lagging bot

SaPrOuZy · Post by **SaPrOuZy** » Mon Sep 19, 2005 1:40 am

hello there,
is there a way for the bot to know if its lagging, i have a script for botnet
floods running perfectly, but sometimes if the bot is lagging it it messes up.
so i want to check if it is or not.
thanks.

demond · Post by **demond** » Mon Sep 19, 2005 1:57 am

wrong forum dude

it's normal for eggdrop to be lagging during mass flood; the important thing is to not flood itself out of irc thereby losing ops, which could still happen even on modern ircds that have msg pace limits

SaPrOuZy · Post by **SaPrOuZy** » Mon Sep 19, 2005 2:03 am

what i meant is the bot thinks there's a flood because it's lagging
i was checking the logs of a chan from 2 different bots,
the one lagging had like 3 minutes packed in 1 minute

so what i have in mind is to if the bots catches a flood, it will perform the
protection, and then checks if it's lagging , if it is it removes the lock/bans

still didn't think deep into it, but a script to know if the bot is lagging will be usefull

demond · Post by **demond** » Mon Sep 19, 2005 2:40 am

hmm that's actually nice idea, I'll think about incorporating something similar in xchannel.tcl

De Kus · Post by **De Kus** » Mon Sep 19, 2005 4:15 am

I wonder how to get the lag value from output in .status?

but what will you do if you notice the bot has lagged 60s and somone triggers a 10s flood? do you want to ignore the flood? it might have been really a flood, it might have not been. How will you determine?

PS: thats why I still believe in '/mode #chan +f 5:2' on networks that support it, the irc server knows usually better than the bot when 5 lines are beiing send within 2s

.

demond · Post by **demond** » Mon Sep 19, 2005 4:35 am

you get the lag not by parsing .status but by pinging the server and measuring its response time

and I will probably have periodic lag checks; on mass flood, I won't issue any server commands if the average lag from previous minute or so has been unacceptable

SaPrOuZy · Post by **SaPrOuZy** » Mon Sep 19, 2005 5:10 am

but what will you do if you notice the bot has lagged 60s and somone triggers a 10s flood? do you want to ignore the flood? it might have been really a flood, it might have not been. How will you determine?

i have more than 1 bot protecting the channel, so if one bot is lagging
i will rely on the others.

and I will probably have periodic lag checks; on mass flood, I won't issue any server commands if the average lag from previous minute or so has been unacceptable

that seems to be a good idea, to rely on an average lag time

SaPrOuZy · Post by **SaPrOuZy** » Wed Sep 21, 2005 8:34 am

Code: Select all

proc lag_check {} {
	global lagging
	set lag0 [ping server1]
	if {$lag0 > 3 } { 
		set lagging 1 
		return
	}
# server1 != server2 != server3
	set lag1 [ping server2]
	set lag2 [ping server3]
	if {$lag1 > 3 && lag2 > 3 } {
		set lagging 1
		return
	}
}

explanation of pseudo code: (EDIT*: it's not really pseudo code, it's more like code hehe)
server1 = the server to which the bot is connected.
if the ping reply is greater than 3 seconds then the bot is lagging due to it's connection or load on box or whatever.
in case it's not, we should also check if the IRC server is lagging (is about to split or something)
so we ping 2 other servers , server2 and server3
if the reply of both is greater than 3 seconds then server1 is lagging
and thus the bot is considered lagging and shouldn't issue commands

what do you guys think about this?

what's needed is:

a function that pings the server and returns the reply
to get the 2 other server names to ping.

i know it's should be in the scripting forum but it's about the pervious.

demond · Post by **demond** » Wed Sep 21, 2005 11:58 am

pinging other servers is pointless since if you are lagged to your own server, you'll be lagged to all others anyway - you have only one connection to IRC in any given moment, and that connection is through the server you are on

SaPrOuZy · Post by **SaPrOuZy** » Wed Sep 21, 2005 12:05 pm

if the ping reply is greater than 3 seconds then the bot is lagging due to it's connection or load on box or whatever.
in case it's not, we should also check if the IRC server is lagging (is about to split or something)

demond · Post by **demond** » Wed Sep 21, 2005 1:39 pm

hmm so you are proposing a complex lag tracking system but you are not elaborating on details

obviously, you'll need to know the structure of the network, i.e. which servers your server connects to (that can be obtained by analyzing LINKS output); and you will have to come up with some algorithm for taking the ultimate decision whether to take action or not; however, that won't be easy since there can be all kind of scenarios; consider for example:

you are on server A, which connects to B, C and D; you have pinged B, C and D and detected no lag; however, D connects to E, and D is lagging to E, and most of your users are either on E or on servers behind E; so E splits, rejoins, and you take action because your averaged lag measurement for the neighbouring servers didn't go above the trhreshold

SaPrOuZy · Post by **SaPrOuZy** » Wed Sep 21, 2005 2:45 pm

i thought about parsing LINKS(to get the list of servers) and WHOIS (to know where the bot is connected)

what am actually checking if the server that the bot IS connected to is lagging or not. not other servers
because if it is then it could cause probs.

and also is not about splits/rejoins
it's about to know forexample : if the bot noticed 6 joins happened in 3 seconds it could be due to 3 possible reasons:
1- the is actually a flood
2- the bot it self is lagging
3- the server on which the bot is connected is lagging and suddenly received the packets and displayed them.

1 and 2 are taken care of, 3 remain to be completed.

hopefuly this clears up my thought more

demond · Post by **demond** » Wed Sep 21, 2005 2:57 pm

SaPrOuZy wrote:i thought about parsing LINKS(to get the list of servers) and WHOIS (to know where the bot is connected)

there is global variable server, don't go with whois'ing self

what am actually checking if the server that the bot IS connected to is lagging or not. not other servers
because if it is then it could cause probs.

lagging to what? to the bot? then what does [ping server2] means in your pseudo-code?

and also is not about splits/rejoins
it's about to know forexample : if the bot noticed 6 joins happened in 3 seconds it could be due to 3 possible reasons:
1- the is actually a flood
2- the bot it self is lagging
3- the server on which the bot is connected is lagging and suddenly received the packets and displayed them.

1 and 2 are taken care of, 3 remain to be completed.

hopefuly this clears up my thought more

3 means pinging other servers

SaPrOuZy · Post by **SaPrOuZy** » Wed Sep 21, 2005 3:11 pm

a-

lagging to what? to the bot? then what does [ping server2] means in your pseudo-code?

i want to check if the bot is lagging to server2 (in case it's not lagging to server1, it's own server)
if the bot is not lagging to it's server
and the bot is lagging to server2 and server3
then this most porbably means that server1 is lagging to other servers.

b-

3 means pinging other servers

yes i want to ping other servers to accomplish a

anyways for now i'll settle for what's done and test it. the case of the lagging server can wait lol

demond · Post by **demond** » Wed Sep 21, 2005 3:25 pm

so we are back to my previous post about complex lag tracking system and zillions scenarios