I cannot contribute to your problem but the problem originally discussed in 
the thread you are referring to at
> https://mail-index.netbsd.org/tech-kern/2012/10/25/msg014236.html
has been tracked down (by me) and solved by hannken@.
It was RAID specific and due to WAPBL not honoring block alignments 
when flushing the log. It saturated the RAID components, not the CPU.

Reply via email to