Weird Problem with Serverlag on Eggdrop 1.6.17

^CJ^ · Post by **^CJ^** » Sat Sep 10, 2005 10:52 am

Hi,

i recently experienced a weird problem on my shell:

using a script that checks the lag the bot has to IRC, i got totally different results on eggdrop 1.6.17 and 1.6.15. I had one bot of each version connected from the same shell to the same irc server, the .15 bot hat a latency of 0.8-4ms, the .17 bot had 48ms+! I used this script: http://www.egghelp.org/cgi-bin/tcl_arch ... oad&id=971 and also another one to which i have no url but which returned exaclty the same results...

I also found a workaround for the problem after a lot of experimenting: after i copied the server.so from the .15 modules/ folder to .17 and restarted, the .17 bot had the same low ping!

I tried that on 2 different shells and the result was the same. Is anyone else experiencing the same problem? Maybe this is just some sort of compatibility problem with the script? I'm not sure whether the bot really has that higher latency or if its just the displayed value thats different.

pls help

demond · Post by **demond** » Sat Sep 10, 2005 12:58 pm

that's probably due to a change in output queue(s) handling in .17

however, the ms latency is not that important; what is important is a proper output rate so the bot won't flood itself out

^CJ^ · Post by **^CJ^** » Sat Sep 10, 2005 1:51 pm

Thanks for the answer!
As the bots are basically used as channelprotection on quakenet, i thought it might be better to have a fast latency in order to react more quickly to overtake attempts. Considering that would you still recommend the slower latency? If not so, would it be a stable solution to use the new eggdrop version with the old server.so?

demond · Post by **demond** » Sat Sep 10, 2005 2:09 pm

if your channel is registered with some form of network-wide channel service, you don't bother with takeovers - you can always easily reverse one, should it ever happen

if your channel is not registered and your enemies manage to somehow gain ops, chances are they will take over the channel anyway, no matter which version eggdrop you use - there isn't a bot or/and script that guarantees your botnet will emerge victorious from the op/deop battle - and if you don't have at least 10-15 opped bots, forget about protection from hacked ops

De Kus · Post by **De Kus** » Tue Sep 13, 2005 4:06 pm

one note, the script doesn't use proper milliseconds, because uses
[clock clicks]
instead of
[clock clicks -m]
which is the proper way to get a milliseconds timestamp. Is it possible that while changing the eggdrop version also the system got some updates or it was compiled with another TCL version, so the clock command returned probably an other clicks type. ah dont forget to edit the expr line to compensade the change

.

just to show you that it actually makes a difference:

[20:55:41] tcl: evaluate (.tcl): clock clicks -m
Tcl: 1126641341021
[20:55:46] tcl: evaluate (.tcl): clock clicks
Tcl: 1126641346986467

as you can see, on the system that was captured we have 3 more digits, so clicks are microseconds instead of milliseconds ^-^.

oh, and one more note, for measering lag, you usually use the "PING" command with a parameter like:
putquick "PING $lag"
and later you can check the return parameter of ping (the server will exactly reply the same parameter) to ensure we are really use the requested ping reply and not one that might have been send by another event.

demond · Post by **demond** » Tue Sep 13, 2005 4:18 pm

milliseconds are almost meaningless on IRC, and it's amazing how many scripts follow BitchX's 133t approach for measuring lag