Yeah, wal can also grow very large, making read operations increasingly slow on wrong checkpoint configuration, which also makes checkpointing slower and it grows out of proportion.
Inefficient queries, no reader gaps and a bad manual checkpointing system can bring the system down.