Im running sentinel 2.7 standalone
and the problem i have is it is run on bots which are on a large channel averaging around 300 - 400 users.
Now the default setting keep thinking there is a flood join every 5 seconds when ppl leave join the channelso i modified them to not do it as much. It has cut down but still does it occasionaly.if i make the settings any higer they keep mode locking and banning users who arnt flooding but joining at the same time
The only problem is when we get a large proxy flood / tsunami flood(which is very very fast) the bots dont seem to respond until the flood is almost finished. also they dont seem to kickban half the flooders as i see them leaving of their own accord after a few minutes . The channel is +miRck at this point so there is lil they can do.. but having the bots able to handle the floods better would solve a great headache
I am pasting the settings i have below
# Bot CTCP flood.
set sl_bcflood 5:30
# Channel CTCP flood.
set sl_ccflood 5:20
# Channel avalanche/tsunami flood.
set sl_avflood 4:20
# Channel text flood.
set sl_txflood 40:30
# Channel bogus username join flood.
set sl_boflood 4:20
The seconds field (i.e. 3 in 4:3) in your join-part and nick flood settings is far too low and against what is recommended in the documentation. It is likely to cause detection and banning problems. Instead, set it to 20 and fiddle with the number field instead (e.g. 15:20 or 20:20).
Bots on very big channels are likely to experience bursts of data from the server making the bot think there's a flood, but there's not much you can do about that. A feature in the next version of sentinel.tcl is likely to reduce detection of those cases.