Add keepalive option to agent server #154

Avatat · 2020-10-20T11:37:14Z

This is a fix to issue #152.

Sometimes, agent can disconnect without closing connection, and server don't know about it. Disconnected agent is still on the available agents list and server tries to route traffic through it.

k8s-ci-robot · 2020-10-20T11:37:17Z

Thanks for your pull request. Before we can look at your pull request, you'll need to sign a Contributor License Agreement (CLA).

📝 Please follow instructions at https://git.k8s.io/community/CLA.md#the-contributor-license-agreement to sign the CLA.

It may take a couple minutes for the CLA signature to be fully registered; after that, please reply here with a new comment and we'll verify. Thanks.

If you've already signed a CLA, it's possible we don't have your GitHub username or you're using a different email address. Check your existing CLA data and verify that your email is set on your git commits.
If you signed the CLA as a corporation, please sign in with your organization's credentials at https://identity.linuxfoundation.org/projects/cncf to be authorized.
If you have done the above and are still having issues with the CLA being reported as unsigned, please log a ticket with the Linux Foundation Helpdesk: https://support.linuxfoundation.org/
Should you encounter any issues with the Linux Foundation Helpdesk, send a message to the backup e-mail support address at: login-issues@jira.linuxfoundation.org

Instructions for interacting with me using PR comments are available here. If you have questions or suggestions related to my behavior, please file an issue against the kubernetes/test-infra repository. I understand the commands that are listed here.

k8s-ci-robot · 2020-10-20T11:37:22Z

Welcome @Avatat!

It looks like this is your first PR to kubernetes-sigs/apiserver-network-proxy 🎉. Please refer to our pull request process documentation to help your PR have a smooth ride to approval.

You will be prompted by a bot to use commands during the review process. Do not be afraid to follow the prompts! It is okay to experiment. Here is the bot commands documentation.

You can also check if kubernetes-sigs/apiserver-network-proxy has its own contribution guidelines.

You may want to refer to our testing guide if you run into trouble with your tests not passing.

If you are having difficulty getting your pull request seen, please follow the recommended escalation practices. Also, for tips and tricks in the contribution process you may want to read the Kubernetes contributor cheat sheet. We want to make sure your contribution gets all the attention it needs!

Thank you, and welcome to Kubernetes. 😃

k8s-ci-robot · 2020-10-20T11:37:22Z

Hi @Avatat. Thanks for your PR.

I'm waiting for a kubernetes-sigs member to verify that this patch is reasonable to test. If it is, they should reply with /ok-to-test on its own line. Until that is done, I will not automatically test new commits in this PR, but the usual testing commands by org members will still work. Regular contributors should join the org to skip this step.

Once the patch is verified, the new status will be reflected by the ok-to-test label.

I understand the commands that are listed here.

Instructions for interacting with me using PR comments are available here. If you have questions or suggestions related to my behavior, please file an issue against the kubernetes/test-infra repository.

Avatat · 2020-10-20T13:12:00Z

I signed CLA.

Jefftree · 2020-10-20T14:31:39Z

/ok-to-test

Avatat · 2020-10-20T15:31:12Z

/retest

Avatat · 2020-10-20T15:37:38Z

@Jefftree, should I write a test, which will cover a bug from #152?
To be honest, I have no idea how to do this, where to begin.

cheftako · 2020-10-20T16:32:12Z

cmd/server/main.go

-	grpcServer := grpc.NewServer(serverOption)
+	serverOptions := []grpc.ServerOption {
+		grpc.Creds(credentials.NewTLS(tlsConfig)),
+		grpc.KeepaliveParams(keepalive.ServerParameters{Time: 30 * time.Second}),


This is way too short to be the default. Currently the max K8s cluster size is 15K nodes. For such a cluster you would be sending 500 keep-alive packets per second to each server. I would suggest making the default 1 hour (the default TCP keepalive on the server is 2 hours (so 1/2 that is good) and the default server enforced minimum is 5 minutes). I'm fine with this being a optional command line flag so you can tune this to your needs. I think we are still likely to also want to mark problematic tunnels as unhealthy but having both seems a good idea.

I wonder if we also want to set a keep alive on the agent side.

Adding an optional command line flag is a good idea - I will figure it out.

As I understand, we don't need a keepalive mechanism on the agent side, because we currently check health there, using grpc.ClientConn.GetState.

I believe (I could certainly be wrong) that that GetState is mostly tracking if closed has been called. It is not tracking if we have either a network outage such that traffic doesn't route or that the server has gone away without notifying us. These are the conditions which keepalive is meant to help. https://godoc.org/google.golang.org/grpc#WithKeepaliveParams client side keep alive is almost identical to server side keepalive and they should work well together.

Just going to add that SSH tunnels have a healthCheckPoll time of 1 minute (https://github.com/kubernetes/kubernetes/blob/master/pkg/ssh/ssh.go#L316), so it's possible that we could go lower than 1 hour for the keepalive time.

We can probably address client side keep alive in a separate PR.

I think so, too. ~1 minute keepalive time isn't something bad for performance - remember, that we send keepalive packets only when the channel is inactive for longer than specified time. In a 15K-node cluster, administrators probably will not deploy konnectivity as daemonset, on every node.
I'm currently working on adding a command line option to allow the administrator to tune this parameter.

After I complete this PR, I would like to try to add keepalive to the agent and then create a new PR :)

@cheftako, @Jefftree, I've just added a --keepalive-time command-line option to the server code and Kubernetes example.

cmd/server/main.go

This is a fix to issue #152. Sometimes, agent can disconnect without closing connection, and server don't know about it. Disconnected agent is still on the available agents list and server tries to route traffic through it.

cheftako · 2020-10-21T15:50:47Z

/lgtm
/approve

k8s-ci-robot · 2020-10-21T15:50:59Z

[APPROVALNOTIFIER] This PR is APPROVED

This pull-request has been approved by: Avatat, cheftako

The full list of commands accepted by this bot can be found here.

The pull request process is described here

Needs approval from an approver in each of these files:

~~OWNERS~~ [cheftako]

Approvers can indicate their approval by writing /approve in a comment
Approvers can cancel approval by writing /approve cancel in a comment

k8s-ci-robot added the cncf-cla: no Indicates the PR's author has not signed the CNCF CLA. label Oct 20, 2020

k8s-ci-robot added the needs-ok-to-test Indicates a PR that requires an org member to verify it is safe to test. label Oct 20, 2020

k8s-ci-robot requested review from Jefftree and Sh4d1 October 20, 2020 11:37

k8s-ci-robot added the size/XS Denotes a PR that changes 0-9 lines, ignoring generated files. label Oct 20, 2020

Avatat mentioned this pull request Oct 20, 2020

Server tries to route traffic through "hard" disconnected agents #152

Closed

k8s-ci-robot added cncf-cla: yes Indicates the PR's author has signed the CNCF CLA. and removed cncf-cla: no Indicates the PR's author has not signed the CNCF CLA. labels Oct 20, 2020

k8s-ci-robot added ok-to-test Indicates a non-member PR verified by an org member that is safe to test. and removed needs-ok-to-test Indicates a PR that requires an org member to verify it is safe to test. labels Oct 20, 2020

cheftako reviewed Oct 20, 2020

View reviewed changes

k8s-ci-robot added size/S Denotes a PR that changes 10-29 lines, ignoring generated files. and removed size/XS Denotes a PR that changes 0-9 lines, ignoring generated files. labels Oct 20, 2020

Jefftree reviewed Oct 20, 2020

View reviewed changes

cmd/server/main.go Outdated Show resolved Hide resolved

Add keepalive option to agent server

b653688

This is a fix to issue #152. Sometimes, agent can disconnect without closing connection, and server don't know about it. Disconnected agent is still on the available agents list and server tries to route traffic through it.

k8s-ci-robot assigned cheftako Oct 21, 2020

k8s-ci-robot added the lgtm "Looks good to me", indicates that a PR is ready to be merged. label Oct 21, 2020

k8s-ci-robot added the approved Indicates a PR has been approved by an approver from all required OWNERS files. label Oct 21, 2020

k8s-ci-robot merged commit 806811f into kubernetes-sigs:master Oct 21, 2020

Avatat deleted the server-keepalive branch October 21, 2020 16:43

Jefftree mentioned this pull request Nov 9, 2020

Bump network proxy to v0.0.14 kubernetes/kubernetes#96389

Merged

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Add keepalive option to agent server #154

Add keepalive option to agent server #154

Avatat commented Oct 20, 2020

k8s-ci-robot commented Oct 20, 2020

k8s-ci-robot commented Oct 20, 2020

k8s-ci-robot commented Oct 20, 2020

Avatat commented Oct 20, 2020

Jefftree commented Oct 20, 2020

Avatat commented Oct 20, 2020

Avatat commented Oct 20, 2020

cheftako Oct 20, 2020 •

edited

cheftako Oct 20, 2020

Avatat Oct 20, 2020

cheftako Oct 20, 2020

Jefftree Oct 20, 2020

Avatat Oct 20, 2020

Avatat Oct 20, 2020

cheftako commented Oct 21, 2020

k8s-ci-robot commented Oct 21, 2020

Add keepalive option to agent server #154

Add keepalive option to agent server #154

Conversation

Avatat commented Oct 20, 2020

k8s-ci-robot commented Oct 20, 2020

k8s-ci-robot commented Oct 20, 2020

k8s-ci-robot commented Oct 20, 2020

Avatat commented Oct 20, 2020

Jefftree commented Oct 20, 2020

Avatat commented Oct 20, 2020

Avatat commented Oct 20, 2020

cheftako Oct 20, 2020 • edited

Choose a reason for hiding this comment

cheftako Oct 20, 2020

Choose a reason for hiding this comment

Avatat Oct 20, 2020

Choose a reason for hiding this comment

cheftako Oct 20, 2020

Choose a reason for hiding this comment

Jefftree Oct 20, 2020

Choose a reason for hiding this comment

Avatat Oct 20, 2020

Choose a reason for hiding this comment

Avatat Oct 20, 2020

Choose a reason for hiding this comment

cheftako commented Oct 21, 2020

k8s-ci-robot commented Oct 21, 2020

cheftako Oct 20, 2020 •

edited