Kernel panic...

[John <john%sixgirls.org@localhost>]

Hello,

For the last few weeks, my server has been crashing after about a week of
uptime. Although I have a duplicate system, I cannot reproduce the problem
locally (probably since I cannot simulate the load).

Anyway, I compiled in the kernel debugger, and when the crash happened
tonight, I got something like:
Out of memory in kmem_map
(something like) in process httpd

According to netstat -m, I have (right now) 7/188 mapped pages in use. How
does this translate to NMBCLUSTERS? If I wanted to make sure this would
NEVER happen again, what's a reasonably ridiculous value for NMBCLUSTERS?

And why isn't this a recoverable error?

Thanks,
John
-- 
The proof of a system's value is its existence.