apiserver-network-proxy
apiserver-network-proxy copied to clipboard
kubernetes-sigs
completed 10000 runs in 24m45s with 8 failures (0.08%) Ran with `stress ./tests.test -test.cpu=4 -test.run=TestProxy_LargeResponse` Looks like goroutine leaks: ``` --- FAIL: TestProxy_LargeResponse (1.48s) leaks.go:78: found unexpected goroutines: [Goroutine 83...
completed 10000 runs in 23m7s with 11 failures (0.11%) Ran with `stress ./tests.test -test.cpu=4 -test.run=TestFailedDial_HTTPCONN` First 10 failures: ``` --- FAIL: TestFailedDial_HTTPCONN (1.15s) proxy_test.go:474: Get "http://127.0.0.1:40199": EOF ``` ``` ---...
Feature Request: support clusters with dynamic number of proxy-servers. Example use case: gracefully add a 2nd control plane node to a 1 control plane node cluster. Current state: * proxy-server...
Examples: - [server.go:721](https://github.com/kubernetes-sigs/apiserver-network-proxy/blob/v0.0.27/pkg/server/server.go#L721) `"DIAL_RSP contains failure" err="dial tcp 10.12.1.4:443: i/o timeout" dialID=8897718490787564857 agentID="dbfdab43-278f-4258-836a-22cbacca1764" connectionID=0` - [server.go:761](https://github.com/kubernetes-sigs/apiserver-network-proxy/blob/v0.0.27/pkg/server/server.go#L761) `"could not get frontend client" err="can't find connID 0 in the frontends[24d90262-6980-476e-9649-88f86404327b]" serverID="70d94c44-f904-466c-b51d-1482564f528f" agentID="24d90262-6980-476e-9649-88f86404327b"...
We have integrated the apiserver-network-proxy into our project and stumbled across some possibly more fundamental issues with it. Don’t get me wrong: we really like this project and really appreciate...
I think this project is currently lacking benchmarks and stress tests. These tests can be useful for identifying major regressions and issues. Some ideas of tests I'd like to see:...
As suggested in #179, this is the second part of making the dial request processing asynchronous. Previously, the net.Dial call based on a dial request may block the receive loop...
This way we can easily control on multi-homed machines which interfaces different endpoint listen on. All but admin endpoint are listening by default on all interfaces. Admin is only binding...
We have resources spanning 2 subnets A. a production subnet that has public IPs B. an experimental subnet that is not publicly addressable, and that sits behind a NAT We...
1. `proxy-server` container gets OOM killed repeatedly preventing exec/log commands from working. ``` $ k logs konnectivity-agent-8d795bf89-2dvgs -f Error from server: Get "[https://10.205.194.144:10250/containerLogs/kube-system/konnectivity-agent-8d795bf89-2dvgs/konnectivity-agent?follow=true":](https://10.205.194.144:10250/containerLogs/kube-system/konnectivity-agent-8d795bf89-2dvgs/konnectivity-agent?follow=true%22:) dial timeout, backstop $ k exec -it...
