XFS write speed drop

From: Jan Engelhardt
Date: Fri May 19 2006 - 05:34:34 EST


I have noticed that after an upgrade from 2.6.16-rcX -> 2.6.17-rc4, writes
to one (hdc) xfs filesystem have become significantly slower (factor 6 to
8), like if -o sync (or funky journal options on ext3) was on. Also, reads
would stall until writes have completed. I would only expect such behavior
when /proc/sys/vm/dirty_* are set to high values (like 95%, like I do on a
notebook). hda remained fast.

It eventually turned out that it are the log barriers; hda does not support
barriers and XFS gave a nice kernel message hint that pointed me to try out
a mount flag. -o nobarrier makes it fast again.

Jan Engelhardt
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/