Re: Mirror issues on HPPA

Hi Christoph,

On Sun, 2026-06-21 at 18:41 +0200, Christoph Biedl wrote:
> Affected are these URLs:
>=20
> http://ftp.de.debian.org/debian-ports/pool-hppa/main/g/gcc-16/libgomp1_16=
.1.0-2_hppa.deb
> http://ftp.de.debian.org/debian-ports/pool-hppa/main/g/gcc-16/libgcc-s4_1=
6.1.0-2_hppa.deb
> http://ftp.de.debian.org/debian-ports/pool-hppa/main/g/gcc-16/libcc1-0_16=
.1.0-2_hppa.deb
> http://ftp.de.debian.org/debian-ports/pool-hppa/main/g/gcc-16/libatomic1_=
16.1.0-2_hppa.deb
> http://ftp.de.debian.org/debian-ports/pool-hppa/main/g/gcc-16/libstdc%2b%=
2b6_16.1.0-2_hppa.deb
>=20
> Now I'm fairly certain this is outside my control. A check on a second
> mirror returned the same result. Can someone take a look into this?
> (...)
> Further information, based on the mtime: The "received" package was
> installed on the mirror on 1781589787 (2026-06-16T06:03:07) and
> appearently replaced with the "expected" content on 1781805793
> (2026-06-18 18:03:13.886190374 +0000). While TTBOMK there is no
> guarantee a Debian package file name is unique in a mirror, violating
> that causes confusion.

The issue is that gcc-16 was built and uploaded twice for the same version:

https://buildd.debian.org/status/logs.php?pkg=3Dgcc-16&arch=3Dhppa

Since Dave was quicker than me, he already triggered the BinNMU to fix the =
issue.

The reason for this to happen is that Debian Ports has just MiniDAK instead
of DAK with the former not being as robust as the latter, so a simple situa=
tion
like this can cause such issues.

Adrian

--=20
 .''`.  John Paul Adrian Glaubitz
: :' :  Debian Developer
`. `'   Physicist
  `-    GPG: 62FF 8A75 84E0 2956 9546  0006 7426 3B37 F5B5 F913