* Neil Williams <[email protected]> [090222 15:16]:
> reprepro 3.8 has apparently started to leave lockfiles in place if the
> previous command ended with errors:
This should not happen. (And in the whole test-suite does not happen).
One should only see a lock-file iff:
1) reprepro segfaulted or otherwise was killed by an uncatched signal.
2) reprepro is still running (like calling reprepro from a hook script
or things like that).
Note that 1) is currently quite likely as
> ii libarchive1 2.6.1-1 Single library to read/write
> tar,
has bug #516577 causing reprepro to segfault (at least when Contents
file generation is enabled).
On the other hand the "There have been errors!" looks like at least that
command did not segfault.
> ERROR: '/opt/reprepro/grip/incoming/acpi-support-base_0.109-11em1_all.deb'
> cannot be included as
> 'pool/main/a/acpi-support/acpi-support-base_0.109-11em1_all.deb'.
> Already existing files can only be included again, if they are the same, but:
> md5 expected: 6b28df4abf8e50f2ed41b4160905cb8b, got:
> 754b37be5d32b1a22b3944d6b95bdde5
> sha1 expected: a44d321331cb940c770c0e89d296c1a90fd98835, got:
> 19d4cdb48e1c4500e80477d5e0b814f5a721ae37
> sha256 expected:
> 92ded78d200ed98c6e0836c5f3ed8401e3bd73286574e276bd0dedaa899aa432, got:
> 96b4d9f2bb19c1372e16c58452d0f7926f0252b2b400abb23d98b498af956826
Even with the same error message, I do not get left over lock files.
So I am a bit confused. Does your script perhaps do some silent export
somewhere that could segfault because of libarchive and then leave the
lockfile or things like that?
> reprepro calls because deleting a lock file by brute force is not a
> particularly good idea but appears necessary.
It normaly is no good idea. As this should only happen when aborting,
the database might be in inconsistent state. I suggest running some of
the checks from file:///usr/share/doc/reprepro/manual.html#maintainance
in that case.
I'd be really eager to hear if you have some minimal example when this
happens or otherwise some idea how I could reproduce that.
Thanks in advance,
Bernhard R. Link
--
To UNSUBSCRIBE, email to [email protected]
with a subject of "unsubscribe". Trouble? Contact [email protected]
