bug fix: After a JUPE is issued by euworld (mod.ccontrol),
gnuworld is not itself aware that the server is not linked
anymore. Added the PostEvent() part to src/server.cc to fix.
Hidden [Thu, 13 Aug 2009 20:32:25 +0000 (20:32 +0000)]
Fixed an old bug that made euworld crash when the LOGIN command was sent
with no parameters while the LOGIN command was logged in the
lastcom/dailylog.
Hidden [Thu, 6 Aug 2009 02:59:24 +0000 (02:59 +0000)]
Fixed the bug that added a space preceeding "Unknown" when euworld
doesnt receive a version reply from a server. That space was added
each time euworld was restarted.
Sending the RPINGS and TIMES with a different interval. Still sending
them every minute, but splitting them in 20 different batches. So
euweorld sending TIMES and RPINGS each 5 seconds now.
Fixed a bug that surferced due to the last bug fix (SF bug id:2824067) which caused X not to be
notified about channel events of newly registered channel.
Fixed a bug that caused X to be notified twice for channel events.
This leads to join count (for pending channel) to be increment twice whenever someone joins a channel.
- Added command RELOAD.
- Require people with access to be opered to use the service. Being logged in to the appropriate username is not sufficient anymore.
Hidden [Sun, 14 Jun 2009 01:29:54 +0000 (01:29 +0000)]
- Changes in the shells delshellco and delshellnb funtions: the fix looks better now.
- It is now a requirement for opers to specify a user@ to remove a gline. remgline <ip> with no *@ in front of it was causing euworld to modify the gline reason
Hidden [Sat, 13 Jun 2009 06:43:34 +0000 (06:43 +0000)]
- Fixed a bug in delShellco() and delShellnb(), which caused a crash on some machines, but not on others.
- Fixed a little thing in the ANNOUNCE command
- ADDCOMMAND / REMCOMMAND: added -allopers, -alladmins, -allsmts and -allcoders parameters. i.e: /msg euworld addcommand -alladmins <command> can be used to add <command> to all opers and admins.
Hidden [Tue, 9 Jun 2009 05:55:54 +0000 (05:55 +0000)]
- Once someone opers up globally from an ip, if that ip is glined using the gline, forcegline, forcechangline or schangline commands, euworld will ask for the gline to be sent a 2nd time.
- Wrote the ANNOUNCE command, level oper, which loads a fake client with the nick configured in the conf.
Hidden [Sun, 31 May 2009 21:31:55 +0000 (21:31 +0000)]
Glines by nick don't show the glined IP address anywhere in the lastcom logs, in the msgchanlogs and in the dailylogs emailed. It is now being showed in all of them.
Glined IP addresses don't appear anywhere when a channel gline is issued. They are only available in the SCANGLINE command, but they don't appear in the msgchanGlined IP addrend dailylogs either. Fixed that.
The last option can be turned on and off via the conf option showCGIpsInLogs.
Hidden [Thu, 28 May 2009 10:37:30 +0000 (10:37 +0000)]
Sometimes E sets a gline, and before the other servers disconnect the client in question, it has time to flood again and then E glines the same IP again. This code avoids that.
Hidden [Sat, 27 Dec 2008 23:34:30 +0000 (23:34 +0000)]
- Added the command "shells clearall"
- Fixed some shells stuff
- If there is an exception for an IP and an oper attempts to gline it using the GLINE, FORCEGLINE, SCHANGLINE, FORCECHANGLINE commands, euworld will ask the oper to send the gline command again. It will be set if the same gline is sent a 2nd time within 5 minutes.
Hidden [Mon, 17 Nov 2008 02:54:42 +0000 (02:54 +0000)]
- The join-flood is now reported (and flooding clients glined) even if the channel became empty at the time E reports it.
- An IP is caught join-flooding <jcMinJoinsPerIPToGline> times => gline even if it was not caught parting (can be different channels)
- <jcMinJoinToGlineJOnly> joins from one IP in a single join-flood => gline even if it was not caught parting
- E used to not gline at all clients that were authed to X. It now glines them if the ip is used too many times to join-flood.
- The number of clients online per IP is now stored in a map. Instead of counting how many clients match a gline before setting it, it already knows how many clients match the *@IP gline (saves resources)
- E won't set glines if a burst ended in the last 25 seconds.
Note: This version of mod.dronescan (E) has been running on undernet for over 2 months now. I should have committed before.
----------------
MrBean [Wed, 12 Nov 2008 20:45:42 +0000 (20:45 +0000)]
Log last used user ip (alongside last used host).
NOTE: add doc/update.last_seen.sql contents into cservice database
or complaints about missing column last_ip will be displayed!
Hidden [Wed, 6 Aug 2008 21:22:36 +0000 (21:22 +0000)]
mod.ccontrol:
Exceptions for anything bigger than a /8 won't be added using the SHELLS addnetblock command.
mod.dronescan:
Now counting all users from an IP in a map. Instead of iterating the Users list to see how many clients match an IP before adding a gline, it now checks the map. The floodbot glines procedure was very CPU expensive (100% on the production network) before this modification.
STATUSCommand.cc:
* Now showing how many glines are in the queue
dronescan.cc:
* Will not set glines if preloadExceptionalChannels() fails
* Will report the j/p flood and set glines even if the channel becomes empty
* Will not set glines for 15 seconds after a burst
* Will count the number of clients glined too, not only the number of ips
Compy [Wed, 16 Jan 2008 02:03:37 +0000 (02:03 +0000)]
Condensed DB handle usage to one global handle. This means that chanfix should work fine and efficiently without threads. ** NOTE: This version should be considered beta.