Peter Weilbacher wrote: > After unmounting I tried another fsck: > > Phase 1 - Check Blocks, Files/Directories, and Directory Entries. > (chklog) FSCK Duplicate reference to 4 block(s) beginning at offset 2894036 >found in file system object IA16. > > (chklog) FSCK Inode A16 has references to cross linked blocks. > > (chklog) FSCK Multiple metadata references to 4 blocks beginning at offset >2894036 have been detected. > > Duplicate block references have been detected in meta-data. > FSCK CANNOT CONTINUE. > File system is dirty. > (chklog) FSCK processing terminated: 3/1/2002 22.10.58 with return code: 0.
The cross linked blocks may explain the "diRead: di_nlink is zero" errors. Inode A16 is a special inode which describes where the inodes are stored. The cross linked blocks means that disk blocks for this file are also being used for something else. This can cause all kinds of corruption. I do believe that fsck should be able to trim the bad stuff out of here and salvage the rest of the file system. We will investigate why it didn't. > > The xchkdmp'ed log is also appended. Stupid of me to unmount it > *before* doing another backup, because now I cannot even mount it > anymore: Shoot me for not responding sooner. You should be able to mount the file system read-only even if it was dirty. You may have had some errors, but should have gotten most of your data. > > > mount -v /users2 > mount: wrong fs type, bad option, bad superblock on /dev/hda1, > or too many mounted file systems > > I also append the file with the xpeek output of the superblocks. Did > anyone write a How-To about cases like this, and what one has to do > to get the data back? Or does one have to be a JFS programmer to > know enough to do it? I have most of the data on last night's > backup, but today's work would be lost... I think the read-only mount is what you were looking for. Maybe we should add a message to fsck when it prints "FSCK CANNOT CONTINUE" to mount the filesystem read-only and recover what you can. -- David Kleikamp IBM Linux Technology Center _______________________________________________ Jfs-discussion mailing list [EMAIL PROTECTED] http://www-124.ibm.com/developerworks/oss/mailman/listinfo/jfs-discussion
