You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Currently, we use AWS AppRunner to run bundler service.
The AppRunner can easily be set up to scale, for example, it can automatically bring up more instance when demand is high. One issue of this auto scale is that every bundler instance is isolated to each other and AppRunner doesn't provide a custom way to load balance the requests, this leads to problems like:
Inconsistency on transaction override.
Conflict on EOA nonce.
This Bug tracks the effort of solving problem (1), to see how things work out if all bundler instance share the same mempool, well technically, share the same userOp Queue.
A few high level steps going forwards:
refactor current bundler implement to modulize the queue
implement a queue server which contains the userOp and offer API to operate on them
implement another queue module which use the remote queue as storage
performance test
The text was updated successfully, but these errors were encountered:
Currently, we use AWS AppRunner to run bundler service.
The AppRunner can easily be set up to scale, for example, it can automatically bring up more instance when demand is high. One issue of this auto scale is that every bundler instance is isolated to each other and AppRunner doesn't provide a custom way to load balance the requests, this leads to problems like:
This Bug tracks the effort of solving problem (1), to see how things work out if all bundler instance share the same mempool, well technically, share the same userOp Queue.
A few high level steps going forwards:
The text was updated successfully, but these errors were encountered: