Open p4fg opened 7 months ago
sometime this call debug_resyncWorldState is not running correctly and trigger a full heal instead of having a correct snapsync. it's an issue we need to fix. I think the best is to resync from scratch with the flat healing flag. it should be good . It's not an issue with the flat db healing but more with debug_resyncWorldState
Description
As an validator, I want flat database healing to be documented to take a very long time and incurring downtime on mainnet so that i can switch beacon-node before starting healing.
Using 24.1.1, i enabled
--Xsnapsync-synchronizer-flat-db-healing-enabled=true
and followed the instructions inhttps://github.com/hyperledger/besu/pull/5319
.I called the RPC-endpoint on a fully synced besu in order to trigger the healing (as per instructions in the PR):
This resulted in the node going offline/out-of-sync while healing, which was fine, for a while. But after 24h of healing it was still not done, and no ETA is presented in the logs:
As this is now the default behaviour, it should probably be documented that this can take a long time and that downtime is expected as this probably will kick in for some users. (if this is the expected behaviour).
Acceptance Criteria
One or more of:
Steps to Reproduce (Bug)
Expected behavior:
Actual behavior:
Frequency: N/A
Logs (if a bug)
Besu
Lighthouse:
Versions (Add all that apply)
Additional Information