An Azure managed PostgreSQL database service for app development and deployment.
Identify Azure-managed azuresu localhost client backend sessions on Azure Database for PostgreSQL Flexible Server
I am using Azure Database for PostgreSQL Flexible Server and need to identify Azure-managed privileged sessions visible in pg_stat_activity.
The server is PostgreSQL 18.6. During read-only inspection, I observed sessions with:
- role:
azuresu - backend type:
client backend - client address:
127.0.0.1 - state:
idle
I also observe clearly identifiable Azure background processes such as pg-availability, pgms_stats_background_worker, pg_qs_background_worker, pg_cron launcher, and logical replication launcher.
My question concerns specifically the azuresu localhost sessions reported as client backend, not those named background workers.
I need to determine whether these azuresu / 127.0.0.1 client-backend sessions are Azure platform-managed service connections and, if so, how they can be distinguished reliably from other privileged client sessions.
Could Microsoft/Azure PostgreSQL engineering please clarify:
- Which Azure component or service creates these
azuresulocalhostclient backendsessions? - What is their purpose?
- Is there a documented or independently verifiable property in PostgreSQL metadata that can reliably identify them as Azure-managed, rather than relying only on the
azuresurole name or127.0.0.1address? - Can these sessions execute concurrent database operations that could affect DDL, role/privilege changes, triggers, functions, or transactions while a customer application is performing a protected database binding/sealing or maintenance transition?
- Are these sessions expected to remain connected during normal operation, and should customer applications treat them differently from customer-created privileged administrator sessions?
I am requesting diagnostic/architectural clarification only. I do not want to terminate these sessions or change Azure service configuration or privileges.
Please do not request credentials, connection strings, or business data. I can provide sanitized pg_stat_activity metadata if needed.I am using Azure Database for PostgreSQL Flexible Server and need to identify Azure-managed privileged sessions visible in pg_stat_activity.
The server is PostgreSQL 18.6. During read-only inspection, I observed sessions with:
- role:
azuresu - backend type:
client backend - client address:
127.0.0.1 - state:
idle
I also observe clearly identifiable Azure background processes such as pg-availability, pgms_stats_background_worker, pg_qs_background_worker, pg_cron launcher, and logical replication launcher.
My question concerns specifically the azuresu localhost sessions reported as client backend, not those named background workers.
I need to determine whether these azuresu / 127.0.0.1 client-backend sessions are Azure platform-managed service connections and, if so, how they can be distinguished reliably from other privileged client sessions.
Could Microsoft/Azure PostgreSQL engineering please clarify:
- Which Azure component or service creates these
azuresulocalhostclient backendsessions? - What is their purpose?
- Is there a documented or independently verifiable property in PostgreSQL metadata that can reliably identify them as Azure-managed, rather than relying only on the
azuresurole name or127.0.0.1address? - Can these sessions execute concurrent database operations that could affect DDL, role/privilege changes, triggers, functions, or transactions while a customer application is performing a protected database binding/sealing or maintenance transition?
- Are these sessions expected to remain connected during normal operation, and should customer applications treat them differently from customer-created privileged administrator sessions?
I am requesting diagnostic/architectural clarification only. I do not want to terminate these sessions or change Azure service configuration or privileges.
Please do not request credentials, connection strings, or business data. I can provide sanitized pg_stat_activity metadata if needed.