In your scenario with a two-node Always On Availability Group (AG) setup without Active Directory, there are several factors to consider regarding the failure of the AG listener to go offline and the subsequent timeout issues you are experiencing.
- DNS Configuration: Since you mentioned that you did not register the A record in DNS, it is crucial to ensure that the DNS settings are correctly configured. The AG listener relies on DNS for name resolution, and without proper DNS registration, clients may struggle to connect, leading to timeouts.
- Static IP Configuration: Using a static IP for the AG listener is valid, but ensure that this IP is correctly configured in both the SQL Server and Windows Failover Cluster settings. If there are any discrepancies, it could lead to the listener failing to transition states properly.
- Cluster Resource Timeout: The error logs indicate that a component did not respond in a timely manner, causing the cluster resource to exceed its timeout threshold. You may want to check the health of the underlying infrastructure, including network connectivity and any services that the cluster resource depends on. This includes ensuring that there are no network issues or high latency between the nodes.
- Failover Cluster Configuration: Since you are using a witness share, ensure that it is accessible and functioning correctly. Any issues with the witness can also impact the failover process.
- Manual Failover Behavior: The random nature of the failover failures suggests that there might be intermittent issues with resource availability or network stability. Monitoring the cluster logs during failover attempts can provide more insights into what might be causing the delays.
- Cluster Resource Management: Consider adjusting the timeout settings for the cluster resource if you find that the defaults are too aggressive for your environment. This can help prevent premature failover attempts during brief periods of unresponsiveness.
By addressing these areas, you may improve the stability and responsiveness of your Always On AG listener during failover operations.
References: