Full nodes, witness nodes, decentralization, and some questions

[karnal]

Full nodes expose an API and thus need the history plugins.
p2p2-consensus-nodes only store the blockchain and verify it

Got it, thanks.

Yes .. plenty of discussion between devs .. and plenty of ways to address it

Any ETAs?

[fluxer555]

My full node runs at 16.5GB RAM now. (It's on the 96GB RAM dedicated server)

Is this using the latest core release?

Yes, it was at 16.9GB RAM before the update.

Have you updated? What is it now?

[xeroc]

In your reply you introduce a new term (for me), "p2p-consensus-nodes" .. how do these differ from full nodes?

Full nodes expose an API and thus need the history plugins.
p2p2-consensus-nodes only store the blockchain and verify it

Regarding the huge memory usage on full nodes, is the issue being actively addressed? Surely there is no need to keep ALL of that information available in RAM, it could be fetched from disk on demand and stay in memory for a certain amount of time based on how often it's fetched, and all such data should of course have a maximum amount of RAM (configurable by the operator) that it can use - that would keep things more predictable for operators.

Yes .. plenty of discussion between devs .. and plenty of ways to address it

[karnal]

Businesses that run on top of BitShares have incentive to run their own nodes (think exchanges)
The memory mostly grows due to the market history and account history plugin. Those are only required for full-nodes, not for block producers, nor for p2p-consensus-nodes.

Correct, guess I should have been a bit more precise in my post, but: (mostly) big business hosting nodes doesn't help that much with decentralization.

In your reply you introduce a new term (for me), "p2p-consensus-nodes" .. how do these differ from full nodes?


Regarding the huge memory usage on full nodes, is the issue being actively addressed? Surely there is no need to keep ALL of that information available in RAM, it could be fetched from disk on demand and stay in memory for a certain amount of time based on how often it's fetched, and all such data should of course have a maximum amount of RAM (configurable by the operator) that it can use - that would keep things more predictable for operators.

[sahkan]

My full node runs at 16.5GB RAM now. (It's on the 96GB RAM dedicated server)

Is this using the latest core release?

Yes, it was at 16.9GB RAM before the update.

[fluxer555]

My full node runs at 16.5GB RAM now. (It's on the 96GB RAM dedicated server)

Is this using the latest core release?

[sahkan]

My full node runs at 16.5GB RAM now. (It's on the 96GB RAM dedicated server)

[xeroc]

Businesses that run on top of BitShares have incentive to run their own nodes (think exchanges)
The memory mostly grows due to the market history and account history plugin. Those are only required for full-nodes, not for block producers, nor for p2p-consensus-nodes.

[karnal]

There does not appear to be any incentive to run a full node (or indeed, a (lowly ranked) standby witness).

From the bits and bobs I remember, memory usage on witness/full nodes is still rising linearly, altough there was some sort of patch around to drastically reduce this?


Some of the documentation talks about the benefits of running a full node for decentralization - correct assumption that running an actual (idle) witness for the same purpose accomplishes exactly the same?


And finally, could a witness operator comment on the require hardware to run a witness/full node ?