>From: [email protected] > On Thu, Mar 25, 2010 at 12:55:50AM -0700, Elliott Mitchell wrote: > > > > Which is why I wasn't writing about the mount count. I was writing about > > the check interval/check time. With systems that reboot less than once a > > month, the mount count will never be reached. Instead it will be the > > check interval and worse yet, with infrequent reboots, *all* of the check > > intervals will expire at once (even if they start spaced out, they will > > tend to merge together). > > Well, fsck (which is the program which runs e2fsck for each partition) > is no longer something I maintain (it's been moved to util-linux-ng, > which is maintainer by Karel Zak). Granted, fsck would need some way > of telling e2fsck not to do time- or mount-based check if some other > check has taken a long time already, but at least part of the > functionality would have to be done in fsck --- and that's not no > longer my problem. :-)
Interesting. Well, as of lenny, both /sbin/fsck and /sbin/e2fsck were part of e2fsprogs, so e2fsprogs was the correct package to file the bug report against. As you yourself admit, this will in fact require a (likely small) change to e2fsck so either way it is correct to report this against e2fsck. > Quite frankly, I can't get terribly excited about this problem, since > if you have file systems on separate disk spindles, the file system > checks are done in parallal anyway. And if you are using LVM, the > right answer is to use an snapshots, and then schedule the checks > using cron during periods of low usage, maybe once a month. The simple approach that comes to mind is simply limiting it to 1 check per spindle per boot; mandatory checks due to filesystem problems shouldn't face this limit, but should consume the check count. I'd allow the limit to be changed, perhaps specifying a limit of filesystems comprising X% of any drive could be an option. Seeing how this now also effects util-linux, should I file a new bug against util-linux? Does the Debian BTS currently allow for cloning a report or a single one being on multiple packages? -- (\___(\___(\______ --=> 8-) EHM <=-- ______/)___/)___/) \BS ( | [email protected] PGP F6B23DE0 | ) / \_CS\ | _____ -O #include <stddisclaimer.h> O- _____ | / _/ 2477\___\_|_/DC21 03A0 5D61 985B <-PGP-> F2BE 6526 ABD2 F6B2\_|_/___/3DE0 -- To UNSUBSCRIBE, email to [email protected] with a subject of "unsubscribe". Trouble? Contact [email protected]
