Am 15.07.2005 um 07:09 schrieb Mitchell Laks: > I noticed that never happened in the old days when I ran redhat/fedora. They > did away with this routine fsck (sometime around 6.2 -> 7 or 7.2 transition > as I recall) when they switched from ext2 to ext3 as I recall. > > This occurs during /etc/rcS.d/S30check-fs entry > entry during boot, right after file system mount, I believe. > > Is this neccessary on debian if we have journaled file systems? After all it > made sense for ext2 but do we need this for ext3?
The regular fsck is absolutely independent of journaling. The effect of journaling is, that you do not have to scan your _complete_ file system for corruptions after your system has been restarted without closing the file systems properly. The file system driver itself can "replay" the journal on mount time, which means it only has to finish or roll back those transactions that were not cleanly finished. A journaling file system will not and cannot protect you from any other corruptions. This means that a regular fsck is still the (only) way to detect corruptions caused by software bugs, hardware errors, etc. I've made the experience that a regular fsck will usually take place exactly at a time when I do not _have_ the time to wait for it. So I usually disable this feature and do a planned fsck from time to time. Regards, Dennis -- Send personal mail to [EMAIL PROTECTED] only. Off-list mails to [EMAIL PROTECTED] will not reach me. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]