If there is a transient failure in one of the hosts when vlcm Parallel remediation is enabled, the next remediations end up failing consistently (on a different set of hosts). This is because those hosts are in a reboot pending state.
Remediation was designed to fail fast, i.e. whenever remediation fails on the host in cluster, the entire remediation task for the cluster was marked as failed without checking other hosts status with in the cluster.
Currently there is no resolution. In the future release, vCenter task would wait and complete the remediation of all these hosts.
In case users notice any host consistently failing in remediation, that host may be in pending reboot state and the user needs to manually trigger the reboot. If the host is failing to remediate even after it was manually rebooted, there might be some other issue causing this. Please refer to the remediation task logs and the vLCM logs to understand the reason for the failure. If it is not possible to determine the cause of the remediation failure from these, please reach out to the vLCM engineering team with the VC and host support bundles to determine the cause of the failure.