Closed gamarin2 closed 3 years ago
Just had a conversation with the Cephalopod team, who's currently running relayers for Osmosis. Notes can be found here https://hackmd.io/8W0tTnqITqe57RgPz7Bnjw
Based on the conversation, here are the important parameters where we need to use something other than default value:
max_msg_num
can be defaulted to 15max_gas
should be set to 400k on low usage networks (akash, sentinel, regen, persistence) and 800k on higher usage networks (IRIS, OSMOSIS, Crypto.org). On the Cosmos Hub, we can push to 1M. clock drift
: we can start with 30min (1800s), and then gradually lower it trusting period
must be lower than unbonding period, which varies from network to network. We can use 14 days by default, but we need to pay attention when we add a new networkclear_packets_interval
: basically a retry timer. Cephalopod set it to 600. filter
must be set to true
(new param added in latest release). We need not configure channels, but we need filter to true to exclude weird clients. Cephalopod insisted on the fact that early on we'll need to monitor logs to adjust params per network (params need to be fine-tuned per network). Also why they recommend running log in "debug" mode.
cc @gsora @helder-moreira
Will work on this today.
We need to update config of relayer with new values before mainnets. Example of params:
max_msg_num
from 30 to 12clock_drift
from 5s to ? (7200s used by Osmosis)