We have a Linux server 16GB RAM and HD SSD with Postgres 9.5 and configuration:
max_connections = 155
shared_buffers = 4GB
work_mem = 14MB
maintenance_work_mem = 744MB
effective_io_concurrency = 200
effective_cache_size = 12GB
With these settings the memory usage is between 10 ~ 13GB and when running a dump occur errors of "out of memory" preventing new requests. By changing the effective_cache_size setting to 6GB, the memory usage is between 8 ~ 12GB, without errors during the dump.
From the manual i understood that the performance of the effective_cache_size configuration would be on the scheduler with not having such a large impact on memory usage. Has anyone done any study on this setup and know the why of this behavior?