So the crash was caused by a real OOM condition, I was testing it in a VM with 512MB RAM. Changed to 4GB and now I have two netsniff-ng instances working and getting portion of the flows, hashed by the kernel, per flow :-)
I will work on the code and submit patches for review in a few days. On Tue, Apr 7, 2015 at 4:19 PM, Daniel Borkmann <[email protected]> wrote: > On 04/07/2015 04:06 PM, Daniel Borkmann wrote: >> >> On 04/04/2015 04:46 PM, Michał Purzyński wrote: >>> >>> I'm trying to implement the PACKET_FANOUT_HASH support. The idea is to >>> have multiple netsniff-ng processes running and each sharing a portion >>> of flows. The fanout hash mode would send all packets from the same >>> flow to the same socket and all processes would join the same socket >>> group. >>> >>> I tried to do it like this (once the code works I will refactor it, >>> it's an ugly hack now ;-) >>> >>> https://gist.github.com/mpurzynski/5be9f5693e66dbe16947 >>> >>> Single process netsniff-ng works with this code, when I start another >>> one the first crashes. >>> >>> What I'm doing wrong here? >> >> >> I'm delighted to see work in this direction! Awesome! >> >> I guess we would need to pass cluster_id from the command line, so >> that the user could configure this. >> >> What's the exact error message/crash? Are you talking about a kernel >> crash? If user land, could you strace -f that? > > > Btw, both cluster_mode and cluster_id should be 32 bit, but that doesn't > explain the crash of the other one. ;) > >> Thanks, >> Daniel -- You received this message because you are subscribed to the Google Groups "netsniff-ng" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. For more options, visit https://groups.google.com/d/optout.
