For larger datasets, the memory used by each thread can grow quite large as ProxyValues are parsed. Once parsed they are kept on the heap for the worker thread between queries.
To avoid OOM Killing the entire server, the worker threads should be limited to a maximum heap size, which must be configurable.
Then when a single worker grows to large it is simply killed and restarted, without the server itself going down.
See: https://nodejs.org/api/worker_threads.html#workerresourcelimits
For larger datasets, the memory used by each thread can grow quite large as ProxyValues are parsed. Once parsed they are kept on the heap for the worker thread between queries.
To avoid OOM Killing the entire server, the worker threads should be limited to a maximum heap size, which must be configurable.
Then when a single worker grows to large it is simply killed and restarted, without the server itself going down.
See: https://nodejs.org/api/worker_threads.html#workerresourcelimits