Yaci-store mithril snapshot

[Kammerlo]

Why

Currently the syncing takes around a day (or more depending on the underlying hardware). The Node can be setup in round about an hour by using a mithril snapshot. The rest of the needed syncing times comes from the indexer.
Currently there is no possibility to replay a snapshot for the indexer database, therefore a normal sync is needed

What

Yaci-store is a modular indexer, therefore it is hard to provide a snapshot, since users are intended to change the behaviour for their needs. We did exactly this for rosetta-java and we have fixed database schema for each release. We need to provide snapshots for each release for mainnet to enable users to setup their environment as fast as possible.

[linconvidal]

Hey @Kammerlo! I actually tried implementing a snapshot restore process last year (yaci-snapshot branch), but I ran into a few challenges, mostly around schema handling, database ownership, and ensuring the restore process worked seamlessly inside a container.

One issue I had was ensuring the correct schema was set up before restoring. At times, tables were missing or permissions needed to be adjusted manually. I also had to deal with PostgreSQL authentication issues when running the restore inside the container.

Sharing this in case any of it is useful. 🚀

[Kammerlo]

Definitely, when we pick this feature up we will take this into account!
Currently I'm just trying to drop the feature requests I have collected offline. Thanks in advance!