What is an NZB file?
Short answer
An NZB is a small XML index listing the Message-IDs of every article in a Usenet post. Your downloader reads it and asks your provider for each article over NNTP. You get NZBs from an indexer, so the NZB, the indexer and the provider are three separate things, and a provider such as Newshosting or Eweka only supplies the articles.
An NZB file tells your downloader what to fetch, but it contains none of the download itself. It is a few kilobytes of XML naming every article a post was split into. Open it in SABnzbd or NZBGet and the client collects those articles from your Usenet provider, checks them, repairs any gaps with PAR2 files and unpacks the result.
What an NZB contains
Usenet servers store a binary as thousands of small articles, each with its own unique Message-ID. The NZB sorts those IDs by file and puts them in order, so your client knows what to request and how to put the pieces back together. It is plain XML, and you can open one in a text editor and read it yourself.
| Element | What it holds | What your client does with it |
|---|---|---|
<head> / <meta> | Optional fields such as a title or a password hint | Names the job and may pass on an unpack password |
<file> | One file of the post (a RAR volume, a PAR2 file), with poster, date and subject | Creates one file in the job |
<groups> | The newsgroups the post went to | Mostly ignored now, because servers find articles by Message-ID |
<segment> | One Message-ID, its size in bytes and its position in the file | Asks the priority 0 server for that article, then lower priorities if it is missing |
There is no video, audio or archive data in the NZB. Delete it after the job has started and nothing happens, because your client has already loaded the list.
Where NZB files come from
Indexers scan Usenet headers, group articles into posts and publish an NZB for each post they recognise. A few are free, but most charge a small yearly fee for API access. Sonarr, Radarr and Prowlarr use that API to talk to the indexer and pass the NZB to your downloader without you lifting a finger. Not sure you need one? Read do I need an NZB indexer.
Normally your provider does not hand out NZBs. Newshosting, Eweka and the other unlimited accounts sell access to the articles, while the indexer sells the map that points to them. You pay for both, on two separate bills.
How the NZB and your provider work together
This is what happens when you add an NZB to SABnzbd with one main account and one block account (a fixed amount of data, used as backup):
- SABnzbd reads every
<segment>and asks the priority 0 server, for example Newshosting onnews.newshosting.comport 563, for each Message-ID. - Any article the main account lacks gets retried on the next priority, for example a NewsDemon account on UsenetExpress set as optional.
- Articles that neither server has are counted as missing in the job history.
- PAR2 repair rebuilds the missing data if enough recovery blocks came in. If not, the job fails.
That second server only helps if it is on a different backbone, because two Highwinds brands are missing the same Message-IDs.
Telling a bad NZB from a provider gap
Open the failed job in the SABnzbd history and read the detail line, then work through these in order:
- Lots of missing articles on every server: the post was taken down or never fully uploaded. Grab another NZB for the same release.
- Missing on the main account, found on the backup server: a takedown or retention gap on one backbone. This is exactly what a backup server (often called a fill server) is for.
- Every article arrived but unpacking fails: there is a password, or the post was mislabelled. The NZB pointed at the right articles, but the content inside was wrong.
- An old post missing everywhere: compare the post date with your provider's retention, how long it keeps files. Newshosting lists 6,618 days in our data, Eweka 6,615.
Fix incomplete Usenet downloads has the complete list of fixes.
FAQ
- Is an NZB the same as a torrent file?
- They do a similar job. A torrent points at peers who share pieces, while an NZB points at articles stored on your provider's servers, so your speed depends on the provider and your line, and you upload nothing.
- Can I open an NZB without SABnzbd?
- Yes. NZBGet and most binary newsreaders import NZB files. With Easynews you can also search in the browser, but it still pulls the same articles from its backbone.
- Why do two NZBs for the same release behave differently?
- They may point at different uploads. One upload can lose articles to a takedown while a later repost is still whole.