Hi Boyuan! On Mi, 2016-09-14 at 10:43 +0800, Boyuan Yang wrote:
(...) > to download *once*. The dir-chroot creating process should either > fail > immediately or re-run the debootstrap process again with a warning > sent to the logger, > but it just continued as if everything is fine *until* the mkdir > error > happens (which is due to the unclean exit of debootstrap, maybe?) Hmm don't know. I tested around quite a lot lately, and did not experience such a behavior. Could you try to reproduce this with 1.0.21? This may have some nicer logging for these calls, and maybe an excerpt of this log might shed some light ;). Thx! P.S.: Ah: <= 1.0.21 has a little bug in the service file (mini-buildd might not use the default log level), so you might want to apply this first (or wait for 1.0.22): diff --git a/debian/mini-buildd.service b/debian/mini-buildd.service index 6809dda..0a184f2 100644 --- a/debian/mini-buildd.service +++ b/debian/mini-buildd.service @@ -4,6 +4,7 @@ Documentation=man:mini-buildd(8) After=remote-fs.target [Service] +Environment=MINI_BUILDD_OPTIONS="--verbose" EnvironmentFile=-/etc/default/mini-buildd User=mini-buildd ExecStart=/usr/sbin/mini-buildd $MINI_BUILDD_OPTIONS