Viewing agent details
Last updated 2023-06-07
You can access a detailed view of each agent from the Agents page. The agent details provide a summary of the status and performance of the agent.
You can access the agent details from the Agents page.
To view agent details:
- Log in to the Signal Sciences console.
- From the Sites menu, select a site if you have more than one site.
- Click Agents in the navigation bar near the top of the screen.
- From the agents list, click the Agent ID of the agent that you want to view details for.
The following tabs appear in the agents detail:
Signal Sciences provides a number of metrics to understand the performance impact on your infrastructure. You can verify the performance impact on your infrastructure directly in the console via the Charts tab.
The Charts tab contains graphs detailing the requests the agent has processed, any errors observed, memory usage, CPU percentage, decision times, and more. Many of these graphs report percentiles. For example, the Decision time (ms) graph uses percentiles to show the amount of time it takes the agent to process requests. When the 99th percentile line is at 10 ms, the agent took between 0 - 10 ms to process 99% of requests and more than 10 ms to process the remaining 1% of requests.
You can use the time selector in the top left to change the time range being reported on all of the charts.
The following graphs appear on the Charts tab:
- Total requests (req/sec): the number of requests per second received for your site.
- Connections (req/sec): the total number of connections and the total number of dropped connections per second.
- Connections open: the total number of open connections handled by the agent.
- CPU percentage: the percentage of CPU used by the host and by the agent.
- Runtime memory total, heap and stack (bytes): the total memory being used by the agent.
- Uptime (seconds): the agent uptime, in seconds.
- Decision time (ms): the amount of time it took for the agent to process requests.
- Agent queue time (ms): the amount of time it took for the agent to begin processing requests where there is a backlog of requests.
- Agent clock skew (seconds): the difference in time between the agent's clock and the platform's clock.
To view advanced charts, click the Show advanced charts button. The following graphs appear.
- Requests uploaded (req/sec): the number of requests per second uploaded by the agent to the platform.
- Requests after sampling (req/sec): the number of incoming requests per second handed to the agent.
- Request size, avg. (bytes): the average size of web request in bytes. This is calculated by dividing the sum of read and written bytes by the number of requests.
- RPC PreRequest (req/sec): the number of incoming requests handed to the agent.
- RPC UpdateRequest (req/sec): the number of remote calls to agent to update details of request, post execution. Reported in requests per second.
- RPC PostRequest (req/sec): the number of remote calls to the agent to note requests that met one of the following conditions: the server returned an HTTP status of 400 or above, the size of the request was abnormally large, or the request took an abnormal amount of time to process.
- Goroutines: the number of goroutines running simultaneously on the agent
- Garbage collections: the number of time the garbage collector ran.
- GC pause time (ms/sec): the garbage collector pause time in milliseconds.
- Host memory available (bytes): the amount of memory available on the host machine in bytes.
- Rule updates: the number of times the agent updated its rules.
- Communication failures: the number of failed uploads and downloads per agent.
- Agent latency time (ms): the total amount of time that requests are in the queue and processed by the agent.
- Agent post time (ms): the amount of time since the agent's last data sync with the cloud engine.
- RPC MissedUpdateRequest (req/sec): the number of missed remote calls to the agent to update details of the request, post execution. Reported in requests per second.
- Agent missed update time (ms): the amount of time that has passed since the agent update timed out.