graphman commands to fetch / clear the call cache

[azf20]

Currently if indexers want to inspect & maybe clear the Ethereum call cache, they need to access and alter Postgres directly. This is error prone. There could be a dedicated graphman command to streamline this process:

graphman chain cache-get <chain> <contract_address> [--block <number>|--from <number> --to <number>] to print what’s in the call cache
graphman chain cache-remove <same args as cache_get>

Example existing process, when the call cache is found to be malformed:

# graphman rewind

# prerequisites
have graph nodes at least v0.23.1 and configured with the config toml files

# stop index node containers, keep query node
docker stop index-node-0 index-node-1 indexer-agent indexer-service

# connect to query-node-0 container
./shell query-node-0

# rewind subgraphs
graphman rewind <IPFS_hash> <block_hash> <block_number>


# delete eth call cache

# connect to postgres container
./shell postgres

# connect to your graph-node database
psql <db_name> <db_user>

# find ethereum chain namespace in the db
select * from chains;

# navigate to the namespace
set search_path to chain1;

# find latest block number in the call_cache table
select block_number from call_cache order by block_number desc limit 1;

# delete eth_call_cache to make sure you don't write the same identical data

DELETE FROM call_cache where block_number between X and Y;

or

DELETE FROM `call_cache` WHERE block_number >=X and id<= Y;

[neysofu]

Closed by #4066.