[Mageia-sysadm] Needs locking iurt chroot tarballs

[Pascal Terjan]

On Tue, Feb 28, 2012 at 09:43, Pascal Terjan <pterjan at gmail.com> wrote:
> On Tue, Feb 28, 2012 at 08:02, Thierry Vignaud
> <thierry.vignaud at gmail.com> wrote:
>> Hi
>>
>> Yesterday evening, iurt chroot was rebuild due to new rpm.
>> However some packages failed to build:
>> http://pkgsubmit.mageia.org/uploads/failure/cauldron/core/release/20120228010948.fwang.valstar.11974/log/botcmd.1330391389.jonund.log
>>
>> D: [iurt_root_command] Running iurt_root_command --untar
>> /home/iurt/chroot_cauldron.x86_64.tar.gz
>> /home/iurt/chroot_tmp/iurt/chroot_cauldron...
>> D: [iurt_root_command] Uncompress tarball
>> gzip: stdin: unexpected end of file
>> tar: Unexpected EOF in archive
>> tar: Unexpected EOF in archive
>> tar: Error is not recoverable: exiting now
>> I: [iurt_root_command] ERROR: Uncompress tarball
>>
>> There're 2 bugs:
>> 1) build continues despite the tar failure
>> 2) we should:
>> - either lock the chroot tarball when rebuilding it and when unpackagint it
>>  in order to prevent that to happen again
>> - ensure chroot tarball is updated atomically (temp tarball + single mv)
>
> There is supposed to be only one iurt per version-arch on a given
> machine at a time, so if two iurt run on the same machine they use
> different chroot tarballs

One case I can thing when this is not true:
If a package fails to build on one arch, it does not get properly
cancelled. ulri just forgets about other arches but does not stop the
build which is in progress, it just deletes the lock and another build
can start in parallel on the same machine...

This does not seem to be the case here, but there may be another
similar situation