Package: ssmtp
Version: 2.64-4
Severity: normal
In some rare cases the "hostname" command can't answer, making the
postinst script crash.
I stumbled upon one of these cases when I tried to install ssmtp in a
Xen vm during image creation, through a "role" script (those are scripts
run after debian base system is installed, but before first boot - it's
the final part of a Xen guest install with xen-tools).
These scripts are run on the host, chrooted in the mounted guest's
system. I'm not sure what happened, but since the hostname and domain
name are correctly configured before xen-create-image runs the role
scripts, I changed the postinst script like this:
hostname="${RET:-`grep -m 1 \`cat /etc/hostname\` /etc/hosts | awk
'{print $2}'`}"
...and could modify the script, run "dpkg --configure -a" and undo the
changes all through the role script to fix things even before the first
boot, but I don't know if this rare case justifies changing the postinst
script for everyone, since in a "normal" setup, if at least the
/etc/hosts is correctly written, the "hostname" command would work and
the postinst script wouldn't crash.
As for the original poster, I can't see any other case where the
"hostname" command would fail, maybe /etc/hosts was corrupted or missing.
Hope it helps.
--
Raph
--
To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
