The short version

The bandwidth figure on your node counts the bytes your node carried. Your earnings count the bytes we could confirm your node carried. Those are usually the same number, and when they aren’t, the difference is always a session we couldn’t settle cleanly.

How a Bandwidth Sharing session is settled

Bandwidth Sharing is different from the other protocols. A ping or a DNS lookup produces one result from one node, but a proxy session has two participants: your node and the customer using it.

At the end of a session, both sides independently report how many bytes passed, and both sign their report. The hub waits for the pair, compares them, and settles the session from what the two reports agree on. Neither side can inflate its own figure, because the other side’s report has to corroborate it.

This is what protects honest operators. Without it, anyone could point a script at the network and claim terabytes they never carried, and the earnings pool would drain to whoever lied hardest.

When the two numbers drift apart

Only one report arrived. The hub waits an hour for the matching report. If your node’s report arrives and the customer’s never does, or vice versa, there is nothing to corroborate the claim, so the session isn’t credited. The bytes still show as carried on your node, because your node did carry them.

This is the most common cause, and it’s usually a node that went offline, restarted, or lost its connection to the hub before its report was delivered.

The two reports disagree. If your node and the customer report materially different byte counts, we can’t tell which side is wrong, so the session settles below both claims. Persistent disagreement also costs reputation, which affects how much work gets routed to you.

The reports failed their integrity check. Bad signatures or mismatched session hashes mean the pair can’t be trusted at all. This should never happen on a healthy node running a stock build.

What you can do about it

Almost all of the gap comes down to your node being reachable long enough to deliver its half of the report.

• Keep the node running. A node that restarts mid-session may never deliver its report. If yours is restarting on its own, see Troubleshooting Connectivity.
• Keep the connection stable. Dropping off the network between the session ending and the report being delivered has the same effect as being offline.
• Stay on a current release. Report signing and settlement are part of the protocol, and old builds can fall behind it.

Proxy sessions and your job count

Bandwidth Sharing earns on bytes, not on jobs, so proxy sessions don’t move your job count. A node serving a lot of proxy traffic and few probes can show a low job count and healthy earnings at the same time. That’s expected, not a fault.

Still doesn’t add up?

If your earnings look wrong on a node that has been online and stable, contact [email protected] with your node’s peer ID and roughly when the sessions ran, and we can look at how those particular sessions settled.