Description of problem:
aardvark-dns, two processes, trying to use the same port on the same interface
Version-Release number of selected component (if applicable):
Rhel8.6
How reproducible:
Reproducible when using customer product on low power hardware.
Steps to Reproduce:
1. podman starts first container, checks if there's aardvark-dns process running .
2. As there is no aardvark-dns instances running it starts it
3. aardvark-dns initializes and writes it's PID to pidfile
4. Another container is being started in parallel with the first one, so podman performs aardvark-dns checks
5. As first aardvark-dns hasn't finished initializing yet it haven't wrote its PID to pidfile
6. Second container assumes there's no aardvark-dns running due to an empty pidfile
7. Attempts to start it
8. We now have two aardvark-dns instances, one (the first one), bound to port 53 (as it was able to do it because it started and requested it first), second overwriting pidfile with its PID (unable to bind as port is already in use).
Actual results:
Observe 2 aardvark-dns processes running on the same bridge interface which is used by containers.
Expected results:
Only one aardvark-dns instances.
I'm closing this as a duplicate of https://bugzilla.redhat.com/show_bug.cgi?id=2116481. If I'm mistaken and this is not a dupe, please reopen and add a comment with clarification please.
*** This bug has been marked as a duplicate of bug 2116481 ***