Andreas Dilger wrote: >On May 14, 2002 16:54 +0100, Stephen C. Tweedie wrote: > > >>On Tue, May 14, 2002 at 09:43:08AM -0600, Simon Gao wrote: >> >> >>>Can someone help explain why sometimes ext3 failed to recover from a >>>sudden loss of power (ie. has to run lengthy fsck on paritions)? This >>>happened to Redhat 7.2, 7.3 with latest stable kernel. >>> >>> >>This can be for several reasons, but the fsck will always tell you why >>it is doing it, so you'll need to keep your eye open for those >>messages. >> >>Reasons can include disk corruption caught by the filesystem --- the >>fs will mark the partition as having seen errors, and the next fsck >>will do a full check to recover. Or, the filesystem may be marked to >>have regular fscks on it, either every so-many mounts or after a >>certain interval has passed --- "man tune2fs" to see how to change >>those limits or to disable them. >> >>e2fsck will always print a message something like "/dev/foo contains a >>filesystem with errors, check forced" or "/dev/foo has been mounted 33 >>times since the last check, check forced" to tell you why it is doing >>the check. >> >> > >It may also be a case of "I _thought_ this filesystem was ext3, but it >is really ext2". Simon, you should check the contents of /proc/mounts >and also the output from "tune2fs -l /dev/X" and see that it has the >flag "has_journal" and a journal inode is listed. > >Cheers, Andreas >-- >Andreas Dilger >http://www-mddsp.enel.ucalgary.ca/People/adilger/ >http://sourceforge.net/projects/ext2resize/ > > Thanks for all good points. I will definitely double check the machines having this kind problem and let you what I found out. Simon Gao
