Enable node e2e accounting on systemd #26289
Conversation
k8s-github-robot
added
size/S
|
@pwittrock - i don't quite know when in the flow or if this script is invoked or not to know how to respond to the pr checks... please assist. |
derekwaynecarr
added
release-note-none
| # tested rhel 7 (3.10.0-229.7.2.el7.x86_64) using systemd 219 | ||
| if [ -d /etc/systemd ]; then | ||
| sudo mkdir -p /etc/systemd/system.conf.d/ | ||
| sudo cat <<EOF >/etc/systemd/system.conf.d/kubernetes-accounting.conf |
There was a problem hiding this comment.
@ingvagabund on rhel 7 with systemd 219, i had to use the above path.
|
This script is only invoked manually as part of the manual process of creating an image. Your changes should not have caused any failures. These are most likely flakes. |
|
@derekwaynecarr If you have not already, the best way to test that this script does what you want is to create a new rhel host, run the script on the pristine host and then run the node e2e test suite against that host using the comment in e2e-node-jenkins.sh with your *.properties file (copied from the template.properties) |
|
@derekwaynecarr Let me know if you are satisfied with any testing you have done using this script to setup the rhel host. |
|
@pwittrock - I was out of office most of yesterday, testing this now on GCE vanilla rhel 7 image. |
|
@derekwaynecarr SGTM. Wanted to make sure I wasn't blocking you. |
Install docker from RHEL repo on RHEL
derekwaynecarr
force-pushed
the
enable_accounting
branch
from
6b866d9 to
67493aa
Compare
k8s-github-robot
added
size/M
/cc @pwittrock @vishh |
|
@pwittrock @derekwaynecarr Are we adding RHEL and CentOS nodes to our node test pool? |
|
We should if this makes it so they pass consistently. |
|
@pwittrock @vishh - I don't know what test pools you run internally, but the goal here was to just get the script updated to enable accounting globally however you guys setup the base images. If you run this script on any systemd system, it should do the right thing for you and let us unblock other PRs. CI stack for Red Hat stuff can be invoked on a per PR basis by just saying: @redhat-k8s-bot test this (if the pr is labeled team/node) |
pwittrock
added
e2e-not-required
lgtm
pwittrock
removed
the
lgtm
|
I am going to setup an instance with this script, and then lgtm it |
|
@derekwaynecarr I tried to verify this today but ran into some trouble. Will pick it up again tomorrow. |
pwittrock
added
the
lgtm
|
@k8s-bot test this Tests are more than 96 hours old. Re-running tests. |
|
GCE e2e build/test failed for commit 67493aa. Please reference the list of currently known flakes when examining this failure. If you request a re-test, you must reference the issue describing the flake. |
|
Automatic merge from submit-queue |
Automatic merge from submit-queue Update Node e2e Core OS image to run systemd with CPU & Memory accounting enabled by default cc @derekwaynecarr For #26289
Automatic merge from submit-queue Fix system container detection ```release-note Fix system container detection in kubelet on systemd. This fixed environments where CPU and Memory Accounting were not enabled on the unit that launched the kubelet or docker from reporting the root cgroup when monitoring usage stats for those components. ``` Fixes #25909 /cc @vishh @ncdc @eparis @wojtek-t - this reverts the revert #26478 This should merge when node e2e setups were updated per #26289
Updated the e2e setup.sh script to enable cpu and memory accounting.
Related to #26198
/cc @pwittrock