3rd January 2011
- Actions from previous meeting
- Brief Progress
- Any other business
Action Items from last Meeting
Action Items this Meeting
take a look at ruote and see if it would be useful
Minutes IRC (Transcript)
Meeting opened by plars at 14:05
<robtaylor> zyga: lets chat more later.
<zyga> robtaylor, and build something that works (the scheduler is just a small part of that) - hopefully then we can make more educated guesses
<plars> [TOPIC] Actions from previous meeting (https://wiki.linaro.org/Platform/Validation/Meetings/2011-01-27)
Actions from previous meeting (https://wiki.linaro.org/Platform/Validation/Meetings/2011-01-27)
<plars> looks like none
<plars> [TOPIC] Brief Progress
<plars> mirsad: Would you like to go first? I know you have a hard stop in 20 min. or so
<mirsad> yes, thanks!
<mirsad> it is not much, finalized first version of scheduler blueprint
<mirsad> sent it for review
<mirsad> done some minor dispatcher review
<mirsad> and that's it - still no code, paul
<plars> ok, I gave it a quick glance, but I need to spend some time looking at it
<plars> would appreciate if others could take a look as well
<zyga> I _promise_ to look at all blueprints on Monday, I'll be a lava golem since Monday
<plars> we don't have everyone here today, so it may be quicker meeting today
<plars> oooo, cool
<mirsad> zyga, be mean
<zyga> I can be mean
<plars> zyga: would you like to go now?
<zyga> do you really want me to ? :]
<mirsad> zyga, love that
<zyga> plars, sure, I'm still working on l-c, as we discussed earlier I'm trying to wrap things up - I'm currently merging deepti's branches - should be all done today.
<zyga> I will still want to hack the generic benchmark view and ship 0.3 alpha 1
<zyga> I'm going to start hacking lava next week so you can probably expect code merges and my annoying design decisions
<zyga> and I'll use JSON for sure; -)
<plars> It looks like that's the direction it's heading anyway, we just need to formalize it for the spec
<plars> and so long as we don't do something silly like requiring floating point numbers instead of strings...
<zyga> I did not hear that
<zyga> but yeah, I want to make sure it actually works
<plars> anything else to add?
<zyga> not of the top of my head
<plars> Hi iansmith
<plars> we started without you
<plars> my turn
<plars> I added a few things there this week
<iansmith> plars: Sorry - I thought no-one was replying to my questions!
<iansmith> plars: It seems I had been disconnected!
<plars> We now have a job to create images (combine hwpack+rootfs) for headless and boot them on beaglexm and panda
<plars> we should be getting some new boards in there soon too
<plars> so expect others to show up as that happens
<zyga> yay, that's awesome!
<plars> is my branch for stuff that I'm working on for this, and experimental stuff for lava
<plars> some of it will be things that we certainly want to cherry-pick and push into lava
<plars> some of it will be temporary scaffolding to just "make it work" until we have a better solution
<plars> some of it will be total garbage
<plars> don't say I didn't warn you
<plars> but please feel free to take a look, it's what I'm using for the scripts to deploy and boot
<zyga> does everyone have access to those scripts?
<plars> there are a couple of things not checked in yet, such as the master image side scripts, they are pretty basic but should go there anyway
<plars> zyga: sure, it's a public branch
<robtaylor> plars: I took a look, and it looks pretty good to me. not a lot but does the job.
<iansmith> plars: I think we need to be careful about the names and analogies - one day it may bite us
<zyga> iansmith, if another volcano blocks UDS it's not going to be our fault
<iansmith> he he!
<plars> we have equipment on order to expand the lab, which will probably happen next week
<plars> this possibly fits more under AOB, but I'll mention it now anyway
<zyga> plars, who's going to assemble that stuff?
<plars> this will add a rack, our own switch rather than piggy- backing off the office one, shelving, more PDUs, etc
<plars> should allow us to host up to 30 or so boards
<plars> and also give us room to grow past that
<plars> zyga: I'll be assisting remotely with people that are going to be there next week, nobody is planning to travel just to do this right now
<plars> eventually, we'll have someone hired *in* cambridge, who can help with these sorts of things. I'm working on that too
<plars> anyway, the image/boot jobs were my big highlight, but wanted everyone to know about the upcoming expansion as it may mean some outages in the lab
<plars> robtaylor: do you have anything you want to add here?
<plars> I know you were looking at BOSS right?
<robtaylor> I've been reading around linaro validation, digesting specs and sourcecode. At the moment I'm flow diagraming the validation process to capture the use cases.
<robtaylor> plars: yes, boss itself isn't of much use, but the infastructure it's built with - ruote and RabbitMQ - look like they might be useful
<plars> robtaylor: we were already looking at maybe using RabbitMQ
<plars> what would ruote give us exactly?
<plars> (other than a headache trying to remember not to spell it route)
<robtaylor> plars: it's absically a process runner. you can describe a process in simple terms (e.g. as a json doc) and it will asynchonously run the process, possibly over multple machines
<zyga> robtaylor, a real process or a business-logic process?
<robtaylor> zyga: business-logic process
<zyga> I see
<zyga> job dispatcher kind of thing
<robtaylor> zyga: exactly
<robtaylor> has support for multiple concurrent tasks, etc and integrates nicely wirh AMQP
http://ruote.rubyforge.org/ seems to be the link if anyone wants to take a look
<plars> http://ruote.rubyforge.org/ seems to be the link if anyone wants to take a look
<zyga> I'll check it out
<robtaylor> at the moment I'm working with another Codethinker to test it out and see if it helps get us a scheduler and dispatcher quickly
<zyga> unless someone wants to pry that from my dead hands
<robtaylor> plars: yes
<plars> zyga: go for it
<robtaylor> zyga: i'll be happy to report on it on the mailing list later
<zyga> robtaylor, ok
<zyga> robtaylor, I'm a little out of sync with the mailing list - I'm best to reach on IRC nowdays -- still mailing list is a good idea
<plars> [ACTION] zyga and robtaylor to take a look at ruote and see if it would be useful
zyga and robtaylor to take a look at ruote and see if it would be useful
<robtaylor> zyga: sure
<plars> ok, anything else you want to add robtaylor?
<robtaylor> plars: no, that's it from me
<plars> [TOPIC] AOB
<plars> mentioned the lab expansion already
<plars> mailing list
<plars> we don't have a validation team mailing list, and unless I'm convinced otherwise, we probably won't
<plars> we do need to make sure to cc linaro-dev on most discussions though
<zyga> yeah, let's just use linaro-dev
<robtaylor> *nod* signal-to-noise is good there, no need to get another ml
<plars> so for feedback on the scheduler spec, do the same thing I did with moving the job dispatcher one there please
<plars> feel free to cc people on the team specifically, as some people (like me) have mail filters that help flag mail like that
<plars> but make sure the mailing list is included so others can get involved if they wish
<plars> I'm not planning to be out any this week, but I may lose power due to weather
<zyga> plars, ! ?
<zyga> plars, snow?
<plars> we have had fun rolling blackouts here
<plars> zyga: not yet, possibly later in the week, but we haven't been above freezing for most of the week, and we *might* get above freezing tomorrow for a little while
<zyga> plars, you should consider greece
<iansmith> plars: That was never shown in Dallas with JR!
<plars> iansmith: and I'm even further south of Dallas!
<plars> anyone else have anything for AOB?
<iansmith> plars: Nothing from me!
<plars> ok then
Meeting closed at 14:35
internal/archive/Platform/LAVA/Meetings/2011-02-03 (last modified 2013-08-23 13:19:28)