25.07 RC - no default gateway being set if default route is set to a gateway group and the Tier 1 member interface is down

dennypage

Effectively, @luckman212’s request is for a static route that only applies to IGMP echos originating from the firewall itself.

marcosm

@dennypage FWIW that doesn't happen currently even with pf. The route-to rule is based on the interface's source address with any destination that's not in the interface's subnet. Still, a rule can be created that applies to the correct traffic.

Given the feedback, it sounds like the issue isn't that a route should not exist, but rather some route is needed to allow pf to force the traffic. That's effectively the workaround @stephenw10 showed. Any potential undesired behavior from that kind of solution needs to be considered.

dennypage

@stephenw10 said in 25.07 RC - no default gateway being set if default route is set to a gateway group and the Tier 1 member interface is down:

Mmm, a good solution here would be some anycast ping targets that aren't DNS servers. But using DNS servers there is really convenient!

Convenient yes, but from time to time, Google and others get annoyed with everyone using their DNS servers as monitor targets and put temporary blocks in place. I generally recommend people to use regional routers in their ISP instead.

luckman212

@dennypage Exactly! I had written a script called hopfinder that I mentioned farther up, which already does this successfully & automatically for the FIOS connection, where traceroute works properly. On the LTE network, no such luck so I've resorted to querying the RDAP database (which has a nice parseable JSON output) for /32 hosts in T-mobile's network, and then iterating over a handful of them to find a few with the lowest latency. "it works" but the script takes about 45 seconds from start to finish, so not something to run every day, but once a week seems about right.

I'm planning to publish the updated script soon, trying to decide if it's worth making into a full package with a GUI.

SteveITS

Would a workaround for the fees be to block from LAN to 8.8.8.8 with a policy routing rule? Or would the static route override that? (haven't looked, just brainstorming)

FWIW since it was mentioned above, pfBlocker can block DoT, which it has tucked under "DNSBL SafeSearch." Though as I've mentioned elsewhere I know that at least the Dish DVR video on demand "app" (though not the DVR software) is hardcoded to use Google DoT, I think it was.

Patch

@marcosm said in 25.07 RC - no default gateway being set if default route is set to a gateway group and the Tier 1 member interface is down:

Let's keep in focus the following: what exactly is the problem that needs to be solved

Probably my understanding of how the system works but from my perspective it’s the required choices for monitoring

the monitoring address must be very reliable as a failure at that site will result in a failure of the monitored interface.so probably needs to be a site with extensive redundancy and probably with more than one option as even major sites like Microsoft go down and assuming otherwise results on extensive secondary failure. Using ISP routing hardware can also be problematic as when this hardware is heavily loaded pings maybe ignored which would result in secondary failure of the monitored interface.
The current approach requires a static route resulting in all traffic to the monitoring site to always use that specific interface. For a major site doing so over rides system interface loading objectives. Which suggests a monitoring site should be almost never used by real users traffic so some back water site.
The above requirements result in diametrically opposite choices. This raises the question of why this choice is actually required. Normal load balancing requires dynamically allocating traffic to one of several interfaces depending on current loading. Surely interface selection for interface monitoring should operate at this level.
imo ideally users would specify a pool of monitoring addresses. For general internet interfaces the monitoring pool for all interfaces would probably be the same however when monitoring a particular interface, that particular interface would be used.

luckman212

@Patch Nice summary, it conjures up memories of this 8-year-old idea (and bounty) of mine: dpinger multiple targets - aka gwmond

dennypage

@Patch said in 25.07 RC - no default gateway being set if default route is set to a gateway group and the Tier 1 member interface is down:

ideally users would specify a pool of monitoring addresses.

The idea of monitoring multiple addresses has been discussed at length previously.

luckman212

Yes, there's been much discussion about this, and for many many years. That it keeps coming up is a testament to the fact that for many people, a more robust solution is warranted.

In the redmine you linked, the final comment (from @jimp himself) sums it up nicely:

dpinger is only a daemon that pings and reports responses. It doesn't make decisions about what is good or bad for a pfSense gateway as a whole only its specific single target. It isn't up to dpinger to handle multiple targets or different protocols.

What is needed is more like some middleware-ish daemon to sit between pfSense and other gateway monitoring daemons like dpinger (cough cough, gwmond) that would be capable of coordinating multiple monitoring techniques for each gateway and making more informed decisions about their status.

Given the responses on the dpinger github it appears its author agrees that it's out of scope for dpinger itself.

I agree with Jim (and you @dennypage) that dpinger already does its job well, and should stay focused and simple. I do think pfSense needs that yet-to-be-coded "middleware" which could do a better job of orchestrating multiple dpinger instances + possibly other check methods such as curl/wget fetches to test under conditions where ICMP isn't good enough to rule out false positives/negatives.

w0w

@stephenw10 said in 25.07 RC - no default gateway being set if default route is set to a gateway group and the Tier 1 member interface is down:

Add a 3rd dummy gateway that always remains up to provide a default route. Add that to the failover group as some high tier.

Maybe I’m doing something wrong, but when I create a dummy interface, set it to the lowest priority (ex, Tier 3—we don’t really use it as a gateway, right?), and then configure the other two gateways with the “Do not create static routes” option enabled, after a reboot I get the LANGW status “pending” and no default route. So this needs to be another option activated on dummy, "Disable Gateway Monitoring Action"?

stephenw10

If it's showing as pending that implies the gateway is not available yet which should never be true for a local interface/IP address. You set something that actually exists I assume?

w0w

@stephenw10
Yes, it is LAN interface. And it was working fine untill I rebooted pfSense.

stephenw10

Hmm, I didn't have that issue in testing. Is the interface linked/UP? I have seen problems with that in the past but my test interface was not.

w0w

@stephenw10
I think "Disable Gateway Monitoring Action" did the trick. It seems to be working as it should now.
Yes, interface is definitely UP, because I am using it to reach the GUI.

stephenw10

Hmm, that's odd. I would expect that always be available.

w0w

@stephenw10
Yep, there’s definitely something going on. I’ve run into this before several times, but unfortunately I don’t remember all the details and the logs are already gone. All I remember is that at first it wasn’t related to “Do not create static routes” in my case—but my memory has failed me more than once.

luckman212

@stephenw10 / @marcosm any chance we can relocate this busy/lively thread to the regular Routing and Multi WAN section? It seems it isn't and probably never was specific to 25.07 RC anymore...