My node is synced and validating new blocks normally, but a few old blocks can’t be read back. getblockstats fails for six consecutive heights with:
error code: -1
error message:
Can't read undo data from disk
and debug.log repeats:
[error] ReadBlockUndo: Checksum mismatch at FlatFilePos(nFile=5123, nPos=1035211)
Anything that needs those blocks’ undo data fails, including getblock at verbosity 3 and the mempool.space backend’s block indexer, which retries forever.
The usual advice is -reindex, which takes days on mainnet. Is there a way to repair just the damaged undo records? I have -txindex enabled.
