hi,all
i got some problems when i download linaro-android 4.2.2 with the following
command:
linaro_android_build_cmds.sh -m pinned-manifest.xml
it stoped at fetching projects:99%(348/351),so i interrupt it by pressing
ctrl +c and run android/repo sync. then i got some errors as follows:
Fetching projects: 2% (8/351) error: Unable to find
3f906009dbb20057c6f6f45069c32d078bd3d27f under
http://android.git.linaro.org/git-ro/platform/build
Cannot obtain needed object 3f906009dbb20057c6f6f45069c32d078bd3d27f
while processing commit eab7126650f75dd9c240574912791b2bafc80bae.
error: Fetch failed.
Fetching projects: 19% (67/351) error: Unable to find
3f906009dbb20057c6f6f45069c32d078bd3d27f under
http://android.git.linaro.org/git-ro/platform/build
Cannot obtain needed object 3f906009dbb20057c6f6f45069c32d078bd3d27f
while processing commit eab7126650f75dd9c240574912791b2bafc80bae.
error: Fetch failed.
Fetching projects: 33% (116/351) error: Cannot fetch platform/build
Fetching projects: 35% (123/351)
error: Exited sync due to fetch errors
does anyone know how can i fix it?
Hello,
I would start with saying that we with Milo and Ben today fixed admin
access issue with Gerrit,
https://bugs.launchpad.net/linaro-android-infrastructure/+bug/1185061 .
It's unclear why it happened, and there're existing 3rd-party reports
of it, so I suggest all Gerrit admin to look at bug to be aware of
symptoms and report them ASAP if they reoccur.
Otherwise, upgrading to Gerrit 2.5 finally allowed us to automatically
mirror new AOSP projects as they appear, without compromising security.
I also improved detection method of new projects - previously we used
manifest from master branch as the source, but as not all projects
listed there, we missed some. I switched to using list of projects as
provided by gitweb at https://android.googlesource.com/ , and that
picked up bunch of new projects. So, from now on, we should have truly
complete AOSP mirror all the time. To remind we sync existing projects
twice a day and pick up new once a day. Follow up if you think that
needs adjustment.
Gerrit/mirror also proved to be quite stable during previous
exploitation, and now should be fairly feature-complete, and as
ITS folks are proactive with supporting services after migration, so I
guess this signifies point where LAVA team can hand over Gerrit/mirror
to them for ongoing maintenance. Philip, Ben, let me know is that's ok
with you. That means that any users of Gerrit and Android upstream
mirror are welcome to report any issues directly to its(a)linaro.org (cc:
infrastructure(a)linaro.org , so we know how well it goes). The rest on
Android infra issue should still go to
https://bugs.launchpad.net/linaro-android-infrastructure/ (well, if you
want LAVA team to react to it first).
Philip, Ben, all notes Gerrit/mirror are reachable from
https://wiki.linaro.org/Platform/Android/Gerrit . That's not intended
to duplicate official docs, but should provide fairly complete
reference of typical usecases and issues we faced. Caveat: it may be
not always up to date. Feel free to "officially" request improvements
to docs via https://bugs.launchpad.net/linaro-android-infrastructure/
(besides just consulting via irc/email of course).
I'm excited about these changes - it's first time in 2-month marathon I
really feel that migration brings stabilization and improvements to our
process and infrastructure. I'm sure that over coming months more
improvements will be made and that feeling will be shared by all teams
in Linaro.
Thanks,
Paul mailto:pmiscml@gmail.com
--
Best Regards,
Paul
Linaro.org | Open source software for ARM SoCs
Follow Linaro: http://www.facebook.com/pages/Linarohttp://twitter.com/#!/linaroorg - http://www.linaro.org/linaro-blog
Hi,
Here is the test result summary of Linaro android jellybean builds on
different boards.
[1] TI Panda 4430;
[2] TI Panda 4460;
[3] ST Ericsson Snowball;
[5] ARM Versatile Express A9;
synopsis:no major difference and new bug in this week tests.
[1] TI Panda 4430 + Linaro Android jellybean (Column : AL)
https://docs.google.com/a/linaro.org/spreadsheet/ccc?key=0AkxwyUNxNaAadGVWd…
it keeps exactly same result as compare to last week tests. No major
changes.
tests result spreadsheet for details.
[2] TI Panda 4460 + Linaro Android jellybean (Column: AL)
https://docs.google.com/a/linaro.org/spreadsheet/ccc?key=0AkxwyUNxNaAadGVWd…
boot ok, browser ok, 1080p hwvideo playback fails, audio ok, others tests
keeps exactly same result as past few weeks test.
[3] ST Ericsson Snowball + Linaro Android jellybean (Column: AK)
https://docs.google.com/a/linaro.org/spreadsheet/ccc?key=0AkxwyUNxNaAadEF1N…
playback of video failed, no audio sound and keeps exactly same tests
result as last few weeks tests
[5] ARM Versatile Express A9; + Linaro android jellybean (Column: AK)
https://docs.google.com/a/linaro.org/spreadsheet/ccc?key=0AkxwyUNxNaAadExQd…
kernel_version is 3.9.1-00199-g6428ef6, toolchain_version: 4.7-2013.04,
suspend-resume test failed, adb-usb not work known issue.device-tree, and
powertop tests pass.
Thank & Regards
Soumya Basak
I was using the command below:
repo init -u http://android.git.linaro.org/git-ro/platform/manifest.git -b
linaro-android-13.04-release -m galaxynexus.xml --repo-url=
http://android.git.linaro.org/git-ro/tools/repo
It can show some branches' messges, but it always halted like below:
From
/home/alexunder/workspace/linaro/galaxy-android/.repo/manifests.git/clone.bundle
* [new branch] master -> origin/master
How do I solve it? Thanks!
Hi,
I’m trying to build Android on an Origen_quad board, but the system keeps
suspending.
I follow the building instruction and use the image here:
https://releases.linaro.org/13.04/android/origenquad/, and my minicom log
is here:
https://docs.google.com/document/d/1YoAvkhAtwRnjXWklQuneBYKQbffjZ3dHJttQx0C…
Even if I build with the bootarg “no_console_suspend”, the CPU will still
suspend anyway. The log is here:
https://docs.google.com/document/d/1ycgOyK7Uv0I07T0FP4g5nBBOuyvIeyZ11FurEZt…
However, I could use interactive command line through serial port. I could
stop the process (entering “stop”) in the middle of booting, which could
prevent the system from suspending (of course the booting is halted too).
But I cannot get video output (from HDMI) in any circumstances.
Any help will be greatly appreciated.
Thanks!!
##- Please type your reply above this line -##
[IT Services] Update: Re: Daily builds are failing during repo init
You are registered as a CC on this support request (linaroithelp.zendesk.com/requests/109).
Reply to this email to add a comment to the request.
----------------------------------------------
Paul Sokolovsky, May 20 09:35 (ADT)
Fruther tries: I commented "RewriteMap prg:" completely, just in case if it applies to each request by error. Didn't help.
I then tried to just wget -r http://android.git.linaro.org/git-http/platform/manifest.git to see how long that will take, and what I found is that there's http://android.git.linaro.org/git-http/platform/manifest.git/objects/pack/p… which is 660MB. Of course, if that pack is download with normal git clone, it got to be slow. And if it isn't, then why such a big pack lies in repo dir at all?
All in all, I didn't have much progress investigating this, so I revert apache config to what it was, suggest Android people to switch to git:// for now, and this bug is stays in queue, waiting for peorper investigation and solution.
----------------------------------------------
Paul Sokolovsky, May 20 09:06 (ADT)
Ok, I did to things:
1) temporarily disabled RewriteLock to see if it will help performance - it did not, took 6 min before it failed.
2) Urls with "/git-http/" are not subject to rewriting, and it still both slow and gives errors:
$ time git clone http://android.git.linaro.org/git-http/platform/manifest.git
Cloning into 'manifest'...
error: Unable to get pack index http://android.git.linaro.org/git-http/platform/manifest.git/objects/pack/p…
error: Unable to get pack index http://android.git.linaro.org/git-http/platform/manifest.git/objects/pack/p…
^C
real 7m18.513s
----------------------------------------------
Paul Sokolovsky, May 20 08:37 (ADT)
Ben:
1) What could be explanation of need for lock - that script doesn't do any context-sensitive processing.
2) We should try to get rid of it at all
----------------------------------------------
Paul Sokolovsky, May 20 08:30 (ADT)
Vishal: this ticket is for issue reported by Jon, and other http access hiccups as confirmed by me.
"branch 'stable' has not been signed" is separate and tracked at https://bugs.launchpad.net/linaro-android-infrastructure/+bug/1181907
----------------------------------------------
Ben Copeland, May 20 08:22 (ADT)
Hi Paul,
I'm not too sure what I can do to speed up the http process. The bottleneck is down to the python script, We have had to put add "RewriteLock" to Apache otherwise users are getting these clone errors. The downside of this is it means it is preventing parallel requests.
When I start to clone http://android.git.linaro.org/git-ro/platform/manifest.git, watching apache log there are many 404 links "GET /git-ro/platform/manifest.git/objects/75/0313d9ddb35959103d75ff8c450f4d7b6596d9 HTTP/1.1" 404 506 "-" "git/1.8.1.2" at the start. These do stop, but its going process each http request per file, its going to be a lengthy wait along with this process lock.
Ben
----------------------------------------------
Vishal Bhoj, May 20 07:59 (ADT)
I still face the same issue which happens on android-build with this command:
repo init -u http://android.git.linaro.org/git-ro/platform/manifest.git -b linaro_android_4.2.2 -m default.xml --repo-url=http://android.git.linaro.org/git-ro/tools/repo
Get http://android.git.linaro.org/git-ro/tools/repo/clone.bundle
fatal: branch 'stable' has not been signed
----------------------------------------------
Paul Sokolovsky, May 20 07:54 (ADT)
Reprod info for the issue I see:
$ time git clone git://android.git.linaro.org/platform/manifest.git
[]
real 0m7.706s
$ time git clone http://android.git.linaro.org/git-ro/platform/manifest.git
[]
real 14m17.214s
----------------------------------------------
Jon Medhurst (Tixy), May 20 07:53 (ADT)
I'm getting no problems now, thanks.
I had already synced up my local system using the git protocol, so I
restored my Android directory from my Friday backup and did a repo sync.
Dozens of git directories were updated by http:// URLs without any
issues.
----------------------------------------------
Ben Copeland, May 20 07:34 (ADT)
Hello Jon,
We had some issues with the git server regarding the git-ro directory. I have made a similar change to the apache config which hopefully is linked to your problem.
Can you try again and let us know if you are still getting problems?
Thanks
Ben
----------------------------------------------
Paul Sokolovsky, May 20 07:34 (ADT)
All in all, I can confirm that there're issues with http:// access on android.git.linaro.org after migration - plain repo init takes very long, and I see errors like:
error: Unable to get pack index http://android.git.linaro.org/git-ro/platform/manifest.git/objects/pack/pac…
error: Unable to get pack index http://android.git.linaro.org/git-ro/platform/manifest.git/objects/pack/pac…
error: Unable to get pack index http://android.git.linaro.org/git-ro/platform/manifest.git/objects/pack/pac…
(Those are non-deterministic, re-trying them manually works.)
----------------------------------------------
Jon Medhurst (Tixy), May 20 07:10 (ADT)
On Mon, 2013-05-20 at 12:43 +0300, Paul Sokolovsky wrote:
> Hello Jon,
>
> On Mon, 20 May 2013 09:35:00 +0100
> "Jon Medhurst (Tixy)" <tixy(a)linaro.org> wrote:
>
> > On Mon, 2013-05-20 at 09:15 +0530, Vishal Bhoj wrote:
> > > Android builds are failing in repo init phase. I have opened a bug
> > > for the same:
> > > https://bugs.launchpad.net/bugs/1181907
> > >
> > > repo init and sync worked for me locally. Can someone please have
> > > a look.
> >
> > I get several errors like this, with various repos...
> >
> > error: Unable to find 51bd0a1525c7a722ad64a8954d90c6df0b8546a7
> > under http://android.git.linaro.org/git-ro/platform/ndk
> > Cannot obtain needed blob
> > 51bd0a1525c7a722ad64a8954d90c6df0b8546a7
> > while processing commit
> > 7b3bda5d569a69c4f9b87225c5d21da4a41bf12e.
> > error: Fetch failed.
>
> This is different issue from one report by Vishal.
Oops, yes. I didn't read the bug, sorry. I jumped to conclusions after
seeing Vishal and Bero's emails appeared at about the same time.
>
> >
> > As I said in reply to bero's mail, I always seem to get these since
> > the move to http URLs
>
> Move to HTTP happened last couple of days during LCA13 in Hong Kong, do
> you want to say that you experienced the error as quoted above since
> then? Was it reported to some bug/ticket? (Do you have a reference?)
I didn't raise a bug as I had a workaround and I assumed if it was a
widespread problem then it would get sorted out eventually.
> The situation known to me is:
>
> 1. After flipping the switch in HK, git:// became unstable (because it
> was throttled to stop wasting resources as it always did). It took some
> time to migrate everyone to http://.
If my memory serves me right, it wasn't long after this I noticed my
problems, which is why it occurred to me to change the URLs in the
manifests back to the old git:// (They'll get restored to http when the
manifests are changed, which is why I notice it as a continuing
problem.)
I perhaps only do a repo sync a couple of times a month as I don't
really work much on Android, which is perhaps why a) I see this problem?
b) I've not been motivated to get to the bottom of it.
--
Tixy
----------------------------------------------
Paul Sokolovsky, May 20 06:44 (ADT)
Hello Jon,
On Mon, 20 May 2013 09:35:00 +0100
"Jon Medhurst (Tixy)" <tixy(a)linaro.org> wrote:
> On Mon, 2013-05-20 at 09:15 +0530, Vishal Bhoj wrote:
> > Android builds are failing in repo init phase. I have opened a bug
> > for the same:
> > https://bugs.launchpad.net/bugs/1181907
> >
> > repo init and sync worked for me locally. Can someone please have
> > a look.
>
> I get several errors like this, with various repos...
>
> error: Unable to find 51bd0a1525c7a722ad64a8954d90c6df0b8546a7
> under http://android.git.linaro.org/git-ro/platform/ndk
> Cannot obtain needed blob
> 51bd0a1525c7a722ad64a8954d90c6df0b8546a7
> while processing commit
> 7b3bda5d569a69c4f9b87225c5d21da4a41bf12e.
> error: Fetch failed.
This is different issue from one report by Vishal.
>
> As I said in reply to bero's mail, I always seem to get these since
> the move to http URLs
Move to HTTP happened last couple of days during LCA13 in Hong Kong, do
you want to say that you experienced the error as quoted above since
then? Was it reported to some bug/ticket? (Do you have a reference?)
The situation known to me is:
1. After flipping the switch in HK, git:// became unstable (because it
was throttled to stop wasting resources as it always did). It took some
time to migrate everyone to http://.
2. http:// was stable since then, until migration of git.linaro.org
happened weekened May 11.
3. Immediately after migration, there were issues with
http://git.linaro.org which manifested pretty much like the error
above. It was resolved Mon/Tue last week.
4. Friday GMT evening, android.git.linaro.org was migrated to new
setup. Initial builds seem to have gone ok. Detailed status is not
known (your telling that error above was received after May 17,
17:00GMT would help).
5. Additionally, today issue reported by Vishal popped up.
All in all, git issues now should be cc: to its(a)linaro.org (added), ITS
guys are now the owners of new infrastructure, and can resolve issues
pretty well (p.3 was resolved by them).
> and I work around it by changing the manifest
> to have the old git URLs.
This may work as a personal workaround, but won't scale to official
manifests - we need to get http:// working right.
--
Best Regards,
Paul
Linaro.org | Open source software for ARM SoCs
Follow Linaro: http://www.facebook.com/pages/Linarohttp://twitter.com/#!/linaroorg - http://www.linaro.org/linaro-blog
--------------------------------
This email is a service from IT Services