Hi,
I am observing "random" pytest failures when building one of my packages (astropy) [1]. This seems hardware related, since the test passes well
on some buildds (osuosl-02, osuosl-04), while it fails on other buildds (osuosl-03, osuosl-05). Checking the buildd info for these machines [2,
3] show however no relevant differences; both are Quad Core Loongson
3A3000 CPUs; with 12 GB or 16 GB RAM, from the same sponsor, and also
(in the latest case) running the same kernel version.
I am now wondering what the differences between these machines are, what
the cause for these failures may be and what the proper handling of
these failures would be. Can/shall one pin the build to a specific set
of buildds that are known to work? Or just giveback the builds until
they succeed? Or is this something upstream can/should fix?
Please Cc me in the answer, because I am not subscribed to the list.
Best
Ole
[1]
https://buildd.debian.org/status/logs.php?pkg=astropy&arch=mips64el
[2]
https://db.debian.org/machines.cgi?host=mipsel-osuosl-04
[3]
https://db.debian.org/machines.cgi?host=mipsel-osuosl-05
--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)