See in-line > -----Original Message----- > From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] Behalf Of > Julien TOUCHE > Sent: Tuesday, December 23, 2003 3:51 AM > To: [EMAIL PROTECTED] > Subject: Re: [Ntop] Unknown DLT types > > > Burton M. Strauss III wrote: > > > The MTU isn't strange - that's the default for unknown devices. > Since all > > we use it for is the buffer size ntop allocates, there's no > major harm in > > being big. > > > > What we need to do to fix the warnings is to add the > appropriate entries to > > the DLT table... > > > > Let's see ... they're all 12s... that's > > > > #define DLT_RAW 12 /* raw IP */ > > > > Note that there are some ugly comments in Linux's > /usr/include/net/bpf.h re > > this... > > > > /* > > * OpenBSD DLT_LOOP, for loopback devices; it's like DLT_NULL, except > > * that the AF_ type in the link-layer header is in network byte order. > > * > > * OpenBSD defines it as 12, but that collides with DLT_RAW, so we > > * define it as 108 here. If OpenBSD picks up this file, it should > > * define DLT_LOOP as 12 in its version, as per the comment above - > > * and should not use 108 as a DLT_ value. > > */ > > #define DLT_LOOP 108 > > > > which might cause problems down the road. > for my tun test, i'm on linux > on openbsd, DLT_RAW is 14 for information
The actual underlying numerical value SHOULD not matter since we use the DLT_ constants. That is we set the raw entry to 1500/0 and whether it's 12 or 14 doesn't matter as long as it's consistent for the os. However, file it away, that we MAY have to do some table tweaking a bit down the road, if we get that far w/ OpenBSD. > > > which should probably become this: > > > > _mtuSize[DLT_RAW] = 1500 > /* raw IP */ > > _headerSize[DLT_RAW] = 0; > > > > Try it and let me know. > > > i changed just this in globals-core.c > > [snip] > 23/Dec/2003 10:42:43 Checking eth1:0 > 23/Dec/2003 10:42:43 Resetting traffic statistics for device eth1 > 23/Dec/2003 10:42:43 DLT: Device 0 [eth1] is 1, mtu 1514, header 14 > 23/Dec/2003 10:42:43 Checking requested device 'sit1' > 23/Dec/2003 10:42:43 Adding network device sit1 > 23/Dec/2003 10:42:43 Interface 'sit1' (netmask 255.255.255.255) > computed network size is 1024 hosts > 23/Dec/2003 10:42:43 MEMORY: ipTrafficMatrix base (no TrafficEntry) for > interface 'sit1' is 4.05MB > 23/Dec/2003 10:42:43 Checking sit1 for additional devices > 23/Dec/2003 10:42:43 Checking sit1:0 > 23/Dec/2003 10:42:43 Resetting traffic statistics for device sit1 > 23/Dec/2003 10:42:43 DLT: Device 1 [sit1] is 12, mtu 1500, header 0 <snip /> > reports seems normal on web interface. > > is there something precise i could check ? It does look perfectly normal, and there's really not much to look for since ntop doesn't do much probing into packets beyond the first few dozen bytes... > > Regards > > > Julien > > note: it may be useful to have an option to ignore interface which are > not up (for example vpn tunnel down). Not sure how you tell this - if the OS reports that the interface exists, ntop adds it. And libpcap would see packets on it. Maybe I don't understand what you mean - post the ifconfig output (start a new thread) and I'll look at it. -----Burton _______________________________________________ Ntop mailing list [EMAIL PROTECTED] http://listgateway.unipi.it/mailman/listinfo/ntop
