Fix Proxmox Update Stuck at Processing Triggers for pve-ha-manager

Update hangs at “processing triggers for pve-ha-manager”? It’s usually the HA stack waiting on a cluster quorum or a stuck service. Check the cluster state and pve-ha-lrm, then finish dpkg safely.

Intricate overhead power lines against a clear sky showcasing urban infrastructure.

Why it hangs

The pve-ha-manager trigger runs after the package is unpacked. If the node is in a cluster, the trigger tries to talk to the cluster filesystem and the HA services. No quorum or a wedged pve-ha-lrm means the trigger never returns.

Check the cluster

First, see if the cluster is healthy:

pvecm status
ha-manager status

If pvecm status shows no quorum, fix that first. A node that can’t see its peers will hang on any trigger that touches the pmxcfs. If you’re not in a cluster, skip this.

Detailed view of a SATA power cable connector with colorful wiring.

Check the HA services

Look at pve-ha-lrm and pve-ha-crm:

systemctl status pve-ha-lrm pve-ha-crm

If pve-ha-lrm is stuck or restarting in a loop, the trigger waits forever. Restart it cleanly:

systemctl restart pve-ha-lrm pve-ha-crm

Sometimes the local resource manager gets confused about a VM it’s managing. Restarting both services usually clears it.

Safely finish dpkg

Don’t kill the dpkg process. That leaves the package half-configured and you’ll have a bigger mess. Instead, in another terminal, run:

dpkg --configure -a

This tells dpkg to finish configuring any pending packages. If the trigger is still stuck, you can tell dpkg to skip triggers temporarily:

dpkg --configure --pending --force-depends

But only do that if you’re sure the package itself is fine. The trigger is just a post-install script; forcing it to skip doesn’t corrupt anything, but you should run it manually later.

My take

This is a cluster problem, not a package problem. The trigger is just the messenger. Fix the quorum, restart the HA services, and the update finishes. If you’re on a single node and still see this, check if pve-ha-lrm is even needed—you can disable it if you don’t use HA. But on a cluster, don’t disable it; just get quorum back.

Related: Proxmox Cluster NFS: Frontend Connects, Backend Fails has more on cluster weirdness.