20110623

June 9th, 2011, 15:00 UTC in #ubuntu-meeting.

Agenda

Action Items from last meeting

  • NCommander to get Documentation spec approved

Standing Items

Meeting Outcome

[15:58] <NCommander> #startmeeting
[15:58] <MootBot> Meeting started at 09:58. The chair is NCommander.
[15:58] <MootBot> Commands Available: [TOPIC], [IDEA], [ACTION], [AGREED], [LINK], [VOTE]
[15:59] <davidm> G'day NCommander
[15:59]  * NCommander sleeply waves
[15:59] <NCommander> [topic] Action Item Review
[15:59] <MootBot> New Topic:  Action Item Review
[15:59] <NCommander> [link] https://wiki.ubuntu.com/MobileTeam/Meeting/2011/20110526https://wiki.ubuntu.com/MobileTeam/Meeting/2011/20110526
[15:59] <MootBot> LINK received:  https://wiki.ubuntu.com/MobileTeam/Meeting/2011/20110526https://wiki.ubuntu.com/MobileTeam/Meeting/2011/20110526
[15:59] <NCommander> er, none
[16:00] <NCommander> [topic] Standing Items
[16:00] <MootBot> New Topic:  Standing Items
[16:00] <NCommander> [link] http://people.canonical.com/~platform/workitems/oneiric/ubuntu-armel.html
[16:00] <MootBot> LINK received:  http://people.canonical.com/~platform/workitems/oneiric/ubuntu-armel.html
[16:00] <NCommander> I just realized I posted the wrong link
[16:00] <NCommander> [link] https://wiki.ubuntu.com/MobileTeam/Meeting/2011/20110623
[16:00] <MootBot> LINK received:  https://wiki.ubuntu.com/MobileTeam/Meeting/2011/20110623
[16:01] <NCommander> [topic] NCommander to get Documentation spec approved
[16:01] <MootBot> New Topic:  NCommander to get Documentation spec approved
[16:01] <NCommander> c/o
[16:01] <NCommander> [topic] http://people.canonical.com/~platform/workitems/oneiric/ubuntu-armel.html
[16:01] <MootBot> New Topic:  http://people.canonical.com/~platform/workitems/oneiric/ubuntu-armel.html
[16:01] <NCommander> [topic] http://people.canonical.com/~platform/workitems/oneiric/ubuntu-armel-oneiric-alpha-2.html
[16:01] <MootBot> New Topic:  http://people.canonical.com/~platform/workitems/oneiric/ubuntu-armel-oneiric-alpha-2.html
[16:02] <ogra> we look bad ...
[16:02] <ogra> but with dublin ahead that should not matter
[16:02] <NCommander> ++
[16:03] <NCommander> anything else or can I move on?
[16:03] <ogra> move
[16:03] <NCommander> [topic] ARM Server Status (NCommander, Daviey)
[16:03] <MootBot> New Topic:  ARM Server Status (NCommander, Daviey)
[16:04] <NCommander> Slow week. Did some work creating udebs, and determined my panda was fried.
[16:04] <ogra> the existing PXE code is uploaded though
[16:04] <NCommander> \o/
[16:05] <ogra> havent tested how complete it is yet
[16:05] <NCommander> we've got the sprint, we'll catch up there
[16:05] <NCommander> [topic] Kernel Status (cooloney, rsalveti, ppisati)
[16:05] <MootBot> New Topic:  Kernel Status (cooloney, rsalveti, ppisati)
[16:05] <ogra> yep
[16:05] <ppisati> nothing exciting: usual tons of cve fixes
[16:05] <ogra> any news regarding the usb issue ?
[16:05] <ppisati> and...
[16:05] <ppisati> the usb issue has been fixed & committed :)
[16:06] <ogra> \o/
[16:06] <GrueMaster> woot
[16:06] <ppisati> yep
[16:06] <NCommander> yay
[16:06] <ppisati> btw i'm bit low on stuff, so if you have any outstading issue
[16:06] <ppisati> feel free to throw it to me
[16:06] <ogra> we'll get you stuff next week, dont worry
[16:07] <ppisati> BTW, bring some otg stuff to dublin since it seems i can't make it work
[16:07]  * NCommander throws a trout at ppisati 
[16:07] <ppisati> :)
[16:07] <ogra> sprints are good for collecting bugs
[16:07] <GrueMaster> Ohhh yeaaa.
[16:07] <ppisati> you can move from my pov
[16:07] <ppisati> ah no
[16:07] <ppisati> wait
[16:07] <ppisati> tegra2?
[16:07] <GrueMaster> ppisati: I'll bring you an OTG cable.
[16:08] <ppisati> GrueMaster: i've all kinds of otg cables but it seems it doesn't make any difference for my xm
[16:08] <ppisati> GrueMaster: anyway, bring it, thanks
[16:08] <NCommander> [topic] ARM Porting/FTBFS status (NCommander, janimo)
[16:08] <MootBot> New Topic:  ARM Porting/FTBFS status (NCommander, janimo)
[16:08] <NCommander> telepathy-glib has kicked out asses
[16:08]  * ogra looks at infinity 
[16:09] <ogra> your sync ftbfs
[16:09] <ogra> sadly empathy gets uninstallable through that
[16:09] <ogra> so we dont get images
[16:10] <NCommander> anything else?
[16:10]  * ogra would like to have a recent image before the sprint ... 
[16:10] <davidm> Yes that would be nice
[16:10] <ogra> should we drop ampathy for now ?
[16:10] <ogra> *empathy
[16:10] <davidm> Yes if need be
[16:11] <NCommander> ++
[16:11] <ogra> i'm not sure if there is more in the desktop seed
[16:11] <davidm> I really don't want to block on that
[16:11] <ogra> more depending on telepathy-glib i mean
[16:11] <infinity> Err, oh.  Right.  Meeting.
[16:11] <infinity> NCommander: tp-glib is suffering from "lp-buildd is broken" issues.
[16:12] <infinity> ogra: No need to drop empathy, I'll just bump the test timeout in tp-glib to something stupidly high to work around twistd.
[16:12] <NCommander> ugh
[16:12]  * infinity will throw a test upload at his PPA right now and see if it sticks.
[16:12] <ogra> infinity, ok, if we dont have a binary by tomorrow evening i'll drop it though
[16:12] <infinity> ogra: *nod*
[16:13] <ogra> so we have a chance to get an image over the weekend
[16:13] <infinity> ogra: If you do, file a bug against soyuz, it's not actually a tp-glib issue.
[16:14] <ogra> well, thats only temporary indeed, we can corner StevenK in dublin to fix it ;)
[16:14]  * NCommander packs duct tape and a staple gun
[16:14] <NCommander> k
[16:14] <NCommander> can we ove?
[16:14] <ogra> nothing else from me
[16:14] <ogra> (and partially also already the image topic :) )
[16:15] <NCommander> [topic] ARM Image Status (ogra, NCommander)
[16:15] <MootBot> New Topic:  ARM Image Status (ogra, NCommander)
[16:15]  * GrueMaster hasn't ove'd in years
[16:15] <ogra> GrueMaster, oh ? why not ?
[16:15] <ogra> well ... image status, headles builds fine, i havent tested a single one since a while though
[16:16] <ogra> desktop doesnt ... see above :)
[16:16] <NCommander> is there anything else?
[16:16] <GrueMaster> Tested headless omap4 on Tuesday.  Still works.
[16:16]  * NCommander doesn't have anything to add
[16:16] <ogra> great
[16:16] <GrueMaster> Good progress on netboot.
[16:16] <ogra> jasper will need preseed file support for server preinstalled
[16:16] <ogra> we need to discuss where to put that file in dublin
[16:17] <ogra> since unlike in d-i images it will persist on the FS
[16:17] <NCommander> ogra: /var/cache/installer/preseed?
[16:17] <ogra> something like that
[16:17] <NCommander> oro something like that
[16:18] <ogra> yeah
[16:18] <NCommander> anything else?
[16:18] <ogra> the coding will be trivial, but i somehow cant make up my mind about the file :)
[16:18] <ogra> nothing else here
[16:18] <NCommander> [topic] QA Status (GrueMaster)
[16:18] <MootBot> New Topic:  QA Status (GrueMaster)
[16:18] <GrueMaster> Oh, that's me.
[16:19] <GrueMaster> I have my relay now fully operational and online.  I can remotely control 2 pandas atm (limited to pandas on hand).
[16:19] <ogra> you should patent that and sell it to zoos ;)
[16:19] <GrueMaster> Work progressing on ramping up jenkins automation environment.  Lots of questions for the QA guys next week.
[16:20] <GrueMaster> filesystem benchmark testing is very slow & manual.  Each run takes 84 hours and then I have to reformat & restore the rootfs.
[16:21] <ogra> how do you benchmark exactly ?
[16:21] <GrueMaster> I'm using the phoronix filesystem benchmarks.
[16:21] <ogra> i think actually just producing different bootcharts for different filesystems should already show some markers
[16:21] <GrueMaster> All the fs benchmarking tools in one convenient package.
[16:22] <GrueMaster> Not reliable.
[16:22] <ogra> reliable for boot time measuring
[16:22] <ogra> indeed not for a full filesystem benchmark
[16:22] <GrueMaster> filesystem testing is more than just booting.
[16:23] <ogra> i didnt say it isnt :)
[16:23] <GrueMaster> And my task is to benchmark filesystems.  This seemed the best way.
[16:23] <ogra> but getting a bootchart is a matter of seconds
[16:23] <NCommander> ogra: but bootchart is mostly read and sequencish IO
[16:23] <ogra> sure
[16:23] <NCommander> not nightmarish abuse that phoronix does
[16:23] <GrueMaster> exactly.
[16:23] <ogra> its only one aspect
[16:24] <ogra> but low hanging fruit wrt data gathering
[16:24] <NCommander> invalid datat though.
[16:24] <ogra> no
[16:24] <NCommander> would give us skrewed results
[16:24] <GrueMaster> Our boot performance isn't the issue (actually boot time is very impressive).
[16:24] <infinity> bootcharting (especially with intelligent readahead stuff in play) isn't testing your filesystem at all, just the speed of your disk.
[16:24] <ogra> would give us tendency about reading on different filesystems while the full suite runs
[16:24] <infinity> Since most boot tricks are designed to work AROUND the filesystem. :P
[16:24] <GrueMaster> My panda boots nearly at the same speed as my netbook with SSD.
[16:25] <ogra> from power on to desktop ?
[16:25] <GrueMaster> ogra: You would be looking at a few milliseconds in differences.
[16:25] <GrueMaster> yes.
[16:25] <infinity> And yeah, my Panda boots impressively quickly.
[16:25] <infinity> And then takes 20 minutes to load Firefox while I go for lunch.
[16:25] <ogra> intresting, your netbook must have a slow bios :)
[16:25] <GrueMaster> The u-boot delay is close to the bios init time on my netbook.
[16:25] <ogra> infinity, far from whats possible
[16:25]  * rsalveti waves
[16:26] <ogra> u-boot steals massive amounts of time
[16:26] <rsalveti> should we change u-boot count time for 1, 2 seconds?
[16:26] <GrueMaster> 4 seconds of u-boot countdown.
[16:26] <GrueMaster> currently
[16:26] <GrueMaster> 10 on beagle.
[16:26] <ogra> plus about 10-12 for loading initrd and kernel
[16:27] <rsalveti> GrueMaster: I believe 10 will be fixed in next u-boot upload
[16:27] <GrueMaster> Really?
[16:27] <rsalveti> that should be coming today or tomorrow
[16:27] <rsalveti> nops, it's fast now
[16:27] <rsalveti> 2, 3 secs to load
[16:27] <rsalveti> at least on panda/beagle
[16:27] <rsalveti> because of the multi block read
[16:27] <ogra> oh, then it improved
[16:27] <GrueMaster> ogra: Not sure why you are seeing such slow kernel & initrd load times.  Mine is maybe .5 seconds.
[16:28] <rsalveti> just firefox that's a pita atm
[16:28] <rsalveti> I can make a fresh coffee when I decided to start it
[16:28] <GrueMaster> heh
[16:28] <ogra> GrueMaster, from end of countdown to the "ucompressing kernel" message its .5s ?
[16:29] <GrueMaster> roughly for each, yes.
[16:29] <ogra> i'm surely not seeing such times
[16:29] <infinity> rsalveti: Firefox may as well be the "vicious filesystem test" we're looking for, once it's been used a bit.  Cause loading the profile and parsing history, etc, is just evil on a randomly-accessed/abused SD.
[16:29] <GrueMaster> I'll try to time it with a stopwatch.
[16:29] <ogra> though i havent tried oeniric on my panda yet
[16:30] <rsalveti> infinity: yeah, did you try to profile it already?
[16:30] <GrueMaster> That was true even with natty beta+
[16:30] <ogra> will pull a headless image, thats actually something i want to see
[16:30] <infinity> rsalveti: No, I just plugged in an external hard drive, thus preventing my urge to launch my Panda out the window.
[16:30] <rsalveti> :-)
[16:30] <GrueMaster> headless & netbook should have the same start times prior to mounting rootfs and running init from there.
[16:30]  * ogra just uses chromium :)
[16:31] <NCommander> anything els or can I move on?
[16:31] <GrueMaster> chromium is not on the image.  invalid test procedure.
[16:31] <GrueMaster> move.
[16:31] <ogra> yeah
[16:32] <ogra> well, surely not invalid :)
[16:32] <NCommander> [topic] AOB
[16:32] <MootBot> New Topic:  AOB
[16:32] <GrueMaster> for filesystem abuse?  yes it is if it doesn't do the same thrashing.
[16:32]  * ogra is pretty sure fta likes to get arm bugs from actual users :)
[16:33] <ogra> GrueMaster, no, for having a usable browser :)
[16:33] <infinity> (The real fun is when you start swapping to a swap file on the SD, and realise that you probably would have rather just had the OOM killer pick something at random to slaughter than put up with swapping to the same SD you're trying to randomly read/write to)
[16:33] <ogra> not for FS abuse testing
[16:33] <NCommander> if nothing else will close the meeting out
[16:33] <NCommander> 3
[16:33] <NCommander> 2
[16:33] <davidm> Folks for the ARM Server Project test against USB HD
[16:33] <ogra> infinity, well, tell that to the guy with the 200 page writer document open he just wrote :)
[16:34] <davidm> not SD card time please
[16:34] <infinity> davidm: Yeah, and I'm VERY pleased with the Panda with a USB HDD.
[16:34] <NCommander> thanks davidm
[16:34] <ogra> i rarther wait 20min for my system to return than having OOM kick in in such a case
[16:34] <GrueMaster> davidm: planned on it.
[16:34] <davidm> I do want to figure out the SD card issues but that may be a hardware issue
[16:34] <davidm> and un fixable by us
[16:34] <NCommander> ogra: you can force that by smacking with knobs with sysctl)
[16:35] <davidm> I'm looking forward to next week
[16:35] <ogra> NCommander, i can also have a swapfile and swppiness set to 1
[16:35] <GrueMaster> davidm: The idea behind the filesystem benchmarking & improvement was doing the best with what we have.
[16:35] <NCommander> anyway, can I close the meeting now?
[16:35] <ogra> which gets me the overflow memory on disk
[16:35] <NCommander> guess not
[16:35] <GrueMaster> NCommander: Wait 5.
[16:35] <ogra> instead of stealoing from another app
[16:35] <ogra> close :)
[16:36] <davidm> Close
[16:36]  * ogra doesnt have anything
[16:36] <GrueMaster> done
[16:36] <infinity> davidm: And I concur.  I'm loooking forward to seeing everyone again.  Face-time makes a huge difference sometimes.
[16:36] <NCommander> #endmeeting

Action Items

  • None

ARM/Meeting/2011/20110623 (last edited 2011-07-28 17:58:54 by davidm)