Plan and manage an Azure AI solution 1.3 Manage, monitor, and secure AI systems
You need to query Azure AI Search operation logs (for example, slow queries and indexing…
Study guide skill: Monitor data ingestion quality, search index health, and relevance performance
Question B1-013
You need to query Azure AI Search operation logs (for example, slow queries and indexing operations) in Log Analytics alongside other application logs. Nothing appears in the workspace. What is required?
Select one answer.
Answer: D
Resource logs aren't collected and stored until you create a diagnostic setting that routes chosen log categories to a destination such as Azure Monitor Logs (Log Analytics). Semantic ranker and replicas affect relevance and capacity, not log collection.
Deep dive
How it works
Azure AI Search generates resource logs for operations such as queries and indexing, but generating isn't the same as keeping. The monitoring page is explicit: logs are generated automatically, but you must route them to Azure Monitor Logs to save or query them, and resource logs aren't collected and stored until you create a diagnostic setting.
A diagnostic setting does two jobs:
- It picks the log categories to collect. When you create the setting, you specify which categories of logs to collect.
- It picks the destination. The suggested default is Azure Monitor Logs, so you can query search logs together with your other log data. Azure Storage, Event Hubs and partner solutions are other options.
Once logs flow to the workspace, you query them in Log Analytics, for example with AzureDiagnostics | summarize count() by OperationName to see which operations ran. Platform metrics behave differently: Azure Monitor provides them for most services, and you can route them to Logs too if you want them alongside other log data.
So an empty workspace has a simple cause. No diagnostic setting exists, or it doesn't send the right categories to that workspace.
The other options, and when you'd use them
- Increase the number of replicas: Replicas are capacity. The capacity guidance has you estimate required storage with partitions and required throughput with replicas. More replicas help query load, not log collection.
- Enable semantic ranker on the index: Semantic ranker reranks search results with language understanding models to improve relevance. It has nothing to do with where logs go.
- Nothing; resource logs are collected automatically: This is the trap. Logs are generated automatically, but they aren't collected and stored until a diagnostic setting routes them somewhere.
Exam tip
"Logs missing in Log Analytics" almost always means "create a diagnostic setting". Generated is not stored: no diagnostic setting, no resource logs in the workspace.
References
Last reviewed: