Stepper motors should hold position when error.

In your case, you should have gotten a servo/stepper with brake for Z axis.
It is a hardware feature, so software can not help, but if the loop is closed in LinuxCNC at least it will know the position.
PCW's idea with enabling fast after fault and if loop is closed in LinuxCNC is very good.
As for your question, drive fault will disable the motor in the drive, so nothing LinuxCNC can do about it (except above), and so will LinuxCNC disable everything when it gets a fault error.

The z axis motor does have a brake but there is s delay between the motor going dead and brake engaging. In that delay time the z axis drops a couple centimeters.

I will try this, but again this will introduce a delay. Currently when the motor disables the brake enables but there is a delay so z axis drops a couple centimeters.

If the drive itself drops the holding current on a fault,
you are always going to lose position, there is really
nothing LinuxCNC can do about this.

How far it drops depends on how fast the drive can be
reset or brake applied. LinuxCNC can react very quickly
to  a fault signal (< 1 ms), but I would expect the external hardware
(brake or drive reset) to be slower.

LeonDolnicar wrote:

Currently when the motor disables the brake enables but there is a delay so z axis drops a couple centimeters.

How heavy is the Z axis?
This seems like to heavy, so it would need a counterweight, and if so, it will also cause to much stress on the motor and drive trying to keep all that weight in position. Also mitigated by said counterweight.

I dont think the weight is the problem maybe more that the ball screw is fast. 20mm per turn. I got a good quality rexroth ballscrew so I used it. Because of the faster screw I used a stronger stepper motor 12Nm nema34.

Still, the counterweight would solve your existing issue, and not hard to implement.