly but let's not obfuscate the issue with cstringio
<plars> so for my part, I've been going through blueprints, and preparing for the public plan review, handled the serial attachment support merge with some minor fixes (thanks springz), did some catch-up fixes to lava-trunk and installed that on the validation server
<plars> the scripts for generating the daily runs are in a junk branch of mine for now (they shouldn't be needed once we have a scheduler)
<plars> so daily runs are in jenkins again
<plars> and now are running on ALL LEB image types
<plars> not just alip
<plars> so we should probably rename that stream
<plars> and fix the reports so they don't just look at that stream
<zyga> plars, some reports would need updates as well, they implicitly look at that stream
<plars> zyga: possible to rename in the admin console? or do I need to export/import?
<plars> zyga: right
<plars> that's what I meant
<zyga> plars, it's trivial to rename via admin panel
<plars> also, been looking at the android stuff
<plars> the first bit is checked in, but for some reason 0xbench doesn't seem to generate results for me
<plars> need to sort that out
<plars> and it will need modifications to the master images
<plars> we need an sdcard partition (vfat) for android
<plars> dave will handle that obviously
<zyga> are master images still manual or can I run a script and get master image for android?
<plars> zyga: it's manual, but once you have one you can just back it up
<plars> zyga: depending on the size of your sd card, you may want to adjust the actual size of the partitions
<zyga> yeah I get it, it's hard to automate
<plars> zyga: but it's basically run lmc, shrink the rootfs, create a couple of new partitions, and install a few packages
<plars> yeah, and we can do it once per board type, and just back up the image
<plars> rather than automate it, I would almost rather just make a tgz of the dd img
<plars> but it would still be specific to the size of the sd card like that
<plars> so... if you have an easy way to automate it, it would be nice, but I don't consider it a high priority item
<plars> it's not an especially hard process, just something that needs to be well documented
<springz> plars, yes, it can't apply to a card which size is smaller
<plars> springz: exactly
<plars> ok, anything else for this section?
<plars> [TOPIC] AOB
<plars> So we have 4 people here for this
<zyga> I'd like to release some lava stuff on monday
<plars> zyga: is it ready?
<zyga> I'll spend tomorrow on fixing unit tests
<zyga> plars, there is not much new stuff
<zyga> just changes
<plars> zyga: but this includes the split up server and dashboard
<zyga> 0.5 will be the lava-rebranding release
<plars> zyga: that was my next question, this is going to include the name change right?
<plars> awesome work
<plars> the AOB I was going to bring up, is this meeting
<zyga> plars, after that I'd like to have a brief brainstorm on the 0.6 focus
<plars> we have 4 here, me, mwhudson, zyga (who shouldn't be here), and springz (who this might normally be too early for, and normally makes the other meeting time)
<zyga> plars, I'd like to see scheduler integrated, xmlrpc unified and some initial documentation work but it's all open for discussion
<plars> zyga: we will target some user stories to the 11.06 release
<plars> milestone rather
<plars> springz: which meeting time is more convenient for you?
<springz> plars, if zyga is not here, it can be some time late
<plars> because if this is normally just going to be me and mwhudson at this time, then I would propose we go back to the regular meeting time and have mwhudson catch up on notes and sync with me
<springz> plars, but it's also convenient for me, I'll try to attend the meeting
<plars> but if springz, or others would normally be here, then it makes sense to keep it
<plars> ok, we'll try it a few more weeks and see how it goes
<plars> just wanted to put the suggestion out there at least
<springz> plars, I can attend Thur. night meeting, which we keep long
<plars> so next week's meeting will be at 13:00 UTC instead
<springz> plars, yep
<plars> but we don't need to kill it now, let's see how it goes for a few more weeks and decide then
<plars> ok, anything else?
<mwhudson> nothing from me
<plars> ok, thanks everyone
Meeting closed at 01:40
internal/archive/Platform/LAVA/Meetings/2011-05-26 (last modified 2013-08-23 13:18:19)