Configuring Deadlock Monitoring on SQL Server
このページは日本語には対応しておりません。随時翻訳に取り組んでいます。翻訳に関してご質問やご意見ございましたら、お気軽にご連絡ください。
The Deadlock view enables you to explore deadlock events in your SQL Server database.
A deadlock occurs when two or more processes are unable to proceed because each is waiting for the other to release resources.
This guide assumes that Database Monitoring has been configured for your SQL Server.
Before you begin
- Supported databases
- SQL Server
- Supported deployments
- All deployment types except Azure DB (Azure Managed Instance is supported)
- Supported Agent versions
- 7.59.0+
Setup
In the SQL Server database instance, create a Datadog Extended Events (XE) session. You can run the session on any database in the instance.
Note: If the Datadog XE session isn’t created in the database, the Agent still attempts to collect deadlock events from a default SQL Server XE view. This view writes to the buffer pool, but there’s a higher chance of missing events because of a size limitation on the XML queried from it. For more information, see You may not see the data you expect in Extended Event Ring Buffer Targets on the SQL Server Support Blog.
CREATE EVENT SESSION datadog
ON SERVER
ADD EVENT sqlserver.xml_deadlock_report
ADD TARGET package0.ring_buffer
WITH (
MAX_MEMORY = 1024 KB,
EVENT_RETENTION_MODE = ALLOW_SINGLE_EVENT_LOSS,
MAX_DISPATCH_LATENCY = 30 SECONDS,
STARTUP_STATE = ON
);
GO
ALTER EVENT SESSION datadog ON SERVER STATE = START;
GO
- In the Datadog Agent, enable deadlocks in
sqlserver.d/conf.yaml
.
deadlocks_collection:
enabled: true
Exploring deadlock events
To access the deadlock view, navigate to the APM > Database Monitoring > Databases tab, select a SQL Server host, and then click the Deadlocks tab.
The deadlock tab displays details about the victim and survivor processes, and includes a link to the deadlock diagram.
Note: Because deadlocks occur infrequently, it’s unlikely that any deadlock information will be visible right away.
Further reading