I was running V2.5.1.16 and have been for 6+ months no issue. The other night the internet went down and I logged into the Orbi via web and verified the provider issue. While there I noticed the upgrade to 2.6.2.104. Since nothing was working (internet down) I figured it was a good time to upgrade and via the web upgraded the router and 3 sats. All came back fine on the new code and later on internet was fine when the provider came back.
Within 24hrs the machines on DHCP stopped having internet access. Checking things out I found internet was fine but DNS was failing. Requests to router IP just timing out. Rebooted and fine. Within 24hrs it happened again. Rebooted, searched for issue, enabled telnet to router and started poking around. Within 24hrs it happened again and this time just restarted dnsmasq via init,d. I grepped through /var/log but I don't see anything there related to DNS. init.d doesn't have any status commands for the service and no systemctl so I can't status the service but the pid is there. Next time it happened I just reloaded the config and that fixed the issue (init,d option to 'reload' instead of 'restart' which just re-reads the config). It happened the 5th time in almost as many days and I just downgraded to the old code (2.5.1.16) on everything. So far everything is fine. If someone wants to debug the issue I'd be happen to collect data as time permits. As a work around while on the new version I can screenscrape to ensure telnet is enabled and run a script to telnet to the router and reload the service as necessary to prevent any outages longer than perhaps 5 minutes.
Anyway, something definitely broken with dnsmasq. I'm sure it passed the mustard to release the code but I've lucked out and have the 'special' setup to cause it to break. It would be nice if:
1) it would check itself, simple task as mentioned
2) NG allow you to set your own DNS servers and only force dnsmasq when secure functions are enabled that require filtering.
