Adrian Sutton

A simpler alternative to getting a native sha256 implementation may be Amazon Corretto Crypto Provider: https://github.com/corretto/amazon-corretto-crypto-provider It only provides Linux/amd bindings currently (Linux/aarch is experimental currently) but that's a lot...

Worth noting here that the most common causes of this are 1) having a -Xmx set too low which causes Java to spend most of it's time running GC (-Xmx5g...

I'd be somewhat cautious about starting from a fresh checkpoint sync for a couple of reasons. It will need to download the historic blocks in the background which increases disk...

@back-87 If you really want to live dangerously, and since you're already running HEAD, we have been working on an optimisation to take database updates off of the critical path...

Great to hear. Thank you for your help testing. We've been seeing good results on our nodes too and have a couple of other improvements we've just merged as well....

Yep the latest head has that arg defaulted to on (arg still exists so it can be toggled off if need be but we'll remove it soon). Thanks for your...

We normally unsubscribe from all gossip channels when we're far enough behind the chain head and sync committee gossip would be included in that. Maybe that cut off is too...

This would also potentially set things up well to not store the execution payload in Teku's database but retrieve it from the EL when required. Probably not the initial goal...

Ah, it took me a bit to work out what the actual problem here was. It looks like we're sending a fCU with a pre-merge beacon block and the terminal...

> but what if, say ttd = 1000 > > ``` > pre-terminal (999) - terminal 1 (1001) > \- terminal 2 (1002) - arrives a bit later > ```...