If you have many lower-end devices and have the uptime reindex method set
when many of the devices go down and come back online the host recache process starts
this process is currently single-threaded and can take time to complete if it does not complete in the max
allotted time its killed and the process restarts

it would be best to allow multi processes on this script to make it quicker so it completes