Each checkpoint compares two totals from the same node: the counter it signed for the checkpoint, and the running total from its latest live traffic report. The diff is how far the checkpoint is above the live total, as a percentage.
How to read it
- Under 20%: accepted. Small gaps are normal, because the checkpoint and the live report are taken at slightly different moments.
- 20% or more: not accepted. Three checkpoints in a row at 20% or more raise a warning flag on the node.
- A checkpoint below the live total shows 0%: only a checkpoint that runs ahead counts.
What it tells you
Because both numbers come from the node, a steady diff only means the node's two reports agree with each other. It cannot catch a node that inflates both. Accuracy is tested by the probes and cross-checks described in How nodes are checked.
Where it counts
- A node's average diff over 7 days is 10% of its health score.
- Checkpoint history is also checked for counters that go backwards (a critical flag), repeated sharp drops, and a node that stops notarizing (warnings).