Your collection file is too large to send to AnkiWeb

It seems the size of my collections has grown too large and my collection has gone over the ankiweb maximum file size. I have tried most of the ideas I’ve read about reducing file size, but still am at a loss. I am not 100% sure how this has happened, but I do have a few ideas.

The specific error I have when syncing is:

Your collection file is too large to send to AnkiWeb. You can reduce its size by removing any unwanted decks (optionally exporting them first), and then using Check Database to shrink the file size down.

⁨387010560 > 314572800⁩ (uncompressed)

I have done the following:

  • Check Media → Deleted Unused → Empty Trash
  • Check Database
  • Export All Decks as .apkg (including scheduling, presets, media, excluding older version support), create new profile, import on new profile.
  • FSRS Helper → Delete redundant manual revlog entries

However, the collection size is still too large, none of the above had a significant impact.

Checking the ~/Library/ApplicationSupport/Anki2/backups folder, all of the backup sizes are around 9MB-16MB in size, nowhere near the 300+MB that the sync error dialog is showing. (There was a large jump from 9MB to 16MB this morning when I added ~6000 new notes this morning, which probably contributed to me going over the limit. Technically speaking I could subtract them for the time being, but this problem will just repeat itself as I do intend on adding far more than 6k new cards in the next few months.) (I’ve been averaging about 230 (!!!) new cards per day for the past 2 months, so I would very quickly run into the problem again within a month.)

I suspect that this issue may be related to me using FSRS Helper to to reschedule all the cards too often, in combination with the ~15k cards currently in the collection. (I recently made huge changes to lowering DR and some times I’d do 300+ reviews at ~50% accuracy, lower than the 70% DR, so I felt the need to reset FSRS and reschedule all cards more often than most other users would be doing it.)

I also suspect that this issue may be related to me importing large numbers of cards through yomitan+AnkiConnect, defaulting to storing information and not to discarding it (for future-proofing), which includes large amounts of superfluous HTML per each card, which… might add up over 10k+ cards? (I figured the superfluous HTML would not compare to the audio data so figured it was “small” enough, but maybe that assumption was incorrect.)

There’s lots of audio media on the cards, but I’ve read that such information would not be applicable in this situation.

Any advice about how to proceed would be greatly appreciated.

1 Like

After additional testing, I’ve discovered the issue was indeed the superfluous HTML information in the cards imported through yomitan. It came up to 276MB uncompressed data for those 6k cards, plus another 100+MB from another ~2000 similarly imported cards that had been done prior.

Pruning the data in those 6k cards down to remove the HTML and media references, it came down to 45MB, 16% of the original size. After removing redundant definitions, etc, I was able to get it down to a few MB.

The problem’s resolved. I’m leaving this post up for anybody else who runs into a similar problem–do be careful about importing data from yomitan, esp. if you were planning on importing 10k+ cards through that method (as I had been intending, and still intend to do). Avoid superfluous HTML. Avoid redundant data. Only import what you actually need.

(This is also likely why backups were only ~16MB, as it was likely compressed and thus the redundant data was greatly compressed.)

2 Likes

I’m glad you worked it out. Yes, excess HTML is almost always the problem. See: Constant Errors Uploading / Sync - #4 by Danika_Dakika .

That won’t ever be the issue because rescheduling through FSRS Helper doesn’t add to the size of the revlog.

That’s correct. Media doesn’t count toward the collection-size limit. Are there limits on file sizes on AnkiWeb? - Anki FAQs