Add fine-grained command flood controls

This reintrouces "Excess Flood" quits for those that prefer it to fakelag,
and allows the maximum command rate to be set in the connect block.

git-svn-id: http://svn.inspircd.org/repository/trunk/inspircd@12093 e03df62e-2008-0410-955e-edbf42e46eb7

1 files changed, 22 insertions, 4 deletions

diff --git a/conf/inspircd.conf.example b/conf/inspircd.conf.example
index 0b4099a40..baadf24d6 100644
--- a/conf/inspircd.conf.example
+++ b/conf/inspircd.conf.example
@@ -307,17 +307,35 @@
          hardsendq="1048576"
 
          # softsendq: amount of data in a client's send queue before the server
-         # begins delaying their commands
+         # begins delaying their commands in order to allow the sendq to drain
          softsendq="8192"
 
          # recvq: amount of data allowed in a client's queue before they are dropped.
          recvq="8192"
 
-         # threshold: This specifies the seconds worth of penalty a user is allowed to have
-         # before fake lag is applied to them.  If this value is set too low, every action will cause throttling.
-         # Set to 0 to disable.
+         # threshold: This specifies the amount of command penalty a user is allowed to have
+         # before being quit or fakelagged due to flood. Normal commands have a penalty of 1,
+         # ones such as /OPER have penalties up to 10.
+         #
+         # If you are not using fakelag, this should be at least 20 to avoid excess flood kills
+         # from processing some commands.
          threshold="10"
 
+         # commandrate: This specifies the maximum rate that commands can be processed.
+         # If commands are sent more rapidly, the user's penalty will increase and they will
+         # either be fakelagged or killed when they reach the threshold
+         #
+         # Units are millicommands per second, so 1000 means one line per second.
+         commandrate="1000"
+
+         # fakelag: Use fakelag instead of killing users for excessive flood
+         #
+         # Fake lag stops command processing for a user when a flood is detected rather than
+         # immediately killing them; their commands are held in the recvq and processed later
+         # as the user's command penalty drops. Note that if this is enabled, flooders will
+         # quit with "RecvQ exceeded" rather than "Excess Flood".
+         fakelag="on"
+
          # localmax: Maximum local connections per IP.
          localmax="3"