run.sh: add snapshot destination in tar cmd

[CharlesJUDITH]

When a SNAPSHOT_URL is specified, the content is stored in $PROJECT_ROOT/data/data. For example: -e SNAPSHOT_URL=https://snapshots.polkachu.com/snapshots/akash/akash_10400871.tar.lz4 And the earliest and latest block time are "earliest_block_time": "1970-01-01T00:00:00Z".

run.sh is deleting the current data directory, with the destitation in the tar command, the snapshot is stored in $PROJECT_ROOT/data and the block data are correct.

[CharlesJUDITH]

How to reproduce:

docker run -e CHAIN_JSON=https://raw.githubusercontent.com/akash-network/net/master/mainnet/meta.json -e FASTSYNC_VERSION=v0  -e MONIKER=chuck-dev -e P2P_POLKACHU=1 -e STATESYNC_POLKACHU=1 -e MINIMUM_GAS_PRICES=0.025uakt -e SNAPSHOT_URL=https://snapshots.polkachu.com/snapshots/akash/akash_10400871.tar.lz4 -e DOWNLOAD_SNAPSHOT=1 -p 1317:1317 -p 26657:26657 -p 9090:9090 ghcr.io/akash-network/cosmos-omnibus:v0.3.27-akash-v0.20.0

[tombeynon]

Thanks for this @CharlesJUDITH! There is a reason for this behaviour - previously snapshots would often contain the data directory contents directly, rather than nesting a data folder. Since this can happen (although less common these days with the addition of the wasm directory) we need to extract into a known directory first and then move the directories as required.

You'll notice a few lines later we use SNAPSHOT_DATA_PATH and SNAPSHOT_WASM_PATH to control extracting of subdirectories correctly. There is an open issue #378 that suggests identifying the subdirectories automatically; that would be a better solution here if you're up for it? Should be pretty simple, I just haven't had a chance to look yet.

I'll close this for now but please reopen if required, and thank you again for the contribution!