Dedicated exec nodes are more or less stateless â Iâve been running a collection of diskless exec nodes using the HPE Insight Cluster Management Utility for a
few years now. The machines boot from an NFS export on the fileserver, and use the disks that used to hold a RAID-1 Linux OS as a RAID-0 /var/lib/condor/execute path, and pulled their configuration from /home/condor/config with some $(HOSTNAME) files and conditional
config. The only backup I need for them was of the PXE and DHCP config and the system boot images.
-Michael Pelletier.
![[Computer Systems Lab]](https://www-auth.cs.wisc.edu/pics/csl_logo.gif){kind=logo}