-
Notifications
You must be signed in to change notification settings - Fork 326
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Packets not clearing #1918
Comments
You might be able to clear all the packets on a given channel using the Nonetheless, it is indeed not normal that Hermes is not clearing any packets on start. As a sanity check, do you have |
Yes, the config is correct. Also, I did use Can hermes clear the packets in batch selected by us? |
Update:
Can you suggest some solution to this? |
Wow, this is an interesting case, thanks for flagging @RohitAudit !
This is because of two reasons:
One thing we can do is to add pagination support in Hermes (ref: #1816). Even with pagination support, I'm not 100% about the error "consensus state not found" in your last comment. We're still looking into that. Can you tell us the SDK version that |
gravity-bridge-3 use cosmos-sdk v0.44.5 and ibc version v2.0.2. |
Thanks for the quick feedback! WRT archival nodes: I think it may speed things up, yes (esp. if the nodes have also slightly higher-compute specs), but will ask some relayer operators if they have ideas of potential optimizations. |
First of all, great work. Testing the network like this is the only way we improve much of anything. Secondly, having an archive node can help. With that said, you will need very fast hard drives for that to work. Something that our team does when we encounter congestion is we run both the go relayer and Hermes. They seem to compliment one another in practice in ways that we don’t fully understand yet but definitely have seen in action. If you do that you will want to have two separate wallets. One for the go relayer and one for Hermes. Concerning speed, my only consistent piece of advice is you want really fast drives. |
in my experience querying archives is less effective because of the sheer db size involved; best would be to have a statesynced node that only retains a chainstate a bit longer than the packets are stuck. @adizere in my opinion having pagination or batching support in such situations would greatly increase relayer production experience |
You know you’re right and I probably should’ve mentioned that the archive node only really helps when you’re dealing with older packets So one thing you probably do not want to do is state sync right now but instead, next time or you might not have the needed history |
I tried using two hermes relayer but both of them picks the same transaction and one fails. |
the only thing that will significantly help this situation are dedicated endpoints for each redundant relayer (particularly on the |
yep, more endpoints will be more better |
We'll be tracking support for query pagination in #1816. Closing this as hopefully addressed. |
Summary of Bug
I am using hermes relayer for gravity-bridge and persistence chain. I brought some tokens from ethereum to gravity and was trying to relay them to the persistence chain. The relayer was stopped and 6000 transaction was sent. When the relayer was started again, all packets got stuck and nothing is getting transmitted. I restarted the relayer after an hour but it is still the same.
Link to the transaction
Debug Logs Printed:
Version
v0.11.0 and v0.12.0
Steps to Reproduce
Start a relayer between gravity and persistence and pool more than 6000 transactions and then start the relayer to clear empty packets.
For Admin Use
The text was updated successfully, but these errors were encountered: