If the worker keeps getting restarted, you might wanna checkout the perf metrics of what you're requesting, it could be your exhausting perf limits, also check what the watchdog limits and behavior are set to