There is nothing INHERENTLY wrong about filling the resolution queue.  If
you get a blast of new hosts, we only queue up the 1st 4K.  As they get
resolved and we see additional packets from the un-resolved hosts, they get
resolved in their turn.  ntop also fails' softly - it just keeps the IP
address w/o resolution for the displays.  That's all FAD (Functioning As
Designed).  

It's when the queue never goes down that you have problems.

Now that DNS cache is absurd - 316M - when each entry is about 75 bytes ... 

FWIW, in the future, you can leave prefsCache.db - that's your run-time
preferences (and ntop_pw.db).  All of the rest of the .db files will be
automatically recreated if necessary.

-----Burton

-----Original Message-----
From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of
Kenneth Porter
Sent: Thursday, March 10, 2005 4:06 PM
To: [email protected]
Subject: [Ntop] ntop-3.1 exits after addressQueue full

This is mostly an FYI. I'm guessing some kind of DB corruption bit me.

I was finding that my ntop-3.1 installation was starting up with 40% CPU
utilitization and then would quit after an hour of churning. The last thing
I see in /var/log/messages when this happens is:

Mar 10 05:17:16 matureasskickers ntop[17098]:   **WARNING** Address 
resolution queue is full [4096 slots]
Mar 10 05:17:16 matureasskickers ntop[17098]:   Addresses in excess won't 
be resolved - ntop continues

I don't see any shutdown messages after this.

I removed (moved to a temp directory) the db files in /usr/share/ntop,
restarted, and I now see more reasonable usage (about 1-2%) and it seems to
be humming along healthy again. These are the files I removed:

-rw-r--r--  1 root root    464965 Mar 10 07:59 addressQueue.db
-rw-r--r--  1 ntop ntop 316977235 Mar 10 07:59 dnsCache.db
-rw-r--r--  1 root root    237568 Mar 10 04:05 fingerprint.db
-rw-------  1 ntop ntop         6 Mar 10 04:06 ntop.pid
-rw-r--r--  1 ntop ntop     13474 Mar  1 08:57 prefsCache.db

Platform is Fedora Core 2 running on a dual Xeon with 2.6.10-1.9_FC2smp
(with HT, effectively a 4-way SMP system).
_______________________________________________
Ntop mailing list
[email protected]
http://listgateway.unipi.it/mailman/listinfo/ntop

_______________________________________________
Ntop mailing list
[email protected]
http://listgateway.unipi.it/mailman/listinfo/ntop

Reply via email to