-t 4 causes a lot more messages, which makes it more likely to have multiple
threads syslog()ing simultaneously which is what the fix is for.  (RH7.2 is
definitely pre NPTL and thus will be running an older glibc which isn't
thread safe in syslog.)

So:

1) Make sure that the -DFORPRENPTL really took - check version.c or in the
About | Configuration page.

Look for: Compile Time: ./configure and then the "compiler (CFLAGS)".

2) If it hangs up, be sure and walk through EACH thread and do the info
stack, e.g.

(gdb) thread 1
(gdb) info stack
(gdb) thread 2
(gdb) info stack

Etc.

That may be the only way to see what was going on when it deadlocks.

-----Burton
 

-----Original Message-----
From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of Rik
Schneider
Sent: Wednesday, March 30, 2005 2:53 PM
To: [email protected]
Subject: RE: [Ntop] Ntop 3.1.1 or cvs on RHEL2.1

It is recognized as redhat2.1.  RHEL2.1 is based on RH 7.2 and uses
pthreads.  Using a source tree from cvs (CVS checkout at about 2245Z Mar 29,
2005) I created configureextra/LINUXredhat2.1 based on the contents of
configureextra/LINUXredhat8.0.

--- begin configureextra/LINUXredhat2.1 --- #!/bin/sh

echo "        Setting RHEL 2.1 specific flag values"

CPPFLAGS="${CPPFLAGS} -DFORPRENPTL"

--- end configureextra/LINUXredhat2.1 ---

After building it I experimented with the -t option.  If I give a value
> 3 the web service does not work.  I seem to have the same problem if I
start the SSL server. 

If I use "--no-fc -u ntop -d -P /var/ntop -i eth1 -m 172.18.32.0/21 -w3000
-W 0 --disable-schedyield -t3" everything works fine for a while.
I have it running now and will do a gdb traceback the next time I get it
hung up.


--
Rik Schneider

Weathernews Americas Inc. 
(405) 310-2840 - Office
(405) 310-2900 - Main
(405) 388-1318 - Mobile
[EMAIL PROTECTED]
 No trees were harmed during the creation of this message; however some
electrons were inconvenienced.


-----Original Message-----
From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of
Burton Strauss
Sent: Monday, March 28, 2005 6:15 AM
To: [email protected]
Subject: RE: [Ntop] Ntop 3.1 or cvs on RHEL2.1

2.1 is probably pre NPTL and so you need the work-around for the
syslog()
hang.

Figure out what the configureextra file should be and copy the lines from
LINUXredhat8.0 (I've posted instructions on this within the last 2 weeks).

-----Burton
 

-----Original Message-----
From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of Rik
Schneider
Sent: Friday, March 25, 2005 3:49 PM
To: [email protected]
Subject: [Ntop] Ntop 3.1 or cvs on RHEL2.1

I don't know if anyone has succeeded in this, I haven't.

Here is what I have tried so far.

Building from source using Red Hat's libpcap fails during configure
with:
*******************************************************************
*
* ERROR: pcap header, bpf or library routines are missing
*           (yes means it was found, no means it was not found)
*
*              pcap.h...yes
*              net/bpf.h...yes
*              pcap-bpf.h...no
*              pcap_open_live() in -lpcap...yes
*              pcap_findalldevs() in -lpcap...no
*
*  (One of the bpf files is required, but either will do)
*
*>>> No way to proceed.
*
*???     1. Install libpcap (http://www.tcpdump.org/)
*???    and Rerun ./configure
*???  or 2. Use the --with-pcap-root= option
*
*******************************************************************
I grabbed and installed a newer libpcap and tcpdump.   I built them as
RPMs.  tcpdump sees the interface fine and gives lots of data.
I them built ntop-3.1 against the new libpcap library and install it.  I
then edit the ntop.conf file setup the password and then start ntop.  It
starts up several threads.  

So far, so good.  

When I attempt to open the web interface at port 3000 it hung my browser
(Firefox on RHEL4).  I stopped all the treads and restarted everything by
hand under strace.

strace -ff -x -s 80 -o /tmp/ntop.strace /usr/bin/ntop -u ntop -d -P
/var/ntop -i eth1 -m 172.18.32.0/21\
-t7 -w3000 -W 0 --disable-schedyield  --use-syslog=local6

The web thread gets hung at rt_sigsuspend.  The log entries for this thread
are:

14:40:22 mon2 ntop[1462]:  [MSGID0316203] [webInterface:8869] Note:
SIGPIPE handler set (ignore)
14:40:22 mon2 ntop[1462]:  [MSGID0986275] [webInterface:8890] WEB:
ntop's web server is now processing requests) for web server

The only thread that seems to increase in size appears to be a polling
thread.

I built the dev version from CVS with the same results.  


Am I running into a known issue or...
_______________________________________________
Ntop mailing list
[email protected]
http://listgateway.unipi.it/mailman/listinfo/ntop

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