-
Notifications
You must be signed in to change notification settings - Fork 118
CNF-10238: NTO render command for SNO boot arguments #844
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
CNF-10238: NTO render command for SNO boot arguments #844
Conversation
|
Skipping CI for Draft Pull Request. |
c99cc52 to
bf92d4c
Compare
|
/test all |
bf92d4c to
a5cfd36
Compare
|
/retest |
fffd0e4 to
da99c2c
Compare
da99c2c to
03c4288
Compare
03c4288 to
c044f65
Compare
|
/cc @vitus133 |
|
/CC @MarSik |
6d33a60 to
f5940f4
Compare
853de8c to
4fabcc5
Compare
| tuneD = append(tuneD, tunedFromPP) | ||
| } | ||
|
|
||
| tuneDrecommended := operator.TunedRecommend(tuneD) |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I wonder if this should go to an MCP loop in case there are multiple pools with multiple profiles.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Actually, I just realized we need to do the opposite. We need to make sure we will select only render the tuned and perf profile for the master MCP (probably). We do not know the proper cpu topology for the other MCPs.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
So ... still need to load all the PerformanceProfiles? or just those which will match with MCP master?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I think master only and only on SNO.
Signed-off-by: Jose Luis Ojosnegros Manchón <[email protected]>
4fabcc5 to
f92e268
Compare
|
/retest |
|
/retitle CNF-10238: NTO render command for SNO boot arguments We want to merge the current working solution (tested) and iterate from there to avoid unnecessary duplication of work wrt OCP branching. |
|
@jlojosnegros: This pull request references CNF-10238 which is a valid jira issue. Warning: The referenced jira issue has an invalid target version for the target branch this PR targets: expected the sub-task to target the "4.15.0" version, but no target version was set. 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/test-infra repository. |
|
/lgtm |
|
/approve |
|
[APPROVALNOTIFIER] This PR is APPROVED This pull-request has been approved by: jlojosnegros, MarSik 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 |
|
@jlojosnegros: all tests passed! Full PR test history. Your PR dashboard. 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/test-infra repository. I understand the commands that are listed here. |
|
[ART PR BUILD NOTIFIER] This PR has been included in build cluster-node-tuning-operator-container-v4.15.0-202312071813.p0.g901f395.assembly.stream for distgit cluster-node-tuning-operator. |
This change combines PRs 970, 998 and 1024 which fixed OCPBUGS-30647 in 4.16.
Summary of changes:
* Even though there is currently no namespace collision with TuneD
using in/var/lib/tuned, change this path to /var/lib/ocp-tuned
* Remove bin/run. While this means a little code duplication across
Containerfiles, we no longer need to do anything special at run
time. This should make things easier for the future.
* Do not inherit --enable-leader-election and --version NTO flags
as they are not handled by subcommands anyway (yet)
* Remove openshift-tuned binary and use NTO subcommand instead.
* /var/lib/tuned/profiles-data is no longer used, remove it.
* Remove openshift-tuned PID file code. It is no longer used.
* Clean up after openshift#844
* Remove TuneD timeout code and reload on ERRORs
* fix logging in updateTunedProfile() and optimize the calls
to update node annotations and update Profile.Status
* clean up tunedStop() to return only one value
* during TuneD process shutdown, handle the fact the TuneD process
might have already exitted
* the openshift-tuned operand now no longer unnecessarily exits when
TuneD process exits; when TuneD process exits, wait for k8s object
changes and only then restart TuneD
* do not use buffered channels
* the indication that TuneD is reloading is now a status bit potentially
reportable back to the operator
* introduce Change type for the TuneD event processor to avoid races, where
it was previously possible to change TuneD configuration during TuneD profile
reload
* register the fact TuneD finished reloading in case the primary TuneD profile
does not exist
* conditional TuneD reload when Cloud Provider changes
* minor logging and comment improvements
Resolves: OCPBUGS-36355
This change combines PRs 970, 998 and 1024 which fixed OCPBUGS-30647 in 4.16.
Summary of changes:
* Even though there is currently no namespace collision with TuneD
using in/var/lib/tuned, change this path to /var/lib/ocp-tuned
* Remove bin/run. While this means a little code duplication across
Containerfiles, we no longer need to do anything special at run
time. This should make things easier for the future.
* Do not inherit --enable-leader-election and --version NTO flags
as they are not handled by subcommands anyway (yet)
* Remove openshift-tuned binary and use NTO subcommand instead.
* /var/lib/tuned/profiles-data is no longer used, remove it.
* Remove openshift-tuned PID file code. It is no longer used.
* Clean up after openshift#844
* Remove TuneD timeout code and reload on ERRORs
* fix logging in updateTunedProfile() and optimize the calls
to update node annotations and update Profile.Status
* clean up tunedStop() to return only one value
* during TuneD process shutdown, handle the fact the TuneD process
might have already exitted
* the openshift-tuned operand now no longer unnecessarily exits when
TuneD process exits; when TuneD process exits, wait for k8s object
changes and only then restart TuneD
* do not use buffered channels
* the indication that TuneD is reloading is now a status bit potentially
reportable back to the operator
* introduce Change type for the TuneD event processor to avoid races, where
it was previously possible to change TuneD configuration during TuneD profile
reload
* register the fact TuneD finished reloading in case the primary TuneD profile
does not exist
* conditional TuneD reload when Cloud Provider changes
* minor logging and comment improvements
Resolves: OCPBUGS-36355
This change combines PRs 970, 998 and 1024 which fixed OCPBUGS-30647 in 4.16.
Summary of changes:
* Even though there is currently no namespace collision with TuneD
using in/var/lib/tuned, change this path to /var/lib/ocp-tuned
* Remove bin/run. While this means a little code duplication across
Containerfiles, we no longer need to do anything special at run
time. This should make things easier for the future.
* Do not inherit --enable-leader-election and --version NTO flags
as they are not handled by subcommands anyway (yet)
* Remove openshift-tuned binary and use NTO subcommand instead.
* /var/lib/tuned/profiles-data is no longer used, remove it.
* Remove openshift-tuned PID file code. It is no longer used.
* Clean up after #844
* Remove TuneD timeout code and reload on ERRORs
* fix logging in updateTunedProfile() and optimize the calls
to update node annotations and update Profile.Status
* clean up tunedStop() to return only one value
* during TuneD process shutdown, handle the fact the TuneD process
might have already exitted
* the openshift-tuned operand now no longer unnecessarily exits when
TuneD process exits; when TuneD process exits, wait for k8s object
changes and only then restart TuneD
* do not use buffered channels
* the indication that TuneD is reloading is now a status bit potentially
reportable back to the operator
* introduce Change type for the TuneD event processor to avoid races, where
it was previously possible to change TuneD configuration during TuneD profile
reload
* register the fact TuneD finished reloading in case the primary TuneD profile
does not exist
* conditional TuneD reload when Cloud Provider changes
* minor logging and comment improvements
Resolves: OCPBUGS-36355
Co-authored-by: Jiri Mencak <[email protected]>
This is a backport of openshift#1095 which fixed OCPBUGS-36355 in 4.15. Summary of changes: * Change the operand's home directory from TuneD's artifacts directory /var/lib/tuned to /var/lib/ocp-tuned * Remove bin/run. While this means a little code duplication across Containerfiles, we no longer need to do anything special at run time. This should make things easier for the future. * Do not inherit --enable-leader-election and --version NTO flags as they are not handled by subcommands anyway (yet) * Remove openshift-tuned binary and use NTO subcommand instead. * /var/lib/tuned/profiles-data is no longer used, remove it. * Remove openshift-tuned PID file code. It is no longer used. * Clean up after openshift#844 * Remove TuneD timeout code and reload on ERRORs * Fix logging in updateTunedProfile() and optimize the calls to update node annotations and update Profile.Status * Clean up tunedStop() to return only one value * During TuneD process shutdown, handle the fact the TuneD process might have already exitted * The openshift-tuned operand now no longer unnecessarily exits when TuneD process exits; when TuneD process exits, wait for k8s object changes and only then restart TuneD * Do not use buffered channels * The indication that TuneD is reloading is now a status bit potentially reportable back to the operator * Introduce Change type for the TuneD event processor to avoid races, where it was previously possible to change TuneD configuration during TuneD profile reload * Register the fact TuneD finished reloading in case the primary TuneD profile does not exist * Conditional TuneD reload when Cloud Provider changes * Minor logging and comment improvements Resolves: OCPBUGS-37734
This is a backport of #1095 which fixed OCPBUGS-36355 in 4.15. Summary of changes: * Change the operand's home directory from TuneD's artifacts directory /var/lib/tuned to /var/lib/ocp-tuned * Remove bin/run. While this means a little code duplication across Containerfiles, we no longer need to do anything special at run time. This should make things easier for the future. * Do not inherit --enable-leader-election and --version NTO flags as they are not handled by subcommands anyway (yet) * Remove openshift-tuned binary and use NTO subcommand instead. * /var/lib/tuned/profiles-data is no longer used, remove it. * Remove openshift-tuned PID file code. It is no longer used. * Clean up after #844 * Remove TuneD timeout code and reload on ERRORs * Fix logging in updateTunedProfile() and optimize the calls to update node annotations and update Profile.Status * Clean up tunedStop() to return only one value * During TuneD process shutdown, handle the fact the TuneD process might have already exitted * The openshift-tuned operand now no longer unnecessarily exits when TuneD process exits; when TuneD process exits, wait for k8s object changes and only then restart TuneD * Do not use buffered channels * The indication that TuneD is reloading is now a status bit potentially reportable back to the operator * Introduce Change type for the TuneD event processor to avoid races, where it was previously possible to change TuneD configuration during TuneD profile reload * Register the fact TuneD finished reloading in case the primary TuneD profile does not exist * Conditional TuneD reload when Cloud Provider changes * Minor logging and comment improvements Resolves: OCPBUGS-37734 Co-authored-by: Jiri Mencak <[email protected]>
We need to speed up bootstrap.
To do so we need to apply kernel boot arguments without restarting node.
Add a new render command to prepare all tuned profiles, run
tuneD, readbootcmdlineand render a MachineConfig to apply those kernel arguments.Warning: running
tuneDwould modify some system files so this command should be executed in a properly isolated environment.see: openshift/installer#7692