Currently gluster processes call bind call only once to bind a socket, If it does not return success the process has exit. Sometime the kernel takes time to cleanup socket fd the situation is normal when a user has started a gluster process again and again usually in case of client process(shd,quota,nfs) happens as graph changes.If socket fd is not cleanup then bind throw an error EAADRINUSE and process exit.In that case a user has to restart the process again and again unless the port is not cleanup, we observed the similar kind of behavior multiple times in downstream on customer environment and for product aspect it does not create good impression. Solution: To avoid an issue we introduce a retries option(https://bugzilla.redhat.com/show_bug.cgi?id=1743020) long time ago in upstream i think same we need to backport in downstream also.
The retries option is not harmful it does tries to call bind retry(5 to 10 times) to avoid an issue. The recent downstream bugzilla(https://bugzilla.redhat.com/show_bug.cgi?id=2015122) because of the similar issue.
Since the problem described in this bug report should be resolved in a recent advisory, it has been closed with a resolution of ERRATA. For information on the advisory (glusterfs bug fix update), and where to find the updated files, follow the link below. If the solution does not work for you, open a new bug report. https://access.redhat.com/errata/RHBA-2022:4840