Discovery Rework Ideas and Shrex Blockage #2191
distractedm1nd
started this conversation in
General
Replies: 0 comments
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
-
There is currently a problem with shrex on the BSR network, where we get context deadlines in two spots. This makes syncing eventually come to a screeching halt. I have not been able to get a FN to sync against BSR fully since discovery refactorings.
Because these can both deadline after 60 seconds (30s on 3rd attempt), even a single bad peer can act as an "attractor" for the daser workers to get stuck on in their routines. There are multiple issues to be resolved here:
(yes I will open single issues and link them here)
The best hotfix I could test against BSR was to add a
ResultDropPeer
, that removes the peer from the FN set and closes the connection with them. But this is a very hacky way to indirectly modify the discovery set from peer manager.@walldiss has a good proposal to:
One other suggestion we tried out is to put these unresponsive peers on a long cooldown, essentially "soft banning" them. This wasn't enough to fix syncing with a timeout of 1 minute and disc peer limit of 15.
All of these needs can be summarized that we need to find a way to have some sort of simple peer scoring for shrex/peerman.
This discussion is here to serve two purposes:
Beta Was this translation helpful? Give feedback.
All reactions