It's an edge case to me - I don't have a pure windows network to test on - everything uses tcp/ip.
There's nothing odd about being in any particular state for any amount of time: Don't think of it has "hung". It's perfectly normal, based on what ntop sees. Remember, it's unlikely that ntop will be started just as everything on the network identifies itself. Rather it's going to learn about the more active hosts first and the less active only when they send periodic announcements or respond to device polls like some router/gateways perform. I adjusted the debug message yesterday so only actual transitions are reported, i.e. you won't see the 6->3 stuff because it never happens. I'm assuming you didn't update your code after that commit?? The time stamp on the ntop-dev message is Mon 2/23/2004 4:51 PM (US Central Time Zone). Can you check the generated HTML and see what's in the cell. With that commit, there should now be <!-- --> (HTML comments) showing how ntop populated the cell. Thanks! -----Burton > -----Original Message----- > From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] Behalf Of > Markus Rehbach > Sent: Tuesday, February 24, 2004 1:24 AM > To: [EMAIL PROTECTED] > Subject: Re: [Ntop] New sorting routines in cvs - these are post 3.0pre1 > - TEST NOW OR HOLD YOUR WATER > > > Hi again, > > I don't think is is an edge case, because many Windows PC will send > those host announcement broadcasts. Here a short extract of the debug log: > > Mon Feb 23 21:59:36 2004 DEBUG_CMPFCTN: > lookupHost(192.168.10.11, 00:40:45:21:B3:5A, m0, f0, dev0) > Mon Feb 23 21:59:36 2004 DEBUG_CMPFCTN: > setResolvedName(0x0831d920) 0 -> 6 192.168.10.11 - hash.c(1140) > Mon Feb 23 21:59:36 2004 DEBUG_CMPFCTN: > setResolvedName(0x0831d920) 6 -> 3 ETHER - pbuf.c(663) > Mon Feb 23 21:59:36 2004 DEBUG_CMPFCTN: > setResolvedName(0x0831d920) 6 -> 3 ETHER - util.c(3065) > > Another host 'hung' in state 6 for a long time and was visible in > the tables as <IP-Address>. > Having sent the SMB BROWSER host announcement resulted in another > empty table entry. > > Hope it is helpful > > Markus > > P.S.: 2nd try, don't know why.... > > ________________________ > > > On Monday 23 February 2004 21:24, Burton M. Strauss III wrote: > > Need to figure out what those entries are, so we can adjust the > edge case. > > > > add > > > > #define DEBUG_CMPFCTN > > > > in globals-defines.h or ntop.h > > > > Then > > > > 1) check the log for the DEBUG_CMPFCTN: messages > > - DEBUG_CMPFCTN: lookupHost shows calls to lookupHost() > > - DEBUG_CMPFCTN: setResolvedName shows the pattern of name resolution > > > > (note, there's another message in util.c around 5816, that shows the > > compares - you might want to delete those lines > > to keep the noise down). > > > > > > 2) look at the generated html code - there may be some <!-- --> > comments (I > > probably need to add more of these) > > > > -----Burton > > > > > -----Original Message----- > > > From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] Behalf Of > > > Markus Rehbach > > > Sent: Monday, February 23, 2004 11:24 AM > > > To: [EMAIL PROTECTED] > > > Subject: Re: [Ntop] New sorting routines in cvs - these are > post 3.0pre1 > > > - TEST NOW OR HOLD YOUR WATER > > > > > > > > > Hi Burton, > > > > > > all Systems in the network which are broadcasting names via the > > > SMB BROWSER > > > Protocol (without having a reverse lookup entry in dns) were > > > shown in 3pre1 > > > by name. But now there is an blank entry in the list, no name nor the > > > IP-Address. > > > > > > Tried it using the -n Parameter. My 'server' is running samba and is > > > broadcasting its name. > > > > > > Cheers > > > > > > Markus > > _______________________________________________ > Ntop mailing list > [EMAIL PROTECTED] > http://listgateway.unipi.it/mailman/listinfo/ntop > _______________________________________________ Ntop mailing list [EMAIL PROTECTED] http://listgateway.unipi.it/mailman/listinfo/ntop
