Negative directory sizes

[subdavis]

After a series of updates, I've ended up with negative directory sizes like this.

I've seen it before. My hypothesis is that it's due to an imbalance of size decrements on deletion, either because deletion is run twice in a race condition or possibly because some allocations are failing to increment. I have no advice for solving it. There is an endpoint to run a database check / repair, but I'm not sure if I personally trust it not to break things worse; maybe others have more confidence in it.

-- Bryon H

I've run PUT system/check to fix size reporting issues a few times. It has always succeeded in doing so.

-- David M

I can't reproduce this, but someone may find this useful in the future.