In message <[EMAIL PROTECTED]>, Gianni Johansson <[EMAIL PROTECTED]> writes
I was looking at my connection time stats last night:

http://localhost:8888/servlet/nodestatus/diagnostics/connectingTime/day

The mean values for the last two days are 11000 and 15000.

This is bad news, since the default routeConnectTime is set to 10000.

If I understand the connection management code correctly, this means that most real (i.e. non cached) outbound connection attempts will fail with a timeout. The fact that the connection will eventually succeed and will be cached doesn't help because the request state machine is already banging on another node reference which will also fail...

wash. rinse. repeat.

It's hard to figure out whether this is a cause or a side effect of the
current massive CPU overuse problem. It definitely isn't helping network congestion any.


Are most people seeing connection times this high?

--gj

Yes. Never less than about 8000, majority are recorded (apparently) as 42180 but took longer or timed out in some way (but I don't know if timed out ones are counted).



p.s.:
Load average: 15
System PII 233, 192M of RAM
env page with system info attached.



Much slower machine, but low limits and works with a low number of connections and load average 0.2 to 0.8 currently.
--
Roger Hayter
_______________________________________________
devl mailing list
[EMAIL PROTECTED]
http://hawk.freenetproject.org:8080/cgi-bin/mailman/listinfo/devl

Reply via email to