I have not had much chance to look dig into the socket code of late, but
here is my situation. If someone loses link for whatever reason and does
not reconnect right away, the load on my system goes to 1.0 average. This
is not a process load, in fact top leads me to belive that the system is
95+% idle, which is certainly not the case. Now, I *think* this has to do
with circle's method of polling cons for "new data", the kernel then polls
and polls and polls, driving up the load.
My system is a FreeBSD 2.2.5 Release running on an AMD K-5 with circle
3.0bpl11 as the base mud. I don't believe that anyone has really mucked
with the socket calls and frankly with finals I don't know if I am going to
have time to figure it out.
Has anyone else experienced this problem or something similar?
-Steve
--
Stephen C. Thompson
stevet@vt.edu
http://www.acm.vt.edu/~stevet
Dark Pawns: pawns.guru.org 4000
+------------------------------------------------------------+
| Ensure that you have read the CircleMUD Mailing List FAQ: |
| http://democracy.queensu.ca/~fletcher/Circle/list-faq.html |
+------------------------------------------------------------+
This archive was generated by hypermail 2b30 : 12/08/00 PST