I was just extracting a 96MB tar file ( tar -xWf backup.tar ), the cpu load was 99% for a long time. I then stopped it and tried again, but this time this popped up in my ssh session: -- Message from syslogd at rock at Sat Feb 18 00:47:05 2006 ... rock kernel: Assertion failure in __journal_unfile_buffer() at fs/jbd/transaction.c:1520: "jh->b_jlist < 9" -- A kernel panic dump is displayed on the servers monitor. Here is what I can remember: Call trace: jurnal_commit_transaction+<something> [jdb] Is the panic dump stored anywhere? When I rebooted, the raid set was dirty and it was re-synced, it did not mark a drive faulty or started using a spare, the raid set still uses the same two disks and now everything seems normal, except this is the second time it does this. It is a fresh install of Cent-OS 4.2 from the DVD image and with the latest updates from yum. The server has 4 80GB software mirrored IDE disks, 2 + 2 spares, one on each channel of a HPT374 controller. I have no idea about what to look for; I have tested the disks many times with the vendor tools and no errors have been found.