I know there’s a good number of Starlink users in here so I thought this might be an interesting discussion. I have a second WAN connection so I do want Starlink to be in a failed status when it’s truly not working… but at the same time it can be really noisy.
For me, it doesn’t seem to matter which health check method I’m using (icmp, dns, or http) – but more so just tuning the timings. At these settings I’m seeing 10-15 failures per day:
Timeout: 5
Interval: 5
Retries: 5
Recovery retries: 3
I am also getting “No cable detected” a few times per day. I have not figured that one out.
I tried just about every combination of settings I could think of with Starlink on WAN1 and my fixed wireless ISP on WAN2, and never found any combination that was truly satisfying. Then I figured out that Starlink just drops - no signal at all - many times each day. It normally only lasts for a few seconds, but my Balance 30 LTE would often read it as “no cable detected”, or at other times as a signal loss. Often, Starlink would be back before the failover completed. It made calls on Teams or Zoom or via VOIP maddening, but the drops were almost always so short that it didn’t matter for video streaming (Netflix, etc.) or file transfers (e.g., Box). I don’t do online gaming so have no experience to offer there.
So… I tried SpeedFusion Cloud, effectively bonding Starlink and the fixed wireless ISP together (both set to Always On/Priority 1". Solved the problem. Now, those short drops just don’t matter because the connection doesn’t break. SpeedFusion Cloud (SFC) keeps it going, and the call doesn’t get interrupted. I have it set up on my “work” computer and my wife’s, for Teams and Zoom. After about four weeks, we’ve used less than 100GB of data, so this might be costing me about $2/week. The rest of our devices don’t go through SFC, as it’s not needed for them.
@Paul_Mossip I just gave that thread a read. Thanks for the link!
I too am using Speedfusion for my work computer. I really like it for critical traffic.
I agree on the short drops not affecting streaming… which is great.
I have a ton of outbound policies… which boil down to…
Streaming services prefer Starlink (to save LTE data)
I have Disney+, Youtube, Hulu, Prime, and Netflix all working successfully which required 15 rules
Facebook/Instagram prefer LTE (because they are awful over Starlink for whatever reason)
My catch-all rule is fastest response
I’m going to tweak my recovery to 1 per your recommendation and see if that improves anything.