You have an Azure Event Grid system topic that delivers events to an HTTPS endpoint on an Azure Function. Occasionally the endpoint is throttled, and events are dropped. You must (1) capture undeliverable events for later inspection in a blob container and (2) reduce unnecessary retries when the endpoint is unhealthy. What should you do?

The problem requires two solutions: capturing undeliverable events and reducing unnecessary retries. 1. Capturing undeliverable events: Configuring a dead-letter destination for the Event Grid subscription is the correct approach. When Event Grid cannot deliver an event after exhausting its retry policy, it sends the event to the dead-letter destination. A blob container in an Azure Storage account is a valid and common choice for this. A managed identity with 'Storage Blob Data Contributor' role provides secure access to the storage account for Event Grid. 2. Reducing unnecessary retries: Adjusting the event subscription's retry policy, specifically MaxDeliveryAttempts and EventTimeToLiveInMinutes, allows you to control how aggressively Event Grid retries delivery. Setting these values appropriately for your endpoint's expected recovery time prevents Event Grid from continually retrying a temporarily unhealthy endpoint, thus reducing unnecessary load and dropped events. Poison message handling for Azure Functions (incorrect option) is for messages processed by the function, not for events delivered to the function by Event Grid. Event Hubs Capture (incorrect option) is for Event Hubs, not Event Grid topics. Using a Service Bus queue as a dead-letter location (incorrect option) is possible, but the problem specifically asks to capture events in a blob container for inspection.