Le Tuesday 09 December 2014 16:36:53, The Wanderer a écrit : > On 12/09/2014 at 10:09 AM, Chris Bannister wrote: > > On Tue, Dec 09, 2014 at 09:48:58AM +0100, Frédéric Marchal wrote: > >> Now, is it possible to run fsck during shutdown? Users have been > >> asking for this for at least 10 years. Is it now acceptable, > >> possible, tolerated? > > > > That sounds like a recipe for disaster. Do you mean *before* > > shutdown? > > Obviously, "during shutdown" means "during the shutdown process", i.e., > during the sequence of shutting-down-the-system steps which takes place > in response to a "shutdown" command. > > That sequence already does several things prior to actually shutting > down the system; perhaps most obviously, it tells various "services" to > stop cleanly, kills other processes, and unmounts filesystems. There > seems as if there should be no conceptual reason why it shouldn't be > possible to add an additional "run a fsck" step into that sequence, > probably after the unmount and before the final shutdown itself. > > ...except that fsck of root during the boot process is possible only > because root hasn't been mounted yet, because we're still in the > initramfs and haven't pivoted into the real root yet. So making that > possible during shutdown would probably require setting up another > ramdisk during the shutdown process (which sounds like a bad idea), > pivoting into it, unmounting the original root, and then triggering the > fsck...
The partition only need to be remounted read only. if I'm understanding it correctly. That's what /etc/init.d/umountroot does during the shutdown sequence. So everything is in place to run fsck just before /etc/rc0.d/K10halt. Now two questions remain: 1) how to invoke an additional hypothetical /etc/rc0.d/K10fsck on demand? 2) is it wise to run fsck at that time? I have seen strong opposition in the past. Mostly turning around the risk that the user would switch the power off or the power supply would fail resulting in a damaged partition. As the risk seems as high during the boot sequence, I don't understand the opposition. > All in all, while it *might* be possible, I don't think it sounds like > that would be worth the trouble. That's probably right too... Frederic -- To UNSUBSCRIBE, email to debian-user-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org Archive: https://lists.debian.org/201412100730.53734.frederic.marc...@wowtechnology.com