In addition to the regular TimescaleDB configuration, it is recommended that you also configure additional settings specific to multi-node operation.
Each of these settings can be configured in the
postgresql.conf file on the
individual node. The
postgresql.conf file is usually in the
but you can locate the correct path by connecting to the node with
giving this command:
After you have modified the
postgresql.conf file, reload the configuration to
see your changes:
If not already set, ensure that
max_prepared_transactions is a non-zero value
on all data nodes is set to
150 as a starting point.
On the access node, set the
enable_partitionwise_aggregate parameter to
This ensures that queries are pushed down to the data nodes, and improves query
On the access node, set
off. Currently, JIT does not work well with
distributed queries. However, you can enable JIT on the data nodes successfully.
On the data nodes, disable
statement_timeout. If you need to enable this,
enable and configure it on the access node only. This setting is disabled by
default in PostgreSQL, but can be useful if your specific environment is suited.
On the data nodes, set the
logical or higher to
move or copy chunks between data nodes. If you
are moving many chunks in parallel, consider increasing
max_replication_slots as well.
For consistency, if the transaction isolation level is set to
it is automatically upgraded to
REPEATABLE READ whenever a distributed
operation occurs. If the isolation level is
SERIALIZABLE, it is not changed.
Found an issue on this page?Report an issue!