BUG!? SamKnows box flooding LAN with DHCPv6 solicit packets after 49 days of uptime

It started last Saturday, but I didn't get a good chance to investigate it until today, now that I'm back from a trip.

The symptom that led me to discover the problem was severe packet loss from my LAN to my WAN. It started with DNS lookups failing. Then, incoming mail connections started to be dropped randomly. The web site started failing to load. A closer look at my managed switches showed high packet activity/loss at the switch on my router.

A power cycle of the SamKnows box resolved the problem.

Not being able to access the SamKnows box to remotely reboot is a big problem that I didn't realize until now. If this kind of disruptive behavior occurs again I am going to have to send it back.

The easiest way to spot the problem is to inspect your router's WAN interface. You may see hundreds of these per second:

13:38:47.947656 IP6 (hlim 1, next-header UDP (17) payload length: 100) fe80::fa1a:67ff:feXX:XXXX.546 > ff02::1:2.547: [udp sum ok] dhcp6 solicit (xid=182693 (elapsed-time 65535) (option-request SIP-servers-domain SIP-servers-address DNS-server DNS-search-list opt_56 opt_64 opt_67) (client-ID hwaddr type 1 f81a67XXXXXX) (reconfigure-accept) (Client-FQDN) (IA_NA IAID:1 T1:0 T2:0) (IA_PD IAID:1 T1:0 T2:0))

Why does the SamKnows box send DHCPv6 solicit packets to the LAN!? Why did this trigger this past Saturday? *ugh*

To mitigate this problem, I am probably going to add a weekly reboot of the SamKnows box to my todo list.

The SamKnows box had been powered on and continuously connected since May 3rd. Interestingly enough, June 21st, 2014 is 49 days after installation.  This sounds a lot like the classic '49.7 day' timer bug, where a 32-bit per-millisecond counter wraps around to zero! I suspect there is a bug in the SamKnows DHCPv6 client related to timers.