-
-
Notifications
You must be signed in to change notification settings - Fork 287
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
getAttesterDuties is being called multiple times in an epoch by 1 validator process #3201
Comments
@g11tech Note that this chart can be very misleading. A single dot does not mean 1 call. A single dot means that the average value in prometheus db for that metric was X, so it could mean 100, 1000 requests that average to that value. Similarly 3 dots may just represent a single call but because of interval polling between grafana and prometheus it renders as three reads for the same scraped value |
@g11tech You can review the validator attester duties flow here
duties are requested twice per epoch
|
fair point, let me look at the counter of the metric: so going up by 2 would make sense (as in the last observation in the above graph) let me figure out the new indices scenario. |
This chart is much better, clearly the validator is calling getAttesterDuties ~8 times during the first 3 epochs or your screen capture |
👍 will investigate further, may be can help us with our memory problem! |
RECOMMENDATION:
But closing this issue, as investigation has concluded this to be of false positive. |
What would be an issue here is how much regen the calls are causing compounded. If the validator clients call getAttesterDuties multiple times but always result in 1 epoch transition run, then it's okay. The problem would be if multiple epoch transitions are run due to those calls |
yes, this looks a good point to be investigated. i will create an issue for this and carry forward the investigation 👍 |
Describe the bug
In each epoch the validator calls for getAttesterDuties, which are supposed to be called once an epoch unless the reorg happens, but lodestar is consistently making 3-4 requests doesn't matter if the 1, 16 or 975 validators are attached (but through ONE VALIDATOR PROCESS) , and it seems to be doing calculation in all of the calls (may be in parallel)
Expected behavior
Should be called just once (per validator) or wait till previous request is completed.
Steps to Reproduce
Run beacon node on prater, attached validator(s) and observe the metrics dashboard.
Screenshots
for 1 validator:
for 16 validators:
for 975 validators:
Desktop (please complete the following information):
v0.30.0/master/+23/6cf7c2c3 (git)
The text was updated successfully, but these errors were encountered: