Hi ,
Thanks for reaching out to Microsoft Q&A.
The reason why filtering and splitting are disabled for your custom metric (numUsed) in Azure Metrics Explorer is likely due to how custom metrics are structured and ingested into Application Insights. Here’s what’s happening and how to resolve it:
Understanding Why Filtering/Splitting is Disabled
Custom Metrics in Metrics Explorer Need Dimensions
- Azure Metrics Explorer supports filtering and splitting only if the custom metric is ingested with dimensions at the time of tracking.
- A custom dimension (like
computerName) added viaTelemetryClient.Context.Propertiesis only available in Log Analytics, not as a native metric dimension.
- When you add `computerName` as a common property, it does not make it a first-class metric dimension. - Metrics Explorer can only filter/split on dimensions that were included when tracking the metric. - A custom dimension (like
Solution: Send computerName as a Dimension
To allow filtering/splitting in Metrics Explorer, you need to send computerName as a dimension when tracking the metric.
Step 1: Modify How You Send the Metric
Instead of using TelemetryClient.TrackMetric(name, value), use MetricIdentifier and dimensions.
Step 2: Validate in Metrics Explorer
- Wait for 5-10 minutes after deployment.
- Go to Azure Portal > Application Insights > Metrics Explorer.
- Select the
numUsedmetric and check ifcomputerNameappears under "Split By".
Alternative: Use KQL (If You Stick With Logs)
If you want to continue using KQL, you can achieve similar results with this query:
customMetrics | where name == "numUsed" | extend instanceId = tostring(customDimensions["computerName"]) | summarize avg(numUsed) by bin(timestamp, 1m), instanceId | render timechart
Note:
- If the metric appears in Log Analytics with
computerNamebut not in Metrics Explorer, it meanscomputerNamewas not sent as a metric dimension. - Once sent correctly, filtering and splitting in Metrics Explorer should work without needing KQL.
Please feel free to click the 'Upvote' (Thumbs-up) button and 'Accept as Answer'. This helps the community by allowing others with similar queries to easily find the solution.