We just moved h1guardian1 to new hardware, with more CPU and memory.  The guardian system is being recovered now.

The new machine has 20 hyperthreaded cores (Intel Xeon 2640 V4 2.4GHz) and 128G of RAM.  This is about twice the resources of the old machine.

The old machine was seeing frequent (couple of times a day) EPICS channel connection drop outs, which would cause EZCA connection errors on the guardian nodes and problems locking.  The load on the old machine was very high (>90% load on all 10 CPUs constantly), and the theory was that that was contributing the EPICS connection issues.  We should have considerably more headroom on this new machine, which will hopefully ameliorate the EPICS issues.

htop on the new machine: