
What This Health Check Tests
The validator tells you whether a feed follows the specification. This check looks at everything around it: how the server delivers the feed and whether the content is alive. A feed can be perfectly valid and still be slow, impossible to cache or quietly abandoned. We fetch the feed the way a reader does, then fetch it a second time to see whether the server supports conditional requests.
| Group | Checks | Why it matters |
|---|---|---|
| Connection | HTTP status, redirects, HTTPS, response time, compression, content type | A reader with hundreds of feeds gives each one a few seconds. Slow, redirecting or uncompressed feeds are fetched less reliably. |
| Caching | ETag, Last-Modified, a real 304 test, Cache-Control, WebSub | Most checks find nothing new. With caching the answer is a few hundred bytes instead of the whole feed. |
| Content | Format, size, item IDs, dates | Missing or repeated IDs cause duplicates; missing dates break sorting. |
| Freshness | Age of the newest item, publishing rhythm | An abandoned feed still loads, but nobody needs it in their reader. |
The score uses the same four parts as the health scores in our directory of feed sources: availability, freshness, regularity and technical quality. The difference is that a one-off check can only measure availability once, while directory profiles use 30 days of daily checks. How the score is calculated.
Fixing the Most Common Problems
Redirects
Feeds that moved often keep working through a redirect, sometimes two or three. That is fine for visitors, but every reader pays for it on every check, forever. If you publish the feed, update the <link rel="alternate"> tag and the subscribe links on your site to the final address. If you follow the feed, replace it in your reader with the address the report shows.
No compression
Feeds are text and compress extremely well: a 300 KB feed often travels as 40 KB. On Apache, enable mod_deflate for application/rss+xml, application/atom+xml and application/feed+json; on nginx, add those types to gzip_types. Many servers compress HTML by default but forget feed types.
No ETag or Last-Modified
Without these headers a reader cannot ask whether the feed changed, so it downloads everything on every check. Static feeds get them from the web server automatically. Feeds generated by a script need them added: send a Last-Modified based on the newest item, compare it with the If-Modified-Since header of the request and answer 304 Not Modified without a body when nothing changed. The publisher best practices include a code example.
The 304 test fails
Some servers send an ETag but answer every request with a full 200 response anyway. Common causes are an ETag that changes on every request because it includes a timestamp, a CDN that strips the conditional headers, or compression that alters the ETag between requests. Check what your server sends with two requests in a row: the ETag should be identical.
Missing or duplicate item IDs
Readers recognise items by their guid (RSS) or id (Atom and JSON Feed). Without one they use the link or title, and every edit then shows up as a new post. Give each item a permanent ID that never changes, not even when you move to HTTPS or a new domain. Why feeds show duplicate posts explains the details.
Questions About Feed Health
What does HTTP 304 mean for an RSS feed?
304 Not Modified is the answer a server gives when a reader asks for the feed with an If-None-Match or If-Modified-Since header and nothing changed since the last visit. The response has no body, so it takes a fraction of the time and bandwidth of a full download. For a feed that is checked every 15 minutes but updated twice a day, more than 95 percent of all requests can be answered with a 304. It is the single most effective way to make a feed cheaper to serve and faster to check.
What is the difference between ETag and Last-Modified?
Both are validators: values a reader stores and sends back on the next request. Last-Modified is a date, usually of the newest change. An ETag is an opaque fingerprint of the content, often a hash. An ETag is more precise because it changes with any change, also within the same second, while a date is easier to generate from a database. You only need one of the two; the check marks a missing ETag as informational when Last-Modified is present.
Why does a feed return 403 or 429 here but work in my reader?
Many sites run bot protection that blocks or rate-limits requests from data centres, where our checker runs, while home connections pass. A 403 means the request was refused, a 429 that too many requests came in. Cloud-based readers such as Feedly and Inoreader run into the same blocks, so if you publish the feed, allow known feed readers through your firewall or CDN rules. If you only follow the feed, a desktop or mobile reader that fetches from your own connection usually still works.
Why is my feed slow?
The usual causes are feeds generated on every request with a heavy database query, very large feeds with full articles and hundreds of items, and missing compression. Cache the generated feed for a few minutes, limit it to the newest 20 to 50 items, and enable compression. If the feed goes through a CDN, make sure the CDN caches it instead of passing every request to your server.
Is my feed data stored?
We keep the result of a check for two minutes so that sharing the report link does not fetch your feed again. Feeds that are part of our directory are checked daily and have a public profile with a 30-day history; other feeds are not stored.
Sources and Further Reading
- RFC 9110: HTTP SemanticsIETF
- RFC 9111: HTTP CachingIETF
- RSS Best Practices ProfileRSS Advisory Board