IPV4 error for dualstack cluster without IPV6 connectivity

[laskoviymishka]

When we try to connect to Postgres inside dual-stack network (ipv4 and ipv6) we face confusing error:

failed to connect to `host=chinook.chiyepnb01it.eu-west-2.rds.amazonaws.com user=tutorial database=chinook`: dial error (timeout: dial tcp 10.0.102.63:5432: i/o timeout)

Given:

Cluster with ipv4 and ipv6 address:

host chinook.chiyepnb01it.eu-west-2.rds.amazonaws.com
chinook.chiyepnb01it.eu-west-2.rds.amazonaws.com has address 10.0.102.63
chinook.chiyepnb01it.eu-west-2.rds.amazonaws.com has IPv6 address 2a05:d01c:d05:6500:3318:e752:f216:3d9b

IPv4 connectivity exists:

telnet 10.0.102.63 5432
Trying 10.0.102.63...
Connected to 10.0.102.63.
Escape character is '^]'.

IPv6 connectivity not exists:

 telnet 2a05:d01c:d05:6500:3318:e752:f216:3d9b 5432
Trying 2a05:d01c:d05:6500:3318:e752:f216:3d9b...
^C

Connection failed (wich is expected), but error should be:

failed to connect to `host=chinook.chiyepnb01it.eu-west-2.rds.amazonaws.com user=tutorial database=chinook`: dial error (timeout: dial tcp [2a05:d01c:d05:6500:3318:e752:f216:3d9b]:5432: i/o timeout)

[laskoviymishka]

Side note:

In pgx/v5 its not reproduceable:

failed to connect to `host=chinook.chiyepnb01it.eu-west-2.rds.amazonaws.com user=tutorial database=chinook`: dial error (timeout: context deadline exceeded)

Error is less verbose