The Application Details page gives you a consolidated view of your application's runtime health: request metrics, JVM internals, thread and connection pool utilization, SQL call patterns, exceptions, and upstream/downstream dependencies.
Access the application details page
Go to the SAE Applications page. Select the destination region and namespace from the top navigation bar, then click the Application ID.
In the left navigation pane, choose Application Monitoring > Application Details.
Overview
The Overview tab shows four time-series charts for the selected time range:
| Metric | What it shows |
|---|---|
| Request Count | Total inbound requests per time interval. A sudden drop may indicate traffic loss or a deployment issue. |
| Response Time | End-to-end response latency per time interval. Sustained increases typically indicate downstream saturation or resource contention. |
| Slow Call Count | Number of calls that exceeded the slow-call threshold per time interval. Correlate with Response Time to identify whether slowness is widespread or isolated. |
| HTTP Status Code | Distribution of HTTP response status codes per time interval. A rise in 5XX codes indicates server-side errors affecting callers. |
Chart interactions
On any chart, you can:
Hover to see point-in-time statistics.
Select a time range on the chart, then click Enter Selected Time Period to focus monitoring on that window, or click View Logs For Selected Time Period to open the corresponding logs.
Click the
icon to view statistics for a specific period or compare the same period across different dates.Click the
icon to view the API details for the metric.Click the
icon to zoom in.
Topology view
The Topology tab visualizes call relationships between the application and its upstream and downstream components. Use the topology graph to quickly identify bottlenecks in the call chain.
JVM monitoring
The JVM Monitoring tab shows time-series curves for seven metric groups:
| Metric group | Metrics | Description |
|---|---|---|
| Instantaneous GC Count | Full GC count; young generation GC count | Number of garbage collection events per interval. Frequent full GCs indicate heap pressure and can cause application pauses. |
| Instantaneous GC Time | Full GC time; young generation GC time | Time spent in GC per interval. For GC types other than ZGC and Shenandoah, this is the Stop-the-World (STW) duration — the period during which all Java threads are paused. For ZGC and Shenandoah, Pauses correspond to STW duration and Cycles represent the total time for one complete GC. |
| Heap Memory Details | Total heap; old generation (bytes); survivor space (bytes); eden space (bytes); committed heap (bytes) | Current heap utilization by memory region. Old generation usage approaching the committed limit typically precedes a full GC. |
| Metaspace Details | Metaspace size (bytes) | Memory used for class metadata. |
| Non-Heap Memory | Committed (bytes); initial (bytes); maximum (bytes) | Non-heap memory tracked by JMX. This is a subset of the actual non-heap regions in the process, so the sum of heap and non-heap shown here may differ from the resident set size (RES) reported by top. See JVM monitoring memory details. |
| Direct Buffer | Total DirectBuffer size (bytes); used DirectBuffer size (bytes) | Off-heap memory allocated directly. |
| JVM Thread Count | Total; deadlocked; new; blocked; runnable; terminated; timed-waiting; waiting | Thread state breakdown. A growing blocked or waiting count suggests lock contention or I/O bottlenecks. |
JVM monitoring data comes from Java Management Extensions (JMX).
Chart controls
In the Instantaneous GC Count and Instantaneous GC Time panels, click Instantaneous Value or Accumulated Value in the upper-right corner to switch between the two curve types.
Click a metric name (for example, FullGC Count) on any panel to show or hide its curve. Each chart must display at least one metric — hiding the last visible metric is not allowed.
Click the
icon on any panel to compare statistics across time periods or dates.Click the
icon to zoom in.Click the
icon on the Heap Memory Details/min, Metaspace Details/min, Non-Heap Memory/min, Direct Buffer/min, or JVM Threads/min panel to view the API details for that metric.
Thread pool and connection pool monitoring
The Thread Pool Monitoring and Connection Pool Monitoring tabs display the following metrics for each pool:
| Metric | Description |
|---|---|
| Core Thread Count | Minimum number of threads kept alive in the pool |
| Maximum Thread Count | Upper limit on threads the pool can create |
| Active Thread Count | Threads currently executing tasks |
| Current Thread Count | Total threads currently in the pool |
| Historical Maximum Thread Count | Peak thread count since the application started |
| Completed Task Count | Total tasks finished since the application started |
| Scheduled Task Count | Tasks queued for execution |
| Queue Size | Number of tasks waiting in the pool queue |
For supported frameworks and the full list of tracked metrics, see Thread pool and connection pool monitoring. For agent version requirements, see Agent management.
To upgrade the agent version, contact the technical support team in the DingTalk group (ID: 32874633).
SQL call analysis
The default threshold for a slow SQL query is 500 ms. To adjust this threshold, see Threshold settings.
The SQL Call Statistics tab shows a time-series curve of SQL call activity over the selected period.
On the chart, you can:
Hover to see point-in-time statistics.
Click the
icon to compare statistics across time periods or dates.Click the
icon to view API details for the metric.Click the
icon to zoom in.
Exception analysis
In ARMS application monitoring, an exception maps to a Java Exception. Because of the try-catch mechanism, an exception caught inside an API call may not surface to the caller — and multiple try-catch blocks can produce multiple exceptions for a single API call. An error is a specific case: an uncaught exception that affects the return value of an API call.
The Exception Statistics area shows a stacked column chart and a list of exceptions during the selected period.
On the chart, you can:
Hover to see point-in-time statistics.
Select a time range to narrow the view.
Click the
icon to compare statistics across time periods or dates.Click the
icon to view API details for the metric.Click the
icon to zoom in.
Upstream and downstream applications
An upstream application calls the current application — it actively sends requests to trigger business logic. A downstream application is called by the current application — it is a service the current application depends on to execute its business logic.
The Upstream Applications and Downstream Applications tabs show Response Time, Request Count, and Error Count for each dependency.
Error analysis and trace query
The SAE console does not provide direct access to error analysis or trace queries. Click the Error Analysis or Trace Query tab to open these views in the ARMS console.
Related topics: