I don’t understand why there needs to be 15 bitcoind threads all reading multiple megabytes/sec from the drive, all the time. Is there some way to curb this? I am running electrs and mempool, but I don’t think it used to be this way before I upgraded to start9 4.0 (currently running beta8, but betas 5-7 showed the same behavior).
Bitcoin Core 29.3:10
Electrs 0.11.1:3
Mempool 3.3.1:2
Killing electrs and mempool stops this behavior. I am going to switch to fulcrum for mempool indexer, supposedly that should help quite a lot. Will report when switched over.
Alright, so electrs wasn’t actually the issue. Or if it was, fulcrum completely solved that part. The actual issue is that mempool keeps querying bitcoind for transaction data instead of just using fulcrum, and forces it to read old blocks from the disk. I think I’m just going to stop running mempool since it’s retarded. I still think it didn’t use to do this before. I might try uninstalling and installing an old version of mempool.
Is it possible you have the indexer disabled in the Mempool config?
It needs to be pointed at Fulcrum.
It is pointed at fulcrum. I tried uninstalling and reinstalling mempool, seems to behave a bit better (maybe had some db corruption?), but still not within acceptable ranges as it likes to pull old blocks from bitcoind directly for some unknown reason. Resorting to running it only when I need it.
That’s a common strategy.
This is probably due to mempool.space. It requests (and indexes) all the blocks in the background, which is probably not finished yet. It can take quite a while. I don’t remember if you can see somewhere if it was fully indexed or not. Even after mempool.space is done indexing, I found it uses too much resources even when idling, so personally don’t have it running anymore.
Indeed. It’s finished indexing a couple hundred thousand blocks for the mining stats. Still consuming excessive resources after that calmed down.
