On Tue, Feb 01, 2022 at 08:20:45PM -0500, Sean Caron wrote: > Thank you for the detailed response, Dave! I downloaded and built the > latest xfsprogs (5.14.2) and tried to run a metadump with the > parameters: > > xfs_metadump -g -o -w /dev/md4 /exports/home/work/md4.metadump > > It says: > > Metadata CRC error detected at 0x56384b41796e, xfs_agf block 0x4d7fffd948/0x1000 > xfs_metadump: cannot init perag data (74). Continuing anyway. > > It starts counting up inodes and gets to "Copied 418624 of 83032768 > inodes (1 of 350 AGs)" > > The it stops with an error: > > xfs_metadump: inode 2216156864 has unexpected extents Not promising - that's a device inode (blk, chr, fifo or sock) that appears to have extents in the data fork. That's indicative of the inode cluster containing garbage, but unfortunately the error propagation from a bad inode appears to abort the rest of the metadump. That looks like a bug in metadump to me. Let me confirm this and work out how it should behave and I'll send you a patch to avoid this issue. Cheers, Dave. -- Dave Chinner david@xxxxxxxxxxxxx
