[receiver/sqlserver] Add Always On Availability Group database replica metrics by splunk-shanu · Pull Request #49476 · open-telemetry/opentelemetry-collector-contrib

splunk-shanu · 2026-07-06T03:41:09Z

Adds 6 new opt-in metrics for monitoring SQL Server Always On Availability Groups, sourced directly from sys.dm_hadr_database_replica_states:

sqlserver.availability_group.database_replica.estimated_data_loss — estimated potential data loss if failover occurred now, expressed as age in seconds of the oldest unsent log record on the primary replica
sqlserver.availability_group.database_replica.estimated_recovery_time — estimated time in seconds to redo all log records in the redo queue on the secondary database replica
sqlserver.availability_group.database_replica.hardened_latency — time in seconds between a log record being generated on the primary and hardened to disk on the secondary (SQL Server 2016+)
sqlserver.availability_group.database_replica.queue.size — log send and redo queue backlog in bytes, distinguished via sqlserver.availability_group.queue.type attribute (log_send, redo)
sqlserver.availability_group.database_replica.log_send.rate — rate at which log data is being sent from the primary database replica to the secondary (bytes/s)
sqlserver.availability_group.database_replica.redo.rate — rate at which log data is being redone on the secondary database replica (bytes/s)

All metrics are scoped per database per replica and include sqlserver.availability_group.name, db.namespace, and sqlserver.replica.name attributes. All metrics are disabled by default and require a direct connection to SQL Server with Always On Availability Groups enabled.

Link to tracking issue

(fill in if applicable)

Testing

Unit tests added covering metric recording from sys.dm_hadr_database_replica_states. Test data files added under testdata/.

Documentation

documentation.md regenerated via make generate to include the 6 new metrics with their attributes, units, and extended documentation notes.

Authorship

I, a human, wrote this pull request description myself.

akgrover · 2026-07-06T20:16:24Z

@splunk-shanu do these metrics require additional permissions to what's documented right now?

thompson-tomo

See previous attempt in #48377

thompson-tomo · 2026-07-07T07:48:36Z

    gauge:
      value_type: double
    attributes: []
+  sqlserver.availability_group.estimated_data_loss:


Naming doesn't follow semconv naming guidance and is ambiguous.

follows {system}.{area}.{property} pattern correctly

It doesn't satisfy

Be precise. Attribute, event, metric, and other names should be descriptive and unambiguous.

For instance why is estimated in the name? What data is being lost?

At the same time, name gives the impression that it is the data has been lost but the measurement is in seconds. Data is not measured in seconds but rather bytes.

updated to more descriptive description, but name will have estimate as it is defined in dmv please refer dmv for more detail.

My quote is taken directly from the naming document https://opentelemetry.io/docs/specs/semconv/general/naming/

The name of the metric doesn't need to match db column name.

thompson-tomo · 2026-07-07T07:49:28Z

+      value_type: double
+    attributes: [availability_group.name, db.namespace, replica.name]
+    extended_documentation: This metric is only available when the receiver is configured to directly connect to SQL Server and the instance has Always On Availability Groups enabled.
+  sqlserver.availability_group.estimated_recovery_time:


Naming doesn't follow semconv naming guidance.

follows {system}.{area}.{property} pattern correctly

It doesn't satisfy

Be precise. Attribute, event, metric, and other names should be descriptive and unambiguous.

For instance why is estimated in the name?

At the same time it is a property of the replica rather than the group.

updated to more descriptive description, but name will have estimate as it is defined in dmv please refer dmv for more detail.

My quote is taken directly from the naming document https://opentelemetry.io/docs/specs/semconv/general/naming/

The name of the metric doesn't need to match db column name.

thompson-tomo · 2026-07-07T07:50:18Z

+      value_type: double
+    attributes: [availability_group.name, db.namespace, replica.name]
+    extended_documentation: This metric is only available when the receiver is configured to directly connect to SQL Server and the instance has Always On Availability Groups enabled.
+  sqlserver.availability_group.hardened_latency:


Naming doesn't follow semconv naming guidance and is ambiguous.

follows {system}.{area}.{property} pattern correctly

Hardened latency is not a property of the group but rather a replica.

thompson-tomo · 2026-07-07T07:51:18Z

+    extended_documentation: This metric is only available when the receiver is configured to directly connect to SQL Server and the instance has Always On Availability Groups enabled.
+  sqlserver.availability_group.log_send.queue.size:
+    enabled: false
+    description: Amount of log records of the primary database that has not been sent to the secondary replicas.


Description contradicts unit.

thompson-tomo · 2026-07-07T07:51:49Z

+    extended_documentation: This metric is only available when the receiver is configured to directly connect to SQL Server and the instance has Always On Availability Groups enabled.
+  sqlserver.availability_group.log_send.rate:
+    enabled: false
+    description: Rate at which log records are being sent from the primary database replica to the secondary replica.


Description contradicts unit.

thompson-tomo · 2026-07-07T07:52:01Z

+    extended_documentation: This metric is only available when the receiver is configured to directly connect to SQL Server and the instance has Always On Availability Groups enabled.
+  sqlserver.availability_group.redo.queue.size:
+    enabled: false
+    description: Amount of log records of the secondary database that has not yet been redone.


Description contradicts unit.

thompson-tomo · 2026-07-07T07:52:17Z

+    extended_documentation: This metric is only available when the receiver is configured to directly connect to SQL Server and the instance has Always On Availability Groups enabled.
+  sqlserver.availability_group.redo.rate:
+    enabled: false
+    description: Rate at which log records are being redone on the secondary database replica.


Description contradicts unit.

splunk-shanu added 5 commits July 3, 2026 12:28

intial commit

1818534

intial commit

4582848

Merge branch 'main' into sqlserver-ag-metrics

1331e03

changes

7aedac5

changes

956169f

sv-splunk reviewed Jul 6, 2026

View reviewed changes

Comment thread .chloggen/sqlserver-ag-metrics.yaml Outdated

thompson-tomo reviewed Jul 7, 2026

View reviewed changes

splunk-shanu added 3 commits July 7, 2026 10:54

changes

0bdf58c

changes

3ee52d5

changes

ffb8b7a

thompson-tomo reviewed Jul 7, 2026

View reviewed changes

splunk-shanu added 3 commits July 7, 2026 13:32

changes

3a52c40

changes

ceff2e0

changes

29b791f

Uh oh!

Conversation

splunk-shanu commented Jul 6, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Authorship

Uh oh!

Uh oh!

akgrover commented Jul 6, 2026

Uh oh!

thompson-tomo left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

splunk-shanu commented Jul 6, 2026 •

edited

Loading