Relayer and Firehose have hard time to get started when network is slow and blocks are fast

[matthewdarwin]

On a blockchain like arbitrum-one where blocks are every 4 seconds, both the relayer and firehose have a challenge to sync to the chain. This is especially the case when the one-blocks/forked-blocks are not local (ie there is network delay)

Sometimes the merger will merge and remove the one-block files before the relayer has a chance to read them.

[billettc]

@matthewdarwin Can you please set your relayer and merge with log set to DEBUG and send me the log when the bug is happening again?

[sduchesneau]

Initial connection has been fixed since v1.5.1
However, one failure scenario is not handled well. Fix is incoming.

[sduchesneau]

Fixed in 1f24594

let me know if you see any "unfixable disconnection" !
The 'hub', upon reconnection with a block in the future, will load all the available 'one-block-files' from that block's LIB and up, to see if it can join.

If it can't, it will (now successfully!) restart the whole process because it needs to restart all its 'pipelines' from merged blocks.

[matthewdarwin]

Firehose still gets stuck on

2024-06-24T14:59:06.089Z INFO (bstream) reconnecting hub after disconnection. expecting to reconnect {"error": "rpc error: code = Unavailable desc = error reading from server: EOF"}