timer spun

[Guest]

I have a shell with loads of users and loads of background processes- you can see some data at the end - this is the state at the moment I post this message.
I recently upgraded my eggdrops from version 1.3.28 to 1.6.6 and now, when the machine is heavily uptimed every now and then there is an error "timer spun 3 minutes" or something like that (they don't give me this error now- the server isn't uptimed enough now).
And when they give that message all the sockets are closed and reopened again- this means that bots looses connection with the other bots and with the irc server...
And here comes my question- how could I disable this? How should I modify the sources, or is there an option to switch it off from the config file? The bots really do they work despite of the uptime of the shell- sometimes a little (or not so little.. ) bit slowler, but things get much worser if they disconnect every now and then...


And here are some stats- I post them because I'm really proud of this machine- you will be too after you read this:) You won't believe how fast it is from inet:)- I don't know it's connection speed, but it's fast... When it gets much more uptimed (this takes from 7 days to 20 and more even! )- they just restart it and everything is OK again.

cat /etc/passwd | wc -l
64755
ps -A |grep eggdrop |wc -l
253
ps -A |grep psybnc |wc -l
76
ps -A| wc -l
427
12:03pm up 4 day(s), 21:29, 11 users, load average: 40.23, 52.67, 36.63

[ppslim]

High numbers, do not make impressive stats. High numbers just make a sluggish system.

Lots of users make a system sluggish.
Low uptimes make for less uptime of your bot.

There nothing you can do at all about timer spin, this is a result of the above so called impressive stats.

The timer spin problem has never atualy been worked out. It is not a bug, just somthing that happens on systems.

This mainly occurs on systems with very high load averages, as the above.

I would switch shell providers straight away, I would not even have waited for the load average to reach 10, let alone 36 to 53.

idealy, systems like the following should work well.

# uptime
12:33pm up 182 days, 21:51, 16 users, load average: 0.39, 0.27, 0.27
# cat /etc/passwd | wc -l
144
# ps -A |grep eggdrop |wc -l
31
# ps -A |grep bnc |wc -l (changed for bnc, ezbounce and ezybnc)
12
# ps -A| wc -l
261

Even this isn't my best, but you get the idea of why this system is better than yours.

[Guest]

I completely agree with you that this machine isn't "the best a man can get", but it's the only I have. The machine I use at my faculty doesn't allow eggdrops... So I have no choice. I'm from BG so I can't buy a shell...
I just asked whether there is a possibility to disable the function checking for timer spun or if I can't there is another question I'd like to ask- what previos version doesn't check for timer spun like 1.3.28...
Thank U again!

[ppslim]

There is no option to disable it, and I am unable to find a version that doesn't check for it.

From my understanding, it is a requirment for the bot to check for "timer spin" (so defined by the message), as failure to check and fix, can cause more problems than it fixes.