Problems when cp/mv large numbers of files/time:...alloc_streaming_cluster...failed
on my ultra5 I'm frequently experiencing problems with the filesystem
(error -> ext3 aborted journal -> remounted readonly)
* following a cutout from kern.log:
kernel: kjournald starting. Commit interval 5 seconds
kernel: EXT3 FS on sdc1, internal journal
kernel: EXT3-fs: mounted filesystem with ordered data mode.
kernel: pci_iommu: alloc_streaming_cluster of npages(131) failed!
kernel: end_request: I/O error, dev sdc, sector 12600
kernel: Aborting journal on device sdc1.
kernel: ext3_abort called.
kernel: EXT3-fs error (device sdc1): ext3_journal_start_sb: Detected
kernel: Remounting filesystem read-only
kernel: __journal_remove_journal_head: freeing b_committed_data
This occurs only when creating a large number of files per time
(tar xvf linux.tar will do ...). Typically the tar will run for some
time (1-2 mins) before the problem occurs. Typical values for
npages(...) failed are around 130. Operations with a lower creation rate
(e.g. larger files of 1-2 GB) run without problems.
The actual version of linux is 220.127.116.11 (from kernel.org) - I didn't
remember seeing this in 2.6.12(.4). The disk is a Maxtor 200GB SATA
disk, the controller is a Silicon Image Sil 3112.
Any hints or clues ?