Error when doing too many calls, pq: remaining connection slots are reserved for non-replication superuser connections

[rogerfar]

When running too many queries at the same (I have a large amount of users and databases) a command like terraform destroy will break the provider with the error:

pq: remaining connection slots are reserved for non-replication superuser connections

It might be needed to more efficiently open and close connections when needed, or limit the amount of open connections.

[tjb900]

We're seeing this also, even with max_connections = 1.
lsof shows several parallel connections to the server:

terraform 3971985 user   11u     IPv4          911087438      0t0       TCP x:60706->y:postgres (SYN_SENT)
terraform 3971985 user   12u     IPv4          911087439      0t0       TCP x:60708->y:postgres (SYN_SENT)
terraform 3971985 user   13u     IPv4          911187131      0t0       TCP x:60674->y:postgres (ESTABLISHED)
terraform 3971985 user   14u     IPv4          911182105      0t0       TCP x:60686->y:postgres (ESTABLISHED)
terraform 3971985 user   15u     IPv4          911224039      0t0       TCP x:60646->y:postgres (ESTABLISHED)
terraform 3971985 user   16u     IPv4          911167979      0t0       TCP x:60716->y:postgres (SYN_SENT)
terraform 3971985 user   17u     IPv4          911165572      0t0       TCP x:60692->y:postgres (ESTABLISHED)

So somehow the max connections setting is not being respected.