On 04/09/14 17:29, Russ Allbery wrote: > Ximin Luo <infini...@pwned.gg> writes: > >> I do have a bunch of /media/x/y/z mounts that fail, but that is not >> easily fixable apart from removing it, but I don't want to do that. > > This is what's creating the problem. I'm not saying that it should be, > just that this is the reason why it's failing. Arguably this is a bug in > systemd, although the counter-argument is that you are supposed to use the > "nofail" option in /etc/fstab if you want boot to continue even if a > particular mount that is configured for automatic mounting at boot is not > available. It's just that sysvinit never implemented those semantics, so > people never bothered and are caught by surprise when this stops working. >
OK, I am sympathetic to this. Perhaps systemd could (at least for now) flash big obvious warnings about this if it detects a failing mount without nofail? I do get some logs about this, but it's completely unclear why it should cause openssh/lighttpd to fail. > I'm unclear on the best way to address this problem. We may need to add > implicit "nofail" on all mounts to deal with the upgrade issue. > >> Besides, sysv-init worked fine with that set up. If this is indeed the >> reason why it is failing, it is still a bug in systemd (or the systemd >> service definition for openssh) that should be fixed. > > The systemd service definition for openssh cannot, so far as I know, do > anything about this. > >> I filed a similar bug for lighttpd too, but I guess less people are >> paying attention to that. > > It's probably the same issue. lighttpd can't fix the problem in the > packaging for that service, I think. > > As a workaround, I believe adding the "nofail" option to mounts that may > or may not succeed will make the problem go away. > Thanks, I will try that! Might we keep this bug open so other people can find this useful bit of info? X -- GPG: 4096R/1318EFAC5FBBDBCE git://github.com/infinity0/pubkeys.git
signature.asc
Description: OpenPGP digital signature