- Newest
- Most votes
- Most comments
Thanks Ashish for your answer. Not sure this Github issue really covers the use case I shared, let me clarify. By default GRPC will open X channels / connections to servers, once done, as long as the connections remain alive it won't try to open any new one.
Let's assume the following case:
We have 1 client and 10 servers Client opened connections to the 10 servers and keep them alive as long as the servers exist Then traffic increases and we auto-scale-out to 2 clients and 12 servers. What happens is that the 2 new servers will receive only very little traffic coming only from the new client. Because the initial client still has all its connections opened. First 10 servers will probably dislike the too big load and problems will start.
Browsing the web a bit more the last days, I think I found what I need: https://github.com/aws-samples/aws-cloudmap-grpc-name-resolver This way, GRPC client load balancing will automatically discover the new servers and open new connections accordingly to properly balance the load.
Probably mixed up things talking about xDS in my initial message, how that worked wasn't so clear at time of writing. Thanks
In case of App Mesh, We are configuring round-robin load balancing and this is not yet configurable in App Mesh API.
This read should really help to get info about how to have my clients connecting to it. https://github.com/aws/aws-app-mesh-roadmap/issues/70
Relevant content
- asked 10 months ago
- asked a year ago
- Accepted Answerasked 4 years ago
- asked 2 years ago
- AWS OFFICIALUpdated 8 months ago
- AWS OFFICIALUpdated 2 months ago
- AWS OFFICIALUpdated 2 years ago
- AWS OFFICIALUpdated 5 months ago