Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

[sysctl] set fs.may_detach_mounts=1 even when CRIs don't set it themselves #8635

Merged

Conversation

cristicalin
Copy link
Contributor

What type of PR is this?

/kind bug

What this PR does / why we need it:
fs.may_detach_mounts sysctl was introduced in the linux kernel (https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=8ed936b5671bfb33d89bc60bdcc7cf0470ba52fe) to facilitate the removal of mounts on unlinked files and directories. This behavior is expected by CRIs which cannot always control the order of layer removals causing race conditions that can result in "file busy" errors when trying to tare down pods. Since not all CRIs set this setting themselves and the impact of not setting it can be failure to terminate pods we should handle it in kubespray.

Which issue(s) this PR fixes:

Fixes #8622

Special notes for your reviewer:
We should backport this to release-2.18 before tagging 2.18.1 since it affects the default deployment with containerd.

Does this PR introduce a user-facing change?:

[sysctl] set fs.may_detach_mounts=1 to address pods stuck in Terminating state

@k8s-ci-robot k8s-ci-robot added kind/bug Categorizes issue or PR as related to a bug. cncf-cla: yes Indicates the PR's author has signed the CNCF CLA. size/S Denotes a PR that changes 10-29 lines, ignoring generated files. labels Mar 19, 2022
@k8s-ci-robot
Copy link
Contributor

[APPROVALNOTIFIER] This PR is APPROVED

This pull-request has been approved by: cristicalin

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:

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 k8s-ci-robot added the approved Indicates a PR has been approved by an approver from all required OWNERS files. label Mar 19, 2022
@bordenit
Copy link

There should maybe be a conditional to check setting. Maybe if RHEL and setting is 0, set to 1. Not sure if the other supported OS options would take the setting, or how long this setting will be needed. Thanks again for the help with this.

@champtar
Copy link
Contributor

I think this sysctl in CentOS7 only, but simple thing to do is test for the presence of the file and set to 1 only if it exists

@cristicalin cristicalin force-pushed the fs_may_detach_mounts branch 2 times, most recently from 95a7afc to ad1935c Compare March 19, 2022 15:36
@cristicalin cristicalin force-pushed the fs_may_detach_mounts branch from ad1935c to b42b743 Compare March 19, 2022 16:08
@floryut floryut added kind/feature Categorizes issue or PR as related to a new feature. and removed kind/bug Categorizes issue or PR as related to a bug. labels Mar 21, 2022
@oomichi
Copy link
Contributor

oomichi commented Mar 22, 2022

/lgtm

@k8s-ci-robot k8s-ci-robot added the lgtm "Looks good to me", indicates that a PR is ready to be merged. label Mar 22, 2022
@k8s-ci-robot k8s-ci-robot merged commit fa9f85c into kubernetes-sigs:master Mar 22, 2022
cristicalin added a commit to simplekube-ro/kubespray that referenced this pull request Mar 22, 2022
Quehenr pushed a commit to Quehenr/unstable-laboratory-infrastructure-kubespray that referenced this pull request Apr 4, 2022
* master: (21 commits)
  Fix quotation of nerdctl_extra_flags (kubernetes-sigs#8668)
  fix: uninstall contailer engine if service is running (kubernetes-sigs#8662)
  fix disk controller type in Vagrantfile (kubernetes-sigs#8656)
  Add ETCD_EXPERIMENTAL_INITIAL_CORRUPT_CHECK flag to etcd config (kubernetes-sigs#8664)
  Update vagrant.md (kubernetes-sigs#8663)
  Add missing 2.10 ansible test (kubernetes-sigs#8665)
  [ansible] make ansible 5.x the new default version (kubernetes-sigs#8660)
  Run 0100-dhclient-hooks if dhcpclient is enabled (kubernetes-sigs#8658)
  terrform/openstack: Fix templating of ansible_ssh_common_args in no_floating.yml if used as TF module (kubernetes-sigs#8646)
  [ansible] add support for ansible 5 (ansible-core 2.12) (kubernetes-sigs#8512)
  [etcd] add 0 hash for arm v3.5.2 to prevent deployment failures
  Update cert-manager to v1.7.2 (kubernetes-sigs#8648)
  fix vagrant parameter (kubernetes-sigs#8650)
  Remove centos7 molecule while opensuse mirror is flaky
  Remove k8s module for ns creation
  Update kubernetes image to 2.18.1
  Vagrantfile: add var to set ansible verbosity level (kubernetes-sigs#8639)
  Have ingress_controller and external_provisioner in upgrade-cluster.yml (kubernetes-sigs#8640)
  [sysctl] set fs.may_detach_mounts=1 even when CRIs don't set it themselves (kubernetes-sigs#8635)
  Fixed cluster roles for openstack cloud controller (kubernetes-sigs#8638)
  ...
sakuraiyuta pushed a commit to sakuraiyuta/kubespray that referenced this pull request Apr 16, 2022
@oomichi oomichi mentioned this pull request May 28, 2022
LuckySB pushed a commit to southbridgeio/kubespray that referenced this pull request Jun 29, 2023
LuckySB pushed a commit to southbridgeio/kubespray that referenced this pull request Oct 23, 2023
sw-cho pushed a commit to sw-cho/kubespray that referenced this pull request Feb 5, 2024
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
approved Indicates a PR has been approved by an approver from all required OWNERS files. cncf-cla: yes Indicates the PR's author has signed the CNCF CLA. kind/feature Categorizes issue or PR as related to a new feature. lgtm "Looks good to me", indicates that a PR is ready to be merged. size/S Denotes a PR that changes 10-29 lines, ignoring generated files.
Projects
None yet
Development

Successfully merging this pull request may close these issues.

fs.may_detach_mounts=1 is needed for successful pod termination
6 participants