Hi,
I have a PostgreSQL server connected to 2 redundant network segments.
Therefore I needed to specify 2 IP addresses in ODBC configuration of a client which is also connected to these 2 networks.
I found this: https://www.enterprisedb.com/blog/improve-database-resiliency-new-jdbc-connectors-and-edb-failover-manager
saying that jdbc supports this feature, syntax is: jdbc:postgresql://host1:port1,host2:port2/database
However, I found no such information for ODBC driver.
I tried anyway to configure multiple addresses .. and it works. If you specify "Server" (in Windows ODBC) or "Servername" (in Linux /etc/odbc.ini or in windows registry configuration of ODBC) as "server1[,server2[,server3...]] then it works (driver tries to connect first to server1, then to server2 etc .. so the connection may take some time, but it is established successfully even if e.g. server1 is down). In windows, whitespaces are not permitted in the string.
So it seems that this is undocumented (or I'm not able to find this info anywhere).
In your release notes (https://odbc.postgresql.org/docs/release.html) there is however one interesting line. In section "psqlODBC 08.02.0101" there is
4) Handle multi-addresses of a specified host properly.
which suggests that this feature is in ODBC drive quite long. Of course, this can be used also for configuring access to multiple redundant PostgreSQL servers.
Can you please add this to documentation/FAQ? Might be interesting for people as PostgreSQL is being used also in environments requiring high availability, redundancy etc (e.g. we use it in SCADA solutions)
best regards
peter
