monorepo
monorepo copied to clipboard
Relayer fees are too high
Background
From feedback recently my Ministro
Tere are many cases where the relayer fee has been excessive. He’s afraid most of the cases are forgotten Seeing lots of relayer fee issues even between L2s. ex/ This user bridged Arbitrum to Optimism and he paid $50 and he hasn’t received a good answer. Had another $18 from optimsm to binance chain
- https://discord.com/channels/454734546869551114/1171878046253465610/1171878870241263637
I'm very concerned that this is actually impacting our pricing and volumes and we dont know it
Analysis
We see sporadic results
- We can look at relayerFee quotes plotted against time, compare with actual gas prices, and implement an anomaly detection scheme to determine how often our quotes are totally off.
- This is something we can build into a data pipeline Quoted relayerFee for transfers with Ethereum as destination chain Gas prices on Ethereum in the same period
- % anomalous estimates by destination chain
- at 30% deviation threshold
- ETH: 10.02%
- BNB: 16.87%
- Arb: 10.81%
- Op: 1.88%
- Poly: 22.11%
- at 50% deviation threshold
- ETH: 6.79%
- BNB: 12.79%
- Arb: 6.95%
- Op: 1.32%
- Poly: 14.54%
- at 30% deviation threshold
Tasks
Next steps:
- Dig into:
-
- transactions with 0 relayer fee → unable to repro
-
- reproduce / look into specific bad transfers (like the socket one thru MM) → unable to repro
-
- close out some of the current reliability bugs
- later) relayer fee calculation (calldata size, gas price oracle, conversion rates, gelato API consistency)
- later) set up archive logs on DD (probably want at least 1 month of lookup)
-
user is saying the issue is optimism origin from convo this would just be a conversion rate bug if truly based on origin chain
next thing to look at is how can we optimize the relayer fee function may be less necessary with V2?
https://lookerstudio.google.com/u/0/reporting/3f9ca497-05b3-4a97-a623-9a776917b579/page/p_oa9ycsgsfd/edit?pli=1
keeping open for a full fix
Fixing individual bugs for this.