do_fetch error with gcc-cross-initial_4.7

do_fetch error with gcc-cross-initial_4.7

[Florin]

Hi all,
using today's poky master branch, sha 
603f6c631e9454d10517599e229943f2b26d08a5 , I get an error when fetching 
gcc.

I first get this warning:

WARNING: Failed to fetch URL 
git://github.com/mirrors/gcc.git;branch=gcc-4_7-branch;protocol=git, 
attempting MIRRORS if available


Then it stays stuck at fetching for a while and ends up with:

git clone --bare --mirror git://github.com/mirrors/gcc.git 
/home/me/poky/build/downloads/git2/github.com.mirrors.gcc.git failed 
with exit code 128, output:
STDOUT: Cloning into bare repository 
/home/me/poky/build/downloads/git2/github.com.mirrors.gcc.git...

STDERR: fatal: The remote end hung up unexpectedly
fatal: early EOF
fatal: index-pack failed

ERROR: Function failed: Fetcher failure for URL: 
'git://github.com/mirrors/gcc.git;branch=gcc-4_7-branch;protocol=git'. 
Unable to fetch URL from any source.


After the build stops with this error, retrying the bitbake command to 
produce the rootfs fetches gcc successfully and gcc (and the rest of the 
packages) is compiled just fine.
Is there anybody else seeing this behavior?

Thank you,
Florin

Re: do_fetch error with gcc-cross-initial_4.7

[Tomas Frydrych]

Hi,

On 22/08/12 22:35, Florin wrote:
> ERROR: Function failed: Fetcher failure for URL:
> 'git://github.com/mirrors/gcc.git;branch=gcc-4_7-branch;protocol=git'.
> Unable to fetch URL from any source.
> 
> 
> After the build stops with this error, retrying the bitbake command to
> produce the rootfs fetches gcc successfully and gcc (and the rest of the
> packages) is compiled just fine.
> Is there anybody else seeing this behavior?

FWIW, over the last few days I have seen intermittent fetch failures
from various packages that pull source from github; they pretty much
always go away on the next attempt; I assume it's github's fault.

Tomas

Re: do_fetch error with gcc-cross-initial_4.7

[Florin]

On 08/23/2012 10:29 AM, Tomas Frydrych wrote:
> Hi,
>
> On 22/08/12 22:35, Florin wrote:
>> ERROR: Function failed: Fetcher failure for URL:
>> 'git://github.com/mirrors/gcc.git;branch=gcc-4_7-branch;protocol=git'.
>> Unable to fetch URL from any source.
>>
>>
>> After the build stops with this error, retrying the bitbake command to
>> produce the rootfs fetches gcc successfully and gcc (and the rest of the
>> packages) is compiled just fine.
>> Is there anybody else seeing this behavior?
> FWIW, over the last few days I have seen intermittent fetch failures
> from various packages that pull source from github; they pretty much
> always go away on the next attempt; I assume it's github's fault.
Hi Tomas,
thanks for your answer. The problem I am seeing is that it always 
behaves like this: in the stages of creating the rootfs it gets to the 
fetching of gcc cross initial, this stalls a while (tens of minutes) in 
which time bitbake fetches (all kind of native and target packages) and 
compiles native packages and it throws an error at the point where the 
dependency chain would have had gcc cross initial compiled and ready to 
be used by other packages. After that, rerunning the bitbake command to 
create the rootfs fetches successfully gcc and the build finishes. So I 
am always seeing this behavior: first fetch fails, second fetch works. 
Always like this, never had to try a third time to fetch gcc. That is 
why I am not suspecting especially a problem with github but rather the 
way bitbake tries the mirrors for gcc.

Thanks,
Florin
> Tomas
> _______________________________________________
> poky mailing list
> poky@yoctoproject.org
> https://lists.yoctoproject.org/listinfo/poky

Re: do_fetch error with gcc-cross-initial_4.7

[Khem Raj]

[-- Attachment #1: Type: text/plain, Size: 2188 bytes --]

On Thursday, August 23, 2012, Florin <flocirel@gmail.com> wrote:
> On 08/23/2012 10:29 AM, Tomas Frydrych wrote:
>>
>> Hi,
>>
>> On 22/08/12 22:35, Florin wrote:
>>>
>>> ERROR: Function failed: Fetcher failure for URL:
>>> 'git://github.com/mirrors/gcc.git;branch=gcc-4_7-branch;protocol=git'.
>>> Unable to fetch URL from any source.
>>>
>>>
>>> After the build stops with this error, retrying the bitbake command to
>>> produce the rootfs fetches gcc successfully and gcc (and the rest of the
>>> packages) is compiled just fine.
>>> Is there anybody else seeing this behavior?
>>
>> FWIW, over the last few days I have seen intermittent fetch failures
>> from various packages that pull source from github; they pretty much
>> always go away on the next attempt; I assume it's github's fault.
>
> Hi Tomas,
> thanks for your answer. The problem I am seeing is that it always behaves
like this: in the stages of creating the rootfs it gets to the fetching of
gcc cross initial, this stalls a while (tens of minutes) in which time
bitbake fetches (all kind of native and target packages) and compiles
native packages and it throws an error at the point where the dependency
chain would have had gcc cross initial compiled and ready to be used by
other packages. After that, rerunning the bitbake command to create the
rootfs fetches successfully gcc and the build finishes. So I am always
seeing this behavior: first fetch fails, second fetch works. Always like
this, never had to try a third time to fetch gcc. That is why I am not
suspecting especially a problem with github but rather the way bitbake
tries the mirrors for gcc.
>

Git archive for gcc is in terabytes so it will take long time to fetch it.
I have sent a patch to switch back to Svn in src_URI
That should avoid long delays in fetching gcc srcs
> Thanks,
> Florin
>>
>> Tomas
>> _______________________________________________
>> poky mailing list
>> poky@yoctoproject.org
>> https://lists.yoctoproject.org/listinfo/poky
>
> _______________________________________________
> poky mailing list
> poky@yoctoproject.org
> https://lists.yoctoproject.org/listinfo/poky
>

[-- Attachment #2: Type: text/html, Size: 2842 bytes --]

Re: do_fetch error with gcc-cross-initial_4.7

[Florin]

[-- Attachment #1: Type: text/plain, Size: 2728 bytes --]

Trying to build something today and still takes forever and crashes at 
fetch. Looking at the gcc-4.7.inc I see it's still from git.
What would the resolution be for this issue?

Thank you

On 08/23/2012 06:35 PM, Khem Raj wrote:
>
>
> On Thursday, August 23, 2012, Florin <flocirel@gmail.com 
> <mailto:flocirel@gmail.com>> wrote:
> > On 08/23/2012 10:29 AM, Tomas Frydrych wrote:
> >>
> >> Hi,
> >>
> >> On 22/08/12 22:35, Florin wrote:
> >>>
> >>> ERROR: Function failed: Fetcher failure for URL:
> >>> 
> 'git://github.com/mirrors/gcc.git;branch=gcc-4_7-branch;protocol=git 
> <http://github.com/mirrors/gcc.git;branch=gcc-4_7-branch;protocol=git>'.
> >>> Unable to fetch URL from any source.
> >>>
> >>>
> >>> After the build stops with this error, retrying the bitbake command to
> >>> produce the rootfs fetches gcc successfully and gcc (and the rest 
> of the
> >>> packages) is compiled just fine.
> >>> Is there anybody else seeing this behavior?
> >>
> >> FWIW, over the last few days I have seen intermittent fetch failures
> >> from various packages that pull source from github; they pretty much
> >> always go away on the next attempt; I assume it's github's fault.
> >
> > Hi Tomas,
> > thanks for your answer. The problem I am seeing is that it always 
> behaves like this: in the stages of creating the rootfs it gets to the 
> fetching of gcc cross initial, this stalls a while (tens of minutes) 
> in which time bitbake fetches (all kind of native and target packages) 
> and compiles native packages and it throws an error at the point where 
> the dependency chain would have had gcc cross initial compiled and 
> ready to be used by other packages. After that, rerunning the bitbake 
> command to create the rootfs fetches successfully gcc and the build 
> finishes. So I am always seeing this behavior: first fetch fails, 
> second fetch works. Always like this, never had to try a third time to 
> fetch gcc. That is why I am not suspecting especially a problem with 
> github but rather the way bitbake tries the mirrors for gcc.
> >
>
> Git archive for gcc is in terabytes so it will take long time to fetch 
> it. I have sent a patch to switch back to Svn in src_URI
> That should avoid long delays in fetching gcc srcs
> > Thanks,
> > Florin
> >>
> >> Tomas
> >> _______________________________________________
> >> poky mailing list
> >> poky@yoctoproject.org <mailto:poky@yoctoproject.org>
> >> https://lists.yoctoproject.org/listinfo/poky
> >
> > _______________________________________________
> > poky mailing list
> > poky@yoctoproject.org <mailto:poky@yoctoproject.org>
> > https://lists.yoctoproject.org/listinfo/poky
> > 


[-- Attachment #2: Type: text/html, Size: 4119 bytes --]

Re: do_fetch error with gcc-cross-initial_4.7

[Richard Purdie]

On Wed, 2012-09-19 at 12:02 +0300, Florin wrote:
> Trying to build something today and still takes forever and crashes at
> fetch. Looking at the gcc-4.7.inc I see it's still from git.
> What would the resolution be for this issue?

The recipe is now using a release tarball. There is talk of ways to
improve the fetcher in this regard in future too but that is probably
1.4 work at this point.

Cheers,

Richard

Re: do_fetch error with gcc-cross-initial_4.7

[Gary Thomas]

On 2012-09-24 04:51, Richard Purdie wrote:
> On Wed, 2012-09-19 at 12:02 +0300, Florin wrote:
>> Trying to build something today and still takes forever and crashes at
>> fetch. Looking at the gcc-4.7.inc I see it's still from git.
>> What would the resolution be for this issue?
>
> The recipe is now using a release tarball. There is talk of ways to
> improve the fetcher in this regard in future too but that is probably
> 1.4 work at this point.

Has this change actually been merged to master?  I just updated and
did not see it.

... anxiously awaiting this change to arrive in master :-)

-- 
------------------------------------------------------------
Gary Thomas                 |  Consulting for the
MLB Associates              |    Embedded world
------------------------------------------------------------

Re: [poky] do_fetch error with gcc-cross-initial_4.7

[Martin Jansa]

On Mon, Sep 24, 2012 at 1:19 PM, Gary Thomas <gary@mlbassoc.com> wrote:
> On 2012-09-24 04:51, Richard Purdie wrote:
>>
>> On Wed, 2012-09-19 at 12:02 +0300, Florin wrote:
>>>
>>> Trying to build something today and still takes forever and crashes at
>>> fetch. Looking at the gcc-4.7.inc I see it's still from git.
>>> What would the resolution be for this issue?
>>
>>
>> The recipe is now using a release tarball. There is talk of ways to
>> improve the fetcher in this regard in future too but that is probably
>> 1.4 work at this point.
>
>
> Has this change actually been merged to master?  I just updated and
> did not see it.
>
> ... anxiously awaiting this change to arrive in master :-)

It is in oe-core/master
http://git.openembedded.org/openembedded-core/commit/?id=5e03d1e83d0536a2fc69a88d3e5407108836203f

Re: do_fetch error with gcc-cross-initial_4.7

[Gary Thomas]

On 2012-09-24 05:19, Gary Thomas wrote:
> On 2012-09-24 04:51, Richard Purdie wrote:
>> On Wed, 2012-09-19 at 12:02 +0300, Florin wrote:
>>> Trying to build something today and still takes forever and crashes at
>>> fetch. Looking at the gcc-4.7.inc I see it's still from git.
>>> What would the resolution be for this issue?
>>
>> The recipe is now using a release tarball. There is talk of ways to
>> improve the fetcher in this regard in future too but that is probably
>> 1.4 work at this point.
>
> Has this change actually been merged to master?  I just updated and
> did not see it.
>
> ... anxiously awaiting this change to arrive in master :-)
>

Indeed, this is already in master (and I missed it!)  Sorry for the
noise, thanks for getting this going.

-- 
------------------------------------------------------------
Gary Thomas                 |  Consulting for the
MLB Associates              |    Embedded world
------------------------------------------------------------