Long pauses in getblk

Anyone heard of this?

FreeBSD 8.1 system with 24GB and a 3ware raid card is under a heavy disk I/O load. ~300MB/s written to a large array.

Pretty regularly, disk I/O drops to zero and nearly every running process is hanging in getblk. Whatever this condition is, it can last up to 10 seconds.
 
SMART status?
Errors in dmesg?
Do you use AHCI? In that case, I would try again without as some discs can have buggy firmware.
 
Thanks for the suggestions. Smart status looks OK and there are no errors in the message logs. I don't believe we're using AHCI but its a good point and I will check.
 
Back
Top