Overused... what happens?

Node Operators troubleshooting
1

it has been discussed here: How can such overusage happen? - #25 by SGC

but since then have there been any changes to try and prevent it? like starting to offload data to reduce the overuse(that we dont get paid for anyway) ?

or why not pay us for it ?

2

You are paid for the pieces the satellite expects you to store as part of the consumer-available data, whether they are part of the “overuse” or not. Where’s the problem?

2 Likes
3

prim the overuse, trash is ok’is, but overuse no…. at least that is how i feel about it

and yeah i know most dont care, but i do

4

We could prevent it by forcing the node operators to never reduce the disk allocation. Is that what you want?

1 Like
5

Yes, lot of things happened since 2022.

Earlier we have some bugs in space calculation (which showed higher overusage). Later we improved piecestore efficiency in different way to make the walkers faster. Finally hashstore was improved to address problems like slow walkers in more generic way.

It’s totally different picture.

But please note:

  1. Hashstore, by default, always has some overusage. But now you can controll it better, and choose between more IO / less overusage.
  2. It’s not as easy as just pay for any “overusage”. Because it’s hard to prove to exact amount. While the current useful used space can be calculated on server side and can be proven by audits.
1 Like