Repository navigation
馃悰 Increase default server IdleTimeout to 120s - #3616
kubernetes-prow[bot] merged 1 commit into
Conversation
|
Welcome @jdymitarai! |
|
Hi @jdymitarai. 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 Regular contributors should join the org to skip this step. Once the patch is verified, the new status will be reflected by the I understand the commands that are listed here. DetailsInstructions 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-sigs/prow repository. |
|
LGTM label has been added. DetailsGit tree hash: 80ab6469f84b86a1e9516464902891eb29f141a1 |
|
[APPROVALNOTIFIER] This PR is APPROVED This pull-request has been approved by: alvaroaleman, jdymitarai The full list of commands accepted by this bot can be found here. The pull request process is described here DetailsNeeds approval from an approver in each of these files:
Approvers can indicate their approval by writing |
|
/hold |
18a9ec6 to
06a9101
Compare
|
/test pull-controller-runtime-test |
|
@jdymitarai: Cannot trigger testing until a trusted user reviews the PR and leaves an DetailsIn response to this:
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-sigs/prow repository. |
|
Updated the unit test to use |
|
/test pull-controller-runtime-test |
06a9101 to
13ba018
Compare
When kube-apiserver makes admission calls to webhook servers over HTTP/1.1, the client-go transport uses http.DefaultTransport's 90s idle timeout. Because controller-runtime's internal httpserver also set IdleTimeout to 90s, both client and server close idle connections at approximately the same time, leading to intermittent EOF / broken pipe failures during admission requests. Increase the default server IdleTimeout to 120s so the client initiates closing idle connections first. Signed-off-by: jdymitarai <o10040115@gmail.com>
13ba018 to
9eede03
Compare
|
Updated the suite and test to \package httpserver\ with Ginkgo \SpecContext\ so that the test binary registers Ginkgo flags properly and complies with \orbidigo. Could an org member please re-trigger /ok-to-test? Thanks! |
|
/ok-to-test /hold cancel Thx! |
|
LGTM label has been added. DetailsGit tree hash: 7aa9da4431f5486120ff55b72a78df660b1c451f |
When kube-apiserver communicates with admission webhooks over HTTP/1.1, client-go uses
http.DefaultTransport's 90s idle timeout. Because controller-runtime's internal httpserver also defaultedIdleTimeoutto 90s, both sides close idle connections at approximately the same time, leading to intermittentEOF/ broken pipe transport errors on reused connections.This increases the default server
IdleTimeoutto 120s so that clients initiate closing idle connections before the server drops them.Fixes #3614