Purpose
Whether your data is staying healthy on the network: the background repair process
that re-shards data away from failed providers, plus loss counters and the
statistical failure probability.
Panels
- Status banner — e.g. “Repair in progress · started 04:53 PM — figures below
are live.” - Repair Progress — Total Rows · Rows Repaired · Rows Failed. (A “row” is the
erasure-coded unit your files are sharded into.) - Lost Rows — Lost Rows · Confirmed Lost. Anything above zero deserves
attention; “confirmed” is the serious one. - Failure Probability — computed probability of data loss, with a “Below
threshold (1e-12)” marker when healthy.
Reading it
- Repair running is normal — providers churn; repair is the system doing its job.
- Rows Failed ticking up repeatedly across passes, or any Confirmed Lost, is a
report-it-now situation: capture the screen and raise it with support/Discord. - Failure Probability above threshold usually accompanies a shrunken matched-host
set — check Storage & Placement.
Screenshots
