Since the 2026.10.21.2302 update, Remote Access has been failing on and off. The DVR's diagnostics say ".u.channelsdvr.net is pointed at (blank) but the current IP for this system is …".
I've been logging a DNS lookup for my DVR's hostname every 10 seconds, through three resolvers: my local resolver, my router and 1.1.1.1. The router's NAT-PMP mapping for port 8089 is checked on the same cycle.
- DNS: the name returns NXDOMAIN for 10–17 minutes about every 3 hours, with 3–6 minute bursts about an hour earlier. From Oct 23 21:00 to Oct 25 08:20 (US Eastern), the windows began at about 21:00, 00:00, 03:00, 06:00, 09:00, 12:00, 15:00, 17:00, 20:00, 23:00, 02:00, 05:00 and 08:00, each at nearly the same minute past the hour.
- Where it fails: all three resolvers fail together, and I get the same NXDOMAIN querying ipdns1.channelsdvr.net and ipdns2.channelsdvr.net directly, so the record is missing on the authoritative side.
- The port mapping is fine: NAT-PMP renews hourly with no gaps. The mapping was missing once, for 10 seconds, right after I opened the DVR diagnostics, and the DVR re-mapped itself.
- A refresh helps a little: calling GET http://:8089/auth/refresh during an outage was followed by recovery within 3–6 minutes in 4 of 4 cases, against 12–17 minutes without it. The sample is small.
Questions: is the DVR expected to re-register its record every 3 hours, and did that change in the Oct 21 release? Can the record be kept in place between re-registrations?
Setup: Docker on Linux, 3 HDHomeRun tuners, Fios router with NAT-PMP, WAN IP in 100.x (CGNAT).