This page describes the database flags that AlloyDB for PostgreSQL uses to enable and manage various service features unique to AlloyDB. For a list of all database flags that AlloyDB supports, see Supported database flags.
Flags marked with Instance restarts mean that AlloyDB restarts an instance whenever you set, remove, or modify this flag on that instance. The flag's value persists for the instance until you modify it again.
alloydb.audit_log_line_prefix
| Type | string |
| Default | "%m [%p]: [%l-1] db=%d,user=%u |
| Instance restarts | Yes |
Controls information prefixed to each audit log line. If left blank, then no prefix is used.
alloydb.enable_pgaudit
| Type | boolean |
| Default | off |
| Instance restarts | Yes |
Controls the availability of the pgaudit extension in an AlloyDB
instance. To use the extension, set the parameter to on and restart the instance. Then add the
pgaudit extension to individual databases in the instance by using the CREATE
EXTENSION command.
The pgaudit extension provides detailed session
and object audit logging using the standard logging facility provided by
PostgreSQL.
alloydb.iam_authentication
| Type | boolean |
| Default | off |
| Instance restarts | Yes |
Enables the use of Identity and Access Management (IAM) (Identity and Access Management) authentication with an AlloyDB instance.
alloydb.logical_decoding
| Type | boolean |
| Default | off |
| Instance restarts | Yes |
Enables the PostgreSQL logical decoding infrastructure in an AlloyDB instance.
Logical decoding provides the infrastructure necessary to stream data modifications to external consumers.
alloydb.log_throttling_window
| Type | integer |
| Default | 0 |
| Instance restarts | No |
Controls whether AlloyDB records duplicate log messages received during a specified time window.
If set to a value greater than 0, then AlloyDB starts a
timer that many seconds long whenever it records a log message. If
AlloyDB receives the identical log message before the timer
expires, it does not record the message. After the timer expires,
AlloyDB appends a count of any skipped messages to the
next occurrence of the same log message that it records.
If set to 0, then AlloyDB records all duplicate log
messages.
alloydb.promote_cancel_to_terminate
| Type | boolean |
| Default | off |
| Instance restarts | No |
If set to true, then any query cancellation requests due to issues such as lock timeouts, statement timeouts, and recovery (replay) conflicts that AlloyDB encounters, are promoted to forced backend termination requests.
By promoting the cancellation requests, AlloyDB helps prevent stuck backends from blocking log replay and other backends.
alloydb.pg_shadow_select_role
| Type | string |
| Default | An empty string |
| Instance restarts | No |
Name of the PostgreSQL role that is to be used to grant permission to query the
pg_shadow view.
scann.enable_preview_features
| Type | boolean |
| Default | off |
| Instance restarts | Yes |
Enables the use of the following features that are available in Preview:
- Adaptive filtering: identify whether inline filtering or pre-filtering provides the best performance at any given point during query execution.
- Auto index maintenance: incrementally manage the index such that when your dataset grows, AlloyDB splits large outlier partitions and tries to provide better QPS and search results.