net/haproxy: multiple map file rules in a single frontend are not working

[viragomann]

Important notices

Before you add a new report, we ask you kindly to acknowledge the following:

• I have read the contributing guide lines at https://github.com/opnsense/core/blob/master/CONTRIBUTING.md
• I am convinced that my issue is new after having checked both open and closed issues at https://github.com/opnsense/core/issues?q=is%3Aissue

Describe the bug
If you create multiple map files of type domain, create rules for each using it, and assign them to a single frontend, only the first one is working, even all domain names are unique.

A use_backend ACL for each is written in the config file one after another. E.g.
use_backend %[req.hdr(host),lower,map_dom(/tmp/haproxy/mapfiles/674879807b26b5.22516524.txt)]
use_backend %[req.hdr(host),lower,map_dom(/tmp/haproxy/mapfiles/67472b11b0a4a9.24401988.txt)]
However, this doesn't work. For domains matching the second or further, you get 503.

To Reproduce

Steps to reproduce the behavior:

1. Create two map files for domain names with a structure "domain backendpool"
2. Create a rule for each using them
3. Create backends as referred in the map files
4. Create a rule for each backend to use
5. Assign this rule to a public server

Expected behavior

The domains of both map files should the routed to the proper backend.

Maybe this would work if OPNsense merges the entries of all map files of the same type into a single temporary file, which is added to the fronted config then.
If this is possible OPNsense should throw a warning.

Describe alternatives you considered

If I manually put all domain mappings into a single map file and state only a single rule for it in the public server settings, all is good.
However, with many domains and many backends, you may lose the track in the map file.

Environment

OPNsense 24.7.9 (amd64).
Intel(R) Xeon(R) Platinum 8461V
virtualized on KVM

[fraenki]

This sounds pretty similar to this report. You may want to check this upstream issue for further details.