Bot fails to connect while reconnecting too fast [solved]

[De Kus]

First I will post some log messages and later I will explain a bit about it:

Code: Select all

Trying server irc.anoxs.net:6667
DNS resolved irc.anoxs.net to 202.3.110.144
[m->] NICK Phoenix
[m->] USER phoenix . . :channel eggdrop bot with the power to ban you!
net: connect! sock 12
Connected to irc.anoxs.net
-NOTICE- *** Looking up your hostname...
-NOTICE- *** Found your hostname (cached)
-NOTICE- *** Checking ident...
-NOTICE- *** Received identd response
[@] zeus.au.oc.swiftirc.net 001 Phoenix :Welcome to the SwiftIRC IRC Network Phoenix!De_Kus@venus.isix.net
Trying server irc.swiftirc.net:6667
DNS resolved irc.swiftirc.net to 85.8.136.114
[m->] NICK Phoenix
[m->] USER phoenix . . :channel eggdrop bot with the power to ban you!
net: connect! sock 12
Connected to irc.swiftirc.net
-NOTICE- *** Looking up your hostname...
-NOTICE- *** Found your hostname (cached)
-NOTICE- *** Checking ident...
-NOTICE- *** Received identd response
[@] manticore.uk.eu.SwiftIRC.net 001 Phoenix :Welcome to the SwiftIRC IRC Network Phoenix!De_Kus@venus.isix.net
Trying server irc.anoxs.net:6667
DNS resolved irc.anoxs.net to 202.3.110.144
[m->] NICK Phoenix
[m->] USER phoenix . . :channel eggdrop bot with the power to ban you!
net: connect! sock 12
Connected to irc.anoxs.net
...
[@]  ERROR :Closing Link: [87.117.192.75] (Throttled: Reconnecting too fast) -Email tim@swiftirc.net for more information.
-ERROR from server- Closing Link: [87.117.192.75] (Throttled: Reconnecting too fast) -Email tim@swiftirc.net for more information.
Disconnecting from server.

Okay a few facts:
- The paste has been cut a little removing duplicated lines due to the raw and normal logging. However it's correct to see the bot connecting to the next server in list (there are only 2) without any notice after the welcome message.
- Everything happens within a few seconds until the throttled error. When this error comes up it waits patiently the server-cycle-wait of 60sec. The server-timeout is set to 30sec.
- The bot runs (or rather tries) a static compilation of 1.6.18 using thread TCL 8.4.16 (Debian unstable).
- Before configuring and compiling I ran make distclean on the source.
- It ran fine using the same configuration on another server (which is down now).
- Maybe an important information is that the bot once joined the 2 channels it is supposed to join for a few seconds without me seen any raw logs about sending the JOIN command... this one was really like a ghost story.
- Others seem to be able to run eggdrops in general on the server
- the error persists since yesterday

So now you have a totally confused user of an eggdrop sitting before the telnet raw-log. Can you give me any valuable hints how to solve this or at least track down the source of the evil?

[nml375]

I don't suppose you could gather some insight using ethereal, tcpdump or other packet-sniffer?

Also, strace might be able to provide some insight in what is going on.

Oh, also, I assume you don't have any custom scripts loaded that might interfere?

[De Kus]

Well, I have running various scripts, but none should cause a thing "server hopping" or "log censorship". Also the same scripts ran on the old server.

PS: about the tcpdump or packet-sniffer, I'd first have to ask the owner about. Or can these be run on user level?

Edit:
I interviewed one of the other eggdrop users, addional facts came up:
- he also compiled 1.6.18, but not static
- since I happen to have also have a modular compilation of the same source, I switched using this one using symlinks which lead to the same unwelcome behavior
- even using his binaries didn't change the result
- the TCL his bot uses, is the same as the one mine uses (verified using ldd and .status). Also I couldn't find another TCL being installed on the server.

Edit2:
After realizing it MUST be something with the scripts loaded, I investigated a little and remembered I wrote a "dirty hack"-script for jump the server in case it entered the wrong network. However it seemed I never uploaded that script on the server, at least it came out I forgot the magic "!" in an if statement making the bot jump on the 001 message... CALL ME A FOOL !!!

[nml375]

One of the very few things that possibly comes to mind is that the kill_server function within the server-module is triggered upon eacy connect, tho I cannot start imagine why that would occur.

As for packet-sniffing, you would generally need root-privileges, unless your shell admin is nuts :p

Kinda sounds like you've ruled out all sources but your scripts. Tried removing them to see wether it works, then loading them one at a time until the error reoccurs?

[nml375]

Ahh, problem solved then