Sharing a recent experience we encountered while upgrading our DNAC cluster. Our upgrade was from 1.3.1.3 directly to 1.3.3.4. The error we were seeing in the maintenance UI was specifically:
Phase:
Node x.x.x.x - Backup NFS server mount failed
The CLI error was:
State : FAILED
Sub-State : INSTALLED_HOST_COMPONENTS
Our issue was that at 34% the upgrade continued to fail. We attempted to remove all configured and scheduled backup jobs, but that did not do the trick. Per TAC during the upgrade process it performs some disk checks before updating the cluster. So during the disk checks it may encounter the error mentioned above. Note that it is not actually trying to mount NFS. It is actually failing because of other dead mount points and utilizing NFS as the scapegoat.
While checking the disk free output, the command (sudo df -h), or (sudo df -h | grep -i trans) will return output that will contain: "Transport endpoint is not connected".
Applying the workaround noted in the bug link ($for i in $(sudo df -h 2>&1 | grep "Transport endpoint is not connected" | awk -F: ' { print $2 } '); do sudo umount $i; done) will unmount all dead mounts that relate to the error. After applying the workaround our upgrade was able to be successfully completed.
Bug link: https://bst.cloudapps.cisco.com/bugsearch/bug/CSCvt95075/?rfs=iqvred
HTH!
