We have a small deployment with between 10 and 15 users using Filr.

The VM is on XEN on SLES 12SP3. The VM has 18GB of RAM and 6 CPUs allocated.

Most of the time the system runs perfectly monitoring with top shows between 0 and 10% CPU being used. Then seeming at ramdom the CPU usage will creep up until the system is running dead slow. Looking at top will reveal dozens of pool-jits-threads running. See images below:

Click image for larger version. 

Name:	jits.PNG 
Views:	37 
Size:	76.1 KB 
ID:	6027

A reboot will clear these out and the system will run perfectly again sometimes for days, sometimes for a couple of weeks before it happens again.

Where can I start looking to try and diagnose what is causing this?