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
Is your feature request related to a problem? Please describe.
One way to reduce or eliminate refinery-to-refinery span traffic would be to allow a smart load balancer to distribute traces to the "correct" refinery. But the distribution algorithm is complex and not easily recreated by an outside service.
Describe the solution you'd like
A query endpoint that a load balancer could use to find out which node should receive a given trace.
Either:
Built into Refinery directly
A separate executable that could be run as a local service (it would have to be listening to the peer gossip conversation to keep up to date with the current state of the nodes)
Describe alternatives you've considered
If Refinery nodes aren't clustered, they won't talk to each other. A smart load balancer could make its own decisions with its choice of algorithm without affecting how Refinery behaves.
Is your feature request related to a problem? Please describe.
One way to reduce or eliminate refinery-to-refinery span traffic would be to allow a smart load balancer to distribute traces to the "correct" refinery. But the distribution algorithm is complex and not easily recreated by an outside service.
Describe the solution you'd like
A query endpoint that a load balancer could use to find out which node should receive a given trace.
Either:
Describe alternatives you've considered
If Refinery nodes aren't clustered, they won't talk to each other. A smart load balancer could make its own decisions with its choice of algorithm without affecting how Refinery behaves.
Additional context
Pollinators discussion
The text was updated successfully, but these errors were encountered: