relayer: Getting "failed to get trusted headers" error
relayer version: 1.0.0-rc I am trying to connect musselnet-2 and stargate-final networks via IBC. this is the error I am getting at the moment:
rly tx link musselnet-2_stargate-final
I[2021-01-18|00:02:01.812] ★ Clients created: client(07-tendermint-0) on chain[musselnet-2] and client(07-tendermint-0) on chain[stargate-final]
I[2021-01-18|00:02:42.784] failed to get trusted headers: chain(stargate-final): post failed: Post "http://x.x.x.x:26657": EOF
I[2021-01-18|00:02:42.784] retrying transaction...
I[2021-01-18|00:05:22.766] failed to get trusted headers: chain(stargate-final): post failed: Post "http://x.x.x.x:26657": EOF
I[2021-01-18|00:05:22.766] retrying transaction...
I[2021-01-18|00:05:52.751] failed to get trusted headers: chain(stargate-final): post failed: Post "http://x.x.x.x:26657": EOF
I[2021-01-18|00:05:52.751] retrying transaction...
When I followed up the bug I found out this line is causing the error https://github.com/cosmos/relayer/blob/v1.0.0-rc1/relayer/headers.go#L122 which looks like might be related to this issue https://github.com/cosmos/cosmos-sdk/pull/8341
About this issue
- Original URL
- State: closed
- Created 3 years ago
- Comments: 30 (26 by maintainers)
Release v0.8.2 appears to have fixed this for us. Haven’t noticed any issues and have been hitting it pretty hard for a couple hours. Nice work!
we should update the error message to indicate that the headers have likely been pruned
@kogisin thanks for the report!! This was very useful. It made me realize I forgot to check if a client is expired or frozen when reusing existing clients.
This is a Tendermint light client security requirement. Trusting period must always be less than the unbonding-period
Here is what is happening:
Fixes:
@orkunkl thanks for trying. Did the heights in the error message change?
I will try to reproduce your error this week and then continue debugging.
As a side note, you’ll need to update your channel order to be
UNORDERED. ICS-20 does not support ordered channelsI solved the issue by changing our full node pruning strategy to
nothing