Open tropicar opened 1 year ago
I've tested pruning on a seperate machine outside of dappnode with nethermind.
I think there should be some documentation to accompany this.
The nethermind docs are not great for this pruning module, they have acknowledged this on their discord.
Some points from my testing:
Re syncing from scratch is faster at the moment. It seems traversing beach node to mirror to a new state db is still quite slow in nethermind. Even in dappnode gen 10 hardware.
2.there is not a good UX in the logging to determine how far along an operator is in the pruning process
There isn't a way to determine how many nodes need mirroring before the process is started so that can't be used as a determination of how long there is to go.
A better explanation than is present on the nethermind docs for the following would save many support requests in dappnode discord: A. What is the different between pruning modes, what is generally viewed as optimal for new setups or existing setups B. How much space does a dappnode need free to successfully prune C. How much and what does more cache ram actually do when it's set in hydrid mode
If the process fails, users need to know to start pruning from the beginning.
Realize this is a quite old PR, but was just checking the Nethermind pruning docs and noticed the Hybrid
pruning mode. Wouldn't that be the better choice compared to Full
pruning mode?
According to Nethermind docs:
Setting Pruning.Mode=Hybrid enables both InMemory and Full modes. InMemory mode helps the node grow at a slower rate compared to Pruning.Mode=Full. With this configuration, Full Pruning is executed less frequently, promoting healthier disk operation. Since Full Pruning is hardware-intensive, this configuration is also beneficial for attestation results.
Please provide this feature. A lot of dappnode users switched to nethermind and everyone will encounter this problem at some point.
Anything blocking this?
This PR is to make user-friendly the process of using the prunning mode of nethermind client.