I set up some scripts to monitor what's up with the connection, and it happened to drop just now.
Downstream remained in-tact (as mentioned before, things like VoIP that aren't using TCP/requiring ACKs keep piping data, and I keep receiving it), but upstream crapped out.
Upstream frequency reported by the modem went from "23800000 Hz Ranged" to "23800000 Hz In Progress", then back to "23800000 Hz Ranged" when the connection came back up. And as I've observed before, upstream power level shot up while downstream remained relatively unaffected (it jumped around a little, but nothing too drastic).
Do these graphs help tell what's going on? I've checked all the connections and crimps, as suggested, and they're all good. Adding some attenuation before the modem to bump upstream power level some has helped stabilize the connection a LOT (it's much more stable now than it was before, and my speeds are better), the upstream drops are much more rare now. My moderately-educated guess tells me to suspect the modem (as any higher upstream power level is bled off by the attenuation I added, so if it were a noise issue, I'd think adding attenuation local to the modem wouldn't help), but I'm not totally sure.
»
dl.dropbox.com/u/1101592 ··· dex.htmlIf it helps, here are the logs surrounding the drop in the graph:
2012-08-27 22:33:00 3-Critical D003.0 DHCP WARNING - Non-critical field invalid in response.
2012-08-27 22:32:51 3-Critical R002.0 No Ranging Response received - T3 time-out
2012-08-27 22:32:45 3-Critical R004.0 Received Response to Broadcast Maintenance Request, But no Unicast Maintenance opportunities received - T4 timeout
2012-08-27 22:32:12 3-Critical R005.0 Started Unicast Maintenance Ranging - No Response received - T3 time-out