"Cristian Rodríguez" írta:
El 25/05/14 11:23, Istvan Gabor escribió:
How can a mount be a dependency? This is stupid.
No, it is not stupid, it is just you do not understand its purpose..
I do understand but I think a setting in the fstab file (noauto) should have precedence to some dependency. If the dependency can not be met the system should claim about it, not disregard the setting.
mount points are dependencies and have always been .
for example:
apache Requires=srv.mount to work nscd requires var-run.mount
and, on and on... you name it.
Yes, and if you remove the hard disk with the file system required, the system claims, and will not reinstall the hard disk. (:
1. Should files be located on the given file system systemd should give error message, not override the administrator's setting.
Systemd does understand what files are where, it only keeps track on what is mounted, what is not and what needs to be at a particular point.
2. systemd should stick to conventional behavior.
because you say so, I see.
Let me explain. Previously (before systemd, or gnome file manager) if I told the system not to touch a file system (noauto setting), the file system was not mounted. Now I have systemd and for some reason I decide to use that setting and expect that the file system will not be mounted. I don't expect that the developers invented some other options, or my setting will not be appreciated. I will discover it only after the mounting (which can lead to data loss, or may prevent successful recovery) occurred.
3 You cannot expect a user to read systemd code generator (I even don't know what it is).
Nope, there are manual pages that describe what they do.
It is still not enough. If I update my openSUSE system which results in replacing sysvinit to systemd, I don't care (as most of the users) what's behind the curtain, I care only if the system works well, and I expect it to work similarly (from a user's view) as before. That is I expect that the setting (noauto) has the same result as before, unless the change is mentioned in the release notes. Was it? (I don't know.)
I think systemd developers should change they program so that it does not touch noauto marked file systems.
There is nothing to change, the OP is using an ancient version in which this behaviour is observed.
If it is so, then it was a fixed bug. But you also mentioned some other option, nofail. This makes me puzzled. Cheers, Istvan -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org