After several weeks without an issue, this problem seems to have reappeared.
I've tried just about every relevant wireless setting. Now, I'm back to mixed 802.11n/g, manually selected an unused channel (1), and channel width at auto 20/40 (tried just 20). I got rid of auto broadcast on the DHCP server, traffic shaping re-enabled (but manually set uplink speed), re-enabled QoS but got rid of dynamic fragmentation, WMM enable on, short GI back on (was off with no difference), WISH and WPS off, UPnP disabled, and got rid of enable multicast streams.
Again, it's only a problem on wireless. I can plug into the router hardwired when this happens, and it's just fine even without a reboot. Interestingly, even when on wireless, I can generally get to the router's homepage without a problem, it just seems to not give me WAN access.
I dumped my router logs to email and noticed a few interesting entries. Can someone help me interpret them?
1. [DATE / TIME / ROUTER NAME] user.warn kernel: [172266.950000] SHAN=== Chaning for PS state change for type 0 subtype 208
2. [DATE / TIME / ROUTER NAME] user.warn kernel: [172138.500000] ->bad
3. [DATE / TIME / ROUTER NAME] user.debug syslog: Debu: Route activate request [WAN ISP IP address] to 18.104.22.168 is from myself, Ignoring it !
4. [DATE / TIME / ROUTER NAME] user.debug syslog: Debu: Joining group 22.214.171.124 upstream on IF address [WAN ISP IP address]
5. [DATE / TIME / ROUTER NAME] user.warn kernel: [ 66.480000] mroute: pending queue full, dropping entries.
6. [DATE / TIME / ROUTER NAME] user.warn kernel: [11254.160000] rx_clear=99, rx_frame=99, tx_frame=0
7. [DATE / TIME / ROUTER NAME] user.warn kernel: [11254.160000] Busy environment detected
I started a ping test and tried to correlate when my Internet connection slowed / dropped. It appears the relevant log entry when that happens is #1.
Any thoughts on the log or what else could be going on? The router has already been RMA'd once, so I'm trying to look for another cause...