agent/xds: Update mesh gateway to use service router timeout #7444
Conversation
There was a problem hiding this comment.
This looks to all function correctly but I found the flow in the cluster generation a little hard to follow. Could you take a look and see if there is a reasonable way to refactor. If all other alternatives are "ugly" then maybe just adding a few comments (like at that continue and outside of the hasResolver block) indicating how the control-flow should be working.
agent/xds/clusters.go
Outdated
| return nil, err | ||
| } | ||
| clusters = append(clusters, cluster) | ||
| continue |
There was a problem hiding this comment.
It was immediately clear to me why there was a continue here. I eventually figured it out that it was to avoid the code below when we already know we don't have a resolver (to prevent some segfaults)
Could there be a way to refactor this code a little to make the code flow a little clearer? Maybe a way that we only generate the unnamed subsets cluster in 1 location?
There was a problem hiding this comment.
Updated! Happy to improve readability
findkim
force-pushed
the
mesh-gateway-connect-timeout
branch
from
5b6dc61 to
a4a9849
Compare
findkim
force-pushed
the
mesh-gateway-connect-timeout
branch
from
a4a9849 to
609ce45
Compare
* website/connect/proxy/envoy: specify timeout precedence for services behind mesh gateway
The PR updates the connection timeout used for services behind a mesh gateway to prioritize the configured
service-resolvertimeout, and falls back to use the timeout of the mesh gateway.Fixes #6370