We're running ntop.x86_64 3.3-1.el5.rf on top of CentOS 5 (Linux version 
2.6.18-53.1.14.el5 ([EMAIL PROTECTED]) (gcc version 4.1.2 
20070626 (Red Hat 4.1.2-14)) #1 SMP).

What we're seeing is that if we leave the "Global Traffic Statistics" 
(trafficStats.html) page open in a web browser (specifically Mozilla 
Firefox 2.x), with the META REFRESH set to 120 seconds (the default) - 
that the ntop server will eventually start throwing errno 10 and errno 
24.  The error will typically occur after 24-48 hours of refreshes.

Once the error occurs, ntop stops processing / recording data until it 
restarted.

Mar 30 06:50:39 fw1-pri ntop[7382]:   errno 10 during sending of page to 
web client
Mar 30 07:19:40 fw1-pri ntop[7382]:   errno 10 during sending of page to 
web client
Mar 30 07:20:58 fw1-pri ntop[7382]:   Unable to accept HTTP(S) request 
(errno=24: Too many open files)
Mar 30 07:21:15 fw1-pri ntop[7382]:   Unable to accept HTTP(S) request 
(errno=24: Too many open files)

We do not see this behavior with "Network Load Statistics" 
(thptStats.html), which is another page that I like to leave open in a 
browser window for days at a time.

# ulimit -a
core file size          (blocks, -c) 0
data seg size           (kbytes, -d) unlimited
max nice                        (-e) 0
file size               (blocks, -f) unlimited
pending signals                 (-i) 36864
max locked memory       (kbytes, -l) 32
max memory size         (kbytes, -m) unlimited
open files                      (-n) 1024
pipe size            (512 bytes, -p) 8
POSIX message queues     (bytes, -q) 819200
max rt priority                 (-r) 0
stack size              (kbytes, -s) 10240
cpu time               (seconds, -t) unlimited
max user processes              (-u) 36864
virtual memory          (kbytes, -v) unlimited
file locks                      (-x) unlimited

For the moment, we've backed off the auto-refresh to 600 seconds (from 
120 seconds), which is still only going to delay the inevitable issue. 
I'm currently running a test on another similar box to see whether I can 
reproduce the issue that the "Global Traffic Statistics" is making ntop 
run out of open files.

The other possibility is that prior to this week, we did not use a 
custom protocol.list file (/etc/ntop/protocol.list).  But we've 
installed the following file, in order to make the Global Traffic 
Statistics report work better for us:

HTTP=http|www|https|3128
FTP=ftp|ftp-data
DNS=name|domain
Mail=pop-2|pop-3|pop3|kpop|smtp|imap|imap2
DHCP-BOOTP=67-68
NNTP=nntp
SSH=22|880
Subversion=922
Jabber=5222-5223
PPTP=1723
NTP=ntp
Gnutella=6346|6347|6348
Kazaa=1214
WinMX=6699|7730
eDonkey=4661-4665
Messenger=1863|5000|5001|5190-5193

Which seems to work well and does what we want.  Except for the little 
issue of the ntop daemon running out of file handles and generating a 
lot of errno 10 messages for the past few days.

(The second box also sees errorno 10 issues... but only when left 
sitting on the Global Traffic Statistics page.)
_______________________________________________
Ntop mailing list
[email protected]
http://listgateway.unipi.it/mailman/listinfo/ntop

Reply via email to