SQL Server 2025 on Linux: Msg 19417 persists after dropping EXTERNAL Availability Group and resetting HADR cluster context

Question

SQL Server 2025 on Linux: Msg 19417 persists after dropping EXTERNAL Availability Group and resetting HADR cluster context

Adil El Gatri 0

Product

SQL Server 2025 Enterprise Edition (Linux)

Ubuntu 24.04.4 LTS

Version: 17.0.4025.3 (CU3)

Secondary server:

SQL Server 2025 Enterprise Edition (Linux)

Ubuntu 25.10

Version: 17.0.4055.5 (CU6)

Environment

Two Linux servers connected through Tailscale.

Primary: Ubuntu VPS

Secondary: Ubuntu Home Server

Always On Availability Groups enabled using:

No Pacemaker.

No Corosync.

No Windows Server Failover Cluster.

Steps to reproduce

Enable HADR.

Create an Availability Group using:

Join the secondary.

Verify synchronization.

Restart SQL Server:

Replica enters RESOLVING because no Pacemaker exists.

Drop the Availability Group.

Execute:

Restart SQL Server.

Verify:

returns zero rows.

Create a new Availability Group using:

Expected result

Availability Group should be created successfully.

Actual result

SQL Server returns:

Diagnostics

sys.availability_groups

Returns:

sys.availability_replicas

Returns:

sys.dm_hadr_database_replica_states

Returns:

sys.dm_hadr_cluster

Returns:

even though no Availability Group exists.

sys.dm_hadr_cluster_members

Returns one row:

mssql.conf

No cluster configuration exists.

ALTER SERVER CONFIGURATION SET HADR CLUSTER CONTEXT = LOCAL

Succeeds without error.

However:

continues to show cluster information.

SQL Server Error Log contains messages such as:

even though SQL Server is running on Linux without Pacemaker.

Impact

The SQL Server instance appears permanently stuck in a remote cluster context.

It becomes impossible to create any new Availability Group.

The only apparent workaround is reinstalling the SQL Server engine.

Product

SQL Server 2025 Enterprise Edition (Linux)

Ubuntu 24.04.4 LTS

Version: 17.0.4025.3 (CU3)

Secondary server:

SQL Server 2025 Enterprise Edition (Linux)

Ubuntu 25.10

Version: 17.0.4055.5 (CU6)

Environment

Two Linux servers connected through Tailscale.

Primary: Ubuntu VPS

Secondary: Ubuntu Home Server

Always On Availability Groups enabled using:

No Pacemaker.

No Corosync.

No Windows Server Failover Cluster.

Steps to reproduce

Enable HADR.

Create an Availability Group using:

Join the secondary.

Verify synchronization.

Restart SQL Server:

Replica enters RESOLVING because no Pacemaker exists.

Drop the Availability Group.

Execute:

Restart SQL Server.

Verify:

returns zero rows.

Create a new Availability Group using:

Expected result

Availability Group should be created successfully.

Actual result

SQL Server returns:

Diagnostics

sys.availability_groups

Returns:

sys.availability_replicas

Returns:

sys.dm_hadr_database_replica_states

Returns:

sys.dm_hadr_cluster

Returns:

even though no Availability Group exists.

sys.dm_hadr_cluster_members

Returns one row:

mssql.conf

No cluster configuration exists.

ALTER SERVER CONFIGURATION SET HADR CLUSTER CONTEXT = LOCAL

Succeeds without error.

However:

continues to show cluster information.

SQL Server Error Log contains messages such as:

even though SQL Server is running on Linux without Pacemaker.

Impact

The SQL Server instance appears permanently stuck in a remote cluster context.

It becomes impossible to create any new Availability Group.

The only apparent workaround is reinstalling the SQL Server engine.

Erland Sommarskog 135.3K Reputation points MVP Volunteer Moderator

2026-07-02T15:27:06.85+00:00

I guess you used AI to compose your post, but then never cared to check the result. First of all the post repeats itself and say everything twice. Second, and more problematic, in all places where you want to show a command or show some output, this is missing. You may want to edit your post to address these issues.Generally, if you want to report a bug, the place to go is https://feedback.azure.com/d365community/forum/04fe6ee0-3b25-ec11-b6e6-000d3a4f0da0. Beware, though, that is only to let Microsoft know about the issue. There is no commitment from Microsoft that they will fix the issue, even less in the near future. If this is a stopping issue for you and you need a fix, you need to open a support case.

Your answer

Erland Sommarskog 135.3K Reputation points MVP Volunteer Moderator

2026-07-02T15:27:06.85+00:00

I guess you used AI to compose your post, but then never cared to check the result. First of all the post repeats itself and say everything twice. Second, and more problematic, in all places where you want to show a command or show some output, this is missing. You may want to edit your post to address these issues.Generally, if you want to report a bug, the place to go is https://feedback.azure.com/d365community/forum/04fe6ee0-3b25-ec11-b6e6-000d3a4f0da0. Beware, though, that is only to let Microsoft know about the issue. There is no commitment from Microsoft that they will fix the issue, even less in the near future. If this is a stopping issue for you and you need a fix, you need to open a support case.