https://bugzilla.novell.com/show_bug.cgi?id=293431#c4
Bernhard Kaindl
PS: Just for the record: the most recent ntfs-3g driver does not make the volume dirty even if it is killed by the halt script.
I'm reducing the severity to normal, because nothing fatal seems to happen tough with the current version, so I see no justification for major anymore after this. Of course we'll address the issue: About automount of ntfs-3g at boot: When it's based on a block device, automounting of ntfs-3g should always work because (at least 1.710) whecks wether fuse has been started already propery and if not, it loads the fuse module, creates /dev/fuse and mounts the fusectl filesystem, if anything of that is missing. As it's running with root privileges at early boot anyways, it also has the user rights to do so. Mounting other fuse filesystems and (generally any fuse filesystem which is not based pure block devices) do have other prerequisites before they can be mounted. E.g. all network-based fuse filesystems can only be mounted after the network has been started and the neccessary network links are up and running and routing packages to the original source of the filesystem. To handle entries for them properly, we'll have to add the respective filesystem types to these lines: $ grep nonfs /etc/init.d/boot.localfs local mount_opts="-t nonfs,nonfs4,noproc,nodevpts,nosmbfs,nocifs -O no_netdev" mount -av -t nonfs,nonfs4,noproc,nodevpts,nosmbfs,nocifs -O no_netdev umount -avt noproc,nonfs,nonfs4,nosmbfs,nocifs,notmpfs,nosysfs || { On shutdown, it's also needed to unmount these filesystems which depend on the network before the network is shut down. To be able to handle all this in one script, we'd have to add # Required-Start: $network To the new /etc/init.d/fuse, so that it starts after network is started, and then we could implement automatic mounting of remote fuse filesystems then, like it's done in the init scripts of nfs, smbfs and cifs. The init script of fuse could take care of automounting and unmounting of all fuse filesystems in one script instead of needing a separate script for each of them. That means that you cannot disable mounting them individually without having to configure auto/noauto each corresponding entry /etc/fstab to enable or disable the mounting, but I think that should be ok. I guess however, that the list of filesystems supported by the fuse fstab-mount script should extendible by installing another fuse filesystem, e.g. by installing the not-yet packages gmailfs for example. Even if SuSE (I don't know if there may be issues) may possibly never ship gmailfs (or may, no idea), additional fuse filesystems can be provided thru the openSUSE buildservice and be automatically integrated into automounting at boot. I'm not sure which way for this would be best, if we should simply add all which are known today or if we should have /etc/sysconfig/fuse. Thanks for the hint about swap files and loop devices: Yes, doing that should be good, we should take care of that. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.