If within your computer provisioning a script is used, you can use the
jamf api to clear the failed commands and then run a Jamf recon. This
article was helpful for me
(https://aporlebeke.wordpress.com/2019/01/04/auto-clearing-failed-mdm-commands-for...
If it helps, I was having the same issue with computers randomly
becoming "unmanaged." Since I know our management account username/
password, I was able to create a script that finds all unmanaged
computers that were formally managed and "re-manage"...
