Monitoring semaphore queue efficiency through Application Insights

  • リリースバージョン: Australia
  • 更新日 2026年03月12日
  • 所要時間：4分

    • Monitor semaphore queue efficiency by tracking the queue depth and number of rejected transactions through the Application Insights semaphore graphs.

    Starting with the Zurich release, Application Insights is no longer deployed, enhanced, or supported. It is recommended to evaluate the Overview of Instance Observer product available with the ServiceNow Impact packages. Work with your Account team to review Impact packages.

    For details, see the Deprecation Process [KB0867184] article in the Now Support Knowledge Base.

    You can monitor performance of the semaphore queues in your instance by comparing the semaphore queue depth and rejection count in the semaphore graphs on the Overview tab.

    Through the semaphore graphs, you can monitor the following metrics:

    • The semaphore queue depth, which enables you to help prevent backlogs
    • Rejection counts, which enables you to identify processing bottlenecks

    Semaphores control the number of simultaneous transactions that can execute on a node. Low available semaphores indicate that the instance is running close to full transaction capacity. Incoming transactions will wait until the semaphores are available.

    You access the semaphore graphs by navigating to All > Application Insights > Application Insights > Overview.

    • Monitor the load level by comparing the queue depth to the queue depth limit that appears as a red line on each semaphore graph.
    • Look for correlations or trends by comparing the queue depth metric to the rejection count metric over time.

    Dig deeper into semaphore processing efficiency by drilling down to analyze issues at the node-level in the detail graphs on the Session Info tab.

    • Look for spikes in the queue depth graphs. A high transaction count indicates the node might be overloaded. Select a data point in the detail graph to view a list of transactions. In the table, look for long-running transactions on the semaphore and stop them. Reduce the transaction count if the queue depth is consistently high.
    • Look for spikes in the rejection count graphs. A high rejection count indicates an unexpected proliferation in calls. If the queue depth is reached, all subsequent requests are rejected, which results in a rejected requests (HTTP 429) error message. Reduce the number of calls or stop long-running transactions that might be backing up the queue.