Check, repair, and unpack

The pipeline

Health check during download. If so many articles are missing that repair can't succeed, HealthCheck acts before you waste more bandwidth.
Rename. Par-rename and RAR-rename restore the real filenames of obfuscated postings.
Verify. Quick verification checks only the files that had download errors.
Repair. Multithreaded par2 repair rebuilds damaged files.
Unpack. RAR and 7-Zip archives are extracted. Passwords come from the NZB file or from a password file.
Clean up and run extensions. Archives and par2 files are deleted, and post-processing extensions run in the order you set.

PostStrategy controls how many downloads can be in these steps at once: sequential, balanced (the default), aggressive, or rocket.

Health check

Health is the share of a download's articles that the news servers actually delivered. Critical health is the lowest health that the download's par2 recovery data can still repair; RustyGet computes it from the number and size of the par2 files.

When health drops below critical health, HealthCheck decides what happens:

Value	Effect
`park`	Default. Moves the download to history with its files kept, so you can add more par2 files or retry later.
`delete`	Deletes the download and moves it to history. With duplicate checks on, the best backup of the title starts.
`pause`	Pauses the download. You start a backup yourself.
`none`	Keeps downloading. Use it when you plan to repair by hand.
`dupe` new	Waits for a duplicate first. If a better copy of the same title waits in history, the download fails over to it; if article fallback can borrow the missing articles, the download keeps going. See Duplicate fleets.

Renaming obfuscated files

Many postings hide their real names, for example a 32-character hexadecimal base name with extensions such as .01, .21, and .par2. Archives with names like these can't be unpacked until the real names come back.

Par-rename: The par2 set stores the original filenames. RustyGet reads them and renames the files in seconds, with no verification pass. The classic approach, running par2cmdline over every file, can take tens of minutes on a large download. Set by ParRename.
RAR-rename: Restores the names of multi-volume RAR archives from the archive headers. It works when a posting has no par2 files, and when the files were obfuscated before the par2 files were made. Set by RarRename. RustyGet parses RAR headers and renames volume sets in Rust.
Direct rename: Renames files while they download, as soon as the first article of each par2 file arrives. Set by DirectRename.

Quick verification

Par2 verification reads and checksums every file, which takes minutes to tens of minutes. On a slow device with a large, lightly damaged download, verification can take longer than the repair itself.

RustyGet doesn't need that pass. It records which files had download errors and hands exactly those to the repair module, so repair starts right away. ParQuick controls this, and ParScan sets which files take part:

`ParScan`	Files scanned
`limited`	Only files that belong to the par2 set.
`extended`	Default. Files of the set first, then other files, until enough recovery blocks are found.
`full`	Every file in the download directory.
`dupe`	Like `extended`, and also borrows par2 blocks from duplicate downloads of the same title in history.

Multicore repair

par2cmdline predates multicore CPUs and uses one core. A popular multicore fork depends on an Intel library and runs only on Intel-compatible CPUs, which leaves out the ARM chips in most NAS boxes and media players.

The repair module has its own threading that works on every platform and CPU. On a four-core ARM device it repairs two to three times faster than par2cmdline. Set the thread count with ParThreads and the memory with ParBuffer.

Unpack

Unpack extracts RAR archives with UnrarCmd and 7-Zip archives with SevenZipCmd. DirectUnpack starts extracting while the download is still running, which saves time when the download needs no repair.

UseTempUnpackDir extracts into a temporary _unpack directory and moves the files when done.
UnpackPassFile points to a file of passwords to try.
UnpackCleanupDisk deletes archives after a successful unpack, and ExtCleanupDisk deletes other leftovers, such as .par2 and .sfv files.
UnpackIgnoreExt leaves some archives alone, for example .cbr comic books.

Retry failed articles

A failed download goes to history, where you can download it again. Besides starting over, you can retry only the articles that failed. That keeps everything already downloaded and fetches just the gaps. It helps when:

A posting hadn't finished propagating to your servers.
An optional news server was unreachable during the first attempt.
You added or turned on another news server.

Post-processing extensions

Extensions are scripts that run at set points, such as after a download finishes. Par check, repair, and unpack are built in; extensions handle everything else, such as deleting samples, sending notifications, or moving files for media managers.

Install extensions from the Extension Manager in the web interface, or put them in ScriptDir.
A current extension is a folder with a manifest.json and an executable. Older extensions are single script files with an NZBGet header comment block, and they still load. Either way, an extension's options are saved in nzbget.conf and passed as NZBPO_ environment variables.
Post-processing extensions read the download's state from NZBPP_ variables such as NZBPP_DIRECTORY, NZBPP_TOTALSTATUS, and NZBPP_STATUS.
Exit with 93 for success, 94 for failure, or 95 to skip. Any other exit code, including 0, marks the script as failed. Exit 92 to ask for a par check after an extension unpacked the files itself.
Print [NZB] DIRECTORY=<path> to tell later scripts that the files moved, or [NZB] FINALDIR=<path> when you move them into a directory that holds other files.

The extension interface is inherited from NZBGet, and NZBGet extensions work unchanged today. The compatibility policy explains what may change.