A share can hold millions of files and tens of terabytes. At that scale a few things behave differently, and knowing them is the difference between taking four files and accidentally starting an 18 TB download.
Read Download a share first for the basics.
Always import index-only first
On the Download page, Import (Index Only) fetches the catalogue and no content. It costs almost nothing regardless of share size.
Import & Download starts downloading the whole share immediately.
On a small share the difference is minutes. On a large one it is the difference between browsing a catalogue and committing to weeks of transfer. Import index-only unless you have already decided you want all of it.
The tree appears before it is finished
Very large shares build their browsable tree in the background. You get the structure straight away, and the app shows a panel with progress while the rest fills in.
Two things you can do while you wait:
Use the other tabs. Verify, Usenet Drive and the share's settings all work during this.
Start downloading anyway. The panel offers Start downloading everything now if you already know you want the lot.
If a listing looks short, check that panel before concluding files are missing — it is very likely still working.
Selecting part of a share
On the Files tab, tick files and folders and press Download selected (N).
Ticking a folder takes everything under it, at any depth. For a season of a show or one platform's files, tick the folder rather than the files.
Your selection is remembered. Leave and come back and the button reads Download N selected, so an interrupted browse does not lose your choices.
Segments are not selectable. A row that looks like an internal segment cannot be downloaded on its own — segments are fetched by downloading the file they belong to. The interface says so where it comes up.
Confirm the scope before you walk away
Once running, the download card shows a scope badge saying what is being fetched — your selection, or the whole share.
Look at it. It is the one place that distinguishes "12 selected files" from "everything", and it is worth two seconds before leaving a transfer running overnight.
If a selective download reports the whole share, cancel it, re-tick your selection and start again. A selection that resolves to nothing is refused with an error rather than being treated as "everything" — but check the badge regardless.
One download at a time
Downloads run one at a time. Start another and it waits, showing Queued — it has not failed.
This is deliberate. Two large downloads sharing one provider's connection limit finish later on average than the same two run in sequence, and they fight over your disk as well.
Queue shows what is running and what is waiting. Cancelling a queued download just removes it from the queue.
Stopping something you started
| Pause | Stops now, keeps everything, resumes exactly where it stopped |
| Cancel | Stops for good, keeps everything already downloaded |
Both work on a running download.
Cancel will not restart itself. The recovery that resumes work after a crash leaves cancelled downloads alone, deliberately — so cancel means cancelled, even across a restart.
Nothing is thrown away. Files already downloaded stay on disk. Download again later reuses all of them and fetches only what is missing.
If you started the wrong thing: press Cancel. Do not delete the share to stop it — cancel is the intended control and it keeps your files.
Percentages during a big download
Progress is measured in bytes, not files, which matters when one file is enormous: a share containing a single 2 TB file would otherwise sit at "0 of 1 files" for a day.
The percentage holds at 99% until everything really is complete, including the final integrity pass. A download is finished when it says so, not when the number looks finished.
Disk space
Check before you start, not after.
Storage page → Model a download (TB) → Check space. It compares against free space on your destination and tells you whether it fits.
If you run out mid-download it pauses rather than corrupting anything, and resumes once you free space.
Prefer a local disk. Network destinations work, but on a large transfer the destination is very often the bottleneck rather than your internet connection, and the final read-back pass is limited by read speed too. The speed test measures your destination separately so you can see this rather than guess: When something is wrong.
Very large individual files
A single multi-terabyte file inside a share is handled specially, and there is one setting worth knowing about.
Settings → Performance → Final check of very large files:
| Option | Behaviour |
|---|---|
| Automatic (recommended) | Chooses per file by size |
| Check during the download | Verifies as it goes; no separate pass at the end |
| Check after the download | One pass at the end |
| Skip the final check | No read-back pass |
Leave it on Automatic. If a final pass on a huge file over a slow network destination is painful, Check during the download spreads that cost instead of adding it at the end.
Skip the final check is not recommended: every piece is still verified as it arrives from Usenet, so you are not unprotected, but you lose the confirmation that what your disk wrote back is what was published — which is exactly the failure a flaky drive or a dropping network share produces.
Streaming beats downloading at this scale
For a very large share, ask whether you need the bytes on your disk at all.
A mounted share behaves like a normal folder and fetches only what programs actually read. A 20 TB media library is usable this way on a machine with a 128 GB disk, because the player reads only the file it streams.
- Mount a share as a drive
- Browse inside a disk image — three files out of a 512 GB image, without the image
Next
- Verify and repair — proving a large download is intact
- Recommended settings
- When something is wrong