Skip to content

No retry logic for transient ERCOT API failures in the downloader #37

@haydenk

Description

@haydenk

Problem

The downloader makes HTTP requests to the ERCOT document list API and download endpoint with no retry logic. A transient 5xx response, TCP reset, or DNS hiccup causes the entire run to fail. Since the downloader is invoked by cron, a single transient failure means that day's files are not downloaded until the next scheduled run.

Expected behavior

Wrap ERCOT HTTP calls with exponential backoff retry (e.g., 3 attempts, starting at 1s). Only persist failure to esiid_documents after all retries are exhausted. Use a crate like backoff or tokio-retry. Network errors and 5xx responses should be retried; 4xx responses (except 429) should fail immediately.

Metadata

Metadata

Assignees

No one assigned

    Labels

    enhancementNew feature or request

    Projects

    No projects

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions