On 08/13/2017 06:00 PM, Neill Miller wrote:
> Hello,
>
> Looks like the store is definitely corrupt if it wasn't shut down
> properly. By chance, did you get a crash during the sync? I ask
> because I have seen something similar twice in a row now on master
> post block 400k+ (the last one being at 441857).
>
> Anytime a crash occurs, there's a chance of store corruption,
> especially if you're not flushing writes (config setting).
>
> If you're adventurous, you can try removing the flush_lock file and
> try restarting bs, but if there was once corruption, even if it
> starts, the corruption persists and may cause future crashes and
> further corruption.

FWIW corruption is *guaranteed* if you synced any data (txs or blocks)
after starting up, and will manifest in unpredictable ways.

> The correct way to remedy this is to re-sync from scratch.
>
> Unless you *need* to use master (which is currently unstable), I
> recommend using version3, which is production/stable.
>
> -Neill.