Thread by @moo9000

1/ Syncing ETH node. So hard they say.

I did it in 24 hours. Running it costs me around 50 EUR/month.

No API limitations. No unplanned downtime. No counterparty risk with the SaaS company.

A thread.

👇👇👇

2/ Your options to run a self-hosted node are

@go_ethereum, one of the original Ethereum nodes

Turbo-Geth from @realLedgerwatch and others

Parity (but their org is more focus on Polkatdot nowadays so not a long term bet)

3/ Go Ethereum works well for a single app use case.

Turbo Geth is needed if you have massive API request requirements. E.g. if you are a wallet or really popular dApp.

4/ You can get a physical server from @Hetzner_Online very cheaply with 1 TB of storage, a price range of 50 EUR.

Equivalent Amazon t2.2xlarge is $267/month, not including storage.

5/ Of course, the difference is that your Hetzner server is not on the same control plane as your AWS/GCP/Azure you cannot easily redeploy it. It is a manual process.

6/ But I do not expect that I need to touch my go-ethereum.

It just runs. It is one of the most stable software I have run. Very boring devops wise.

7/ After you get go-ethereum running and you have a JSON-RPC API URL and you are done. No maintenance.

It would be the same getting JSON-RPC URL from Infura. They are interchangeable.

8/ One thing is that Go-ethereum does not natively support JSON-RPC HTTP endpoint authentication.

A cheap trick to get authentication is to exploit --http.vhosts and setting a server name that is not in a public DNS. Then set this in /etc/hosts file.

9/ A the real way to authenticate on your JSON-RPC node is Nginx.

nginx.com/resources/wiki…

And I am pretty sure there has to be docker-compose file that does all this for you.

10/ Funnily enough, Geth only servers public data, which an adversary could get from the Ethereum mempool any way, so there is no need to bother with HTTPS.

11/ Only times I need to poke the server is when there was is a hard fork. Which is like once a year.

But in the case of a hard fork I would need to touch my cloud-hosted managed instance as well.

12/ So how long it took to sync?

"I cannot sync my Ethereum nodes it has been weeks"

👇👇👇

13/ Less than 24 hours.

I know @go_ethereum team has optimised the sync a lot. But still, it was fast and less than I expected from seeing all the complainers in Crypto Twitter.

14/ After the sync, geth is generating state snapshot for others node to sync and this process may take some days.

But waiting the snapshot to be ready is not necessary for the node to operate.

More information about snapshots: blog.ethereum.org/2020/07/17/ask…

15/ If I need to complain about something in @go_ethereum it is that the console output is not very clear if the node is synced or not. There no estimate when the sync is complete in hours/minutes.

Or at least I did not spot one.

17/ Disk usage space?

300 GB.

(That includes some other stuff on the server, like Prysm and ETH2 chain)

18/ That being said if you want to have an Ethereum API as a service (EaaS) my recommendations are that you get it from @POKTnetwork (more or less decentralised) or @QuikNode

/ FIN

TL;DR If you feel #Ethereum node vendor pricing is high and you are constantly hitting API limits, it is not that hard to do it yourself.

Now time to get some exercise.