I have a cluster of four servers based on next.
I created 400 tables in the test database, and then ran
r.db('test').reconfigure({shards: 8, replicas: 4})

Two things happened:

• One of the servers started using a lot of memory. According to htop, it used in the range of 20 GB, with >40 GB virtual memory allocated to that process. I think this part is ok, though at some point we might want to look into it (and maybe throttle the query or something).
• The server started printing lots of these messages: <jemalloc>: Error in munmap(): Cannot allocate memory. It kept running though and everything still seems to be fine after the reconfiguration has finished.

Three questions:

• What does this message mean exactly?
• If an allocation failed, why is the server not crashing? Is jemalloc internally recovering from it somehow?
• If this is just a sort of warning without any consequences, can we suppress it?