host chain artifacts like archives, reindexer dbs, and genesis files

[conorsch]

Is your feature request related to a problem? Please describe. Reindexing workflows depend on access to hard-to-find data, like the original genesis file for a pre-upgrade chain. Also, newly joining nodes must bootstrap from node state backup.

Describe the solution you'd like There should be a URL like https://artifacts.plinfra.net/ that hosts a directory tree of useful data, separated by chain.

Describe alternatives you've considered Right now we just encourage folks to find the data on their own, like in the pindexer docs:

The <ORIGINAL_GENESIS_FILE> must be the genesis.json file for the chain, before any upgrades. Note, in particular, that after an upgrade, there will be a new genesis file containing only a checkpoint. pindexer specifically needs the original genesis file, because it needs to read information about the start of the chain, such as initial allocations, to track changes over time accurately.

Additional context Here's a list of artifact types for consideration:

• pre-upgrade node state, sanitized (i.e. node key material removed)
• post-upgrade node state, sanitized (i.e. node key material removed)
• reindexer_archive.bin for storing reindexer sqlite3 dbs
• genesis.json files; especially important for chains that have upgraded
• db dumps of event data? ideally not, since the above is sufficient to regenerate the rest, but might be useful as a reference

We don't need to go crazy about completeness, just having a directory to point to would greatly improve a lot of first-run experience.

[conorsch]

Started collecting this data here: https://artifacts.plinfra.net/ It's got subdirs for chains like penumbra-1 and penumbra-testnet-phobos-2. Not 100% complete, but it's a great start. Will continue to add more assets.

One thing we've lost track on is the ability for a new node to bootstrap onto a post-upgrade chain easily. We should revisit the --archive-url logic in pd to make that a bit easier.