URL:
https://github.com/SSSD/sssd/pull/5407
Title: #5407: kcm: check socket path loaded from configuration
pbrezina commented:
"""
> Does this make sense?
Yes, now it makes sense why systemd activated sssd-kcm works correctly even though the
socket path in sssd.conf is incorrect.
> Therefore IMHO the solution would be to print error if systemd socket != sssd.conf
socket and document that socket patch should be changed in the unit file and not in
sssd.conf.
I also agree that it is best to print an error if the two paths are different. But what
happens to those distributions that don't use systemd? Should we add another check to
see whether systemd socket is not empty and then compare it with sssd.conf?
If systemd is not used, or sssd-kcm.socket is inactive then SSSD will create to socket on
its own (`activate_unix_sockets() -> ... -> create_pipe_fd()`) and `bind()` to
filesystem path will fail if the socket path is invalid.
Moreover, does the previous check also apply to other socket
initializations?
I'm sorry, I am not entirely sure what you ask. Can you rephrase it?
"""
See the full comment at
https://github.com/SSSD/sssd/pull/5407#issuecomment-756678534