A genesis.json .initial_height causes rly to query nonexistent blocks

[michaelfig]

Here's a report for:

# rly version
version: 2.5.2
commit: 250cb55eb164733539846c45fbff6fc97a135361
cosmos-sdk: v0.50.4
go: go1.21.5 linux/amd64

When I start a new chain with genesis.json having .initial_height: 17, the relayer tries (and fails) to query for blocks 1-16, retrying several times on each block, until it finally finds the first block produced by the chain.

Here are some of the logs:

relayer      | 2024-07-05T22:12:39.364126Z      warn    Reached max retries querying for block, skipping       {"chain_name": "agoriclocal", "chain_id": "agoriclocal", "height": 15}
relayer      | 2024-07-05T22:12:40.360969Z      info    Chain is not yet in sync      {"chain_name": "agoriclocal", "chain_id": "agoriclocal", "latest_queried_block": 15, "latest_height": 32}
relayer      | 2024-07-05T22:12:41.365149Z      info    Chain is not yet in sync      {"chain_name": "agoriclocal", "chain_id": "agoriclocal", "latest_queried_block": 15, "latest_height": 32}
agoriclocal  | 2024-07-05T22:12:42.347Z block-manager: block 33 begin
agoriclocal  | 2024-07-05T22:12:42.350Z block-manager: block 33 commit
relayer      | 2024-07-05T22:12:42.364923Z      info    Chain is not yet in sync      {"chain_name": "agoriclocal", "chain_id": "agoriclocal", "latest_queried_block": 15, "latest_height": 33}
relayer      | 2024-07-05T22:12:43.365997Z      info    Chain is not yet in sync      {"chain_name": "agoriclocal", "chain_id": "agoriclocal", "latest_queried_block": 15, "latest_height": 33}
relayer      | 2024-07-05T22:12:43.367214Z      warn    Reached max retries querying for block, skipping       {"chain_name": "agoriclocal", "chain_id": "agoriclocal", "height": 16}
relayer      | 2024-07-05T22:12:44.365633Z      info    Chain is in sync        {"chain_name": "agoriclocal", "chain_id": "agoriclocal"}

This may be exacerbated if a chain is hard-forked and started from a genesis export with a high .initial_height.

Do you have advice on the best way to avoid all the retries?

Thanks,
Michael.