Gateway falsely marked as down after restart. Works again after clicking "apply"

[Underpay6703]

I have a gateway to another router to use its ISP in a gateway group.
A problem I found is that when either of the routers is rebooted, the other router marks the gateway (between the routers) as down even after reboot is finished.

When I hit "apply" on the router's Gateway configuration page (without changing anything) the gateway is back online.

Is this a bug? Does anyone have time to reproduce this?

[djr92]

Have you tried using a monitor IP on the gateway? If you set a gateway to "monitor" the IP of the opposite gateway, does it still do this?

[tracerrx]

I'm also having this problem, but only in IPv6 gateways... and across multiple installs... The Ipv6 gateway for all my starlink connections has to be "edited" then applied before its considered up.  All installs have other default gateways and starlink IPv4 gateways that dont require this.

[Underpay6703]

Have you tried using a monitor IP on the gateway? If you set a gateway to "monitor" the IP of the opposite gateway, does it still do this?

I am using a monitor IP, but since this is a gateway to another router, I use an internet address (1.1.1.0) to verify that the gateway (internet access on the other router) works as intended.

If i understand you correctly, I now set the monitor ip to the address of the target gateway instead. This does come back online as expected after a reboot/poweroff on the other router.
But this will not tell me if the gateway (ISP of the other router) is online. So the failover group won't work. 


I am suspecting something is up with the routing table.
With the gateway down, the routing table also seems to mess with the monitor IP entry (1.1.1.0) that is supposed to statically use the gateway as next hop.
However even with the gateway reported down I can still reach 1.1.1.0 and the traceroute shows that it now uses the master router's ISP instead of the gateway I statically imposed in both the gateway monitor and Routes configuration.

[doktornotor]

These are DNS servers, not ping servers. I would suggest to not (ab)use public DNS for ping tests. They will eventually rate-limit or block pings when you do this.

[Underpay6703]

These are DNS servers, not ping servers. I would suggest to not (ab)use public DNS for ping tests. They will eventually rate-limit or block pings when you do this.

I've heard this said at times. If you could, is there an article/source where instances like google or cloudflare say that they don't want people to ping for uptime?
The most I could find describe it as "bad etiquette", never heard about them rate limiting or blocking public IPs.

I could change it to the Gateway Monitor IP I find on my router's automatic gateway for now. 

[doktornotor]

It's not just about bad etiquette. Google seriously DOES rate limit these pings. I have been debugging multiple boxes where using 8.8.8.8 for monitoring caused intermittent "gateway down" issues. And I am not alone.

I have also seen a local ISP with couple of thousands customers redirecting all DNS queries to Google -- because their servers broke down. Well, it did not work exactly well. https://developers.google.com/speed/public-dns/docs/isp

[Underpay6703]

It's not just about bad etiquette. Google seriously DOES rate limit these pings. I have been debugging multiple boxes where using 8.8.8.8 for monitoring caused intermittent "gateway down" issues. And I am not alone.

I have also seen a local ISP with couple of thousands customers redirecting all DNS queries to Google -- because their servers broke down. Well, it did not work exactly well. https://developers.google.com/speed/public-dns/docs/isp

Awesome, thank you for the links. Especially the one from Google.
Would you say that using the monitor IP of each of my routers' WAN gateway is a more reliable way to check uptime?
To illustrate that:
My master router has a WAN gateway to 'a' and also monitors it by pinging 'a'
My backup router has a WAN gateway to 'b' and also monitors it by pinging 'b'

If I want the Master router to have a gateway to backup router('s internet), I would then use 'b' as the monitoring IP and vice versa?
Or are there services more suitable for this in your experience?

[tracerrx]

@doktornotor I think its safe to use 8.8.8.8 to monitor your gateway... While not perfect, it's probably the best most of us can achieve.  Google understands they get lots of pings and mostly respond to ICMP way before it actually gets to their DNS servers.  If you have a PUBLIC alternative thats as reliable as google or cloudflare dns I suggest you post it.

This is also why you should make sure you miss more than X pings in a row before considering a gateway down...

My issue appears to be related to "dhcp6c_script: RENEW on igb3 executing"