On Solaris 2.x, why does cflowdmux report a shmget() error at startup (and then fail to work)?

Solaris' default shared memory segment size maximum is very low (1MB). cflowdmux wants a shared memory segment for packet buffers that's just over 1MB, which is beyond the default maximum limit imposed by the operating system. To permit a larger shared memory packet buffer, you should add this to /etc/system: set shmsys:shminfo_shmmax = 4194304 and reboot the host. The setting above is 4MB; a higher setting is acceptable as well (no resources are allocated by this change, since shared memory is allocated on demand). Note that the packet buffer size may be configured using the PKTBUFSIZE setting in the OPTIONS stanza in cflowd.conf.

Data Problems

What causes entries in the AS matrix to have the source or destination AS set to zero?

Flows for traffic destined for the router's local AS will have a destination AS of 0. Flows for traffic coming from the router's local AS will have a source AS of 0. Flows for traffic destined for the router will have a destination AS of 0 and a destination interface index of 0. Flows for unroutable traffic will have a destination AS of 0, a nexthop of 0.0.0.0, an output interface index of 0 and a destination netmask length of 0. Prefix cache misses when not running CEF. This can cause failed AS and netmask lookups on both the source and destination. Flows for this traffic will have a source AS of 0 and a source netmask length of 0 and/or a destination AS of 0 and destination netmask length of 0. In the case of a source AS of 0 and netmask length of 0, the typical cause is asymmetric traffic (the router is only seeing one direction of the flow because traffic in the other direction takes a different path which circumvents the router or the flow is unidrectional). Solution: use CEF. Flows for multicast traffic appear to always have both source and destination AS set to 0.