Gluster heal errors cannot be healed - glustershd logs are filling up with `remote operation failed. Path: (null) [Invalid argument]`

Issue

• After upgrade to RHGS 3.5.4 , Self-heal failing with remote operation failed: ( -> (null)) [Invalid argument]
• Glustershd log file is full with the below errors. The error does not specify which argument is invalid.

[client-rpc-fops_v2.c:216:client4_0_mkdir_cbk] 0-testvol-client-30: remote operation failed. Path: (null) [Invalid argument]

• At the same time, we can see on the brick logs errors like:

E [MSGID: 113002] [posix-entry-ops.c:661:posix_mkdir] 0-vol01-posix: gfid is null for (null) [Invalid argument]

• It has been found that in some cases heal errors are not progressing or cannot be healed , even when doing the following procedure:

[1] Usually, heal errors are fixed by enabling data, metadata, and entry self-heal, and the self-heal daemon:

# gluster volume set VOLNAME cluster.data-self-heal on
# gluster volume set VOLNAME cluster.metadata-self-heal on 
# gluster volume set VOLNAME cluster.entry-self-heal on
# gluster volume set VOLNAME self-heal-daemon on
+ mounting the volume on a gluster node on /testheal (other mount point name is valid)
+ doing a lookup (ls) on the heal files or directories