Closed elvece closed 2 years ago
And you said that disabling the txindex directive caused it to recover itself?
And you said that disabling the txindex directive caused it to recover itself?
Yes. Did not have to reinstall bitcoin core. Toggling txindex off allowed progress to continue. Started sync on 3/28, noticed this issue morning of 3/29, afternoon of 3/30 sync is at 87.33%.
It appears from your logs that there is a leveldb file that is corrupted in the txindex data.
Fatal LevelDB error: Corruption: block checksum mismatch: /root/.bitcoin/indexes/txindex/032765.ldb
Corruption can happen if it is aggressively killed, perhaps the solution here is to have an action wipe the txindex.
This is almost certainly a hardware issue, and nothing to do with bitcoind. Disk corruptions from I/O errors affect all data, not just txindex
yeah but unfortunately we have to be able to recover from these sorts of corruptions. We could do it on an adhoc basis but as it stands right now, if @elvece was a customer, this would be a support issue and one that would require a post-init script or ssh to resolve.
Sounds like she just turned off txindex in the config
Right but she can't turn it back on anymore which means it is "stuck" until the corruption is cleaned up. Realistically this feels like a bug in Core, but we have a limited ability to do anything about that.
Ok let's add an action to delete the txindex if it gets corrupted
Environment:
Steps to recreate:
Other notes: