-
Notifications
You must be signed in to change notification settings - Fork 63
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
Autoperf #358
Autoperf #358
Conversation
1c6db36
to
db9d6a8
Compare
a33f0de
to
df9cef1
Compare
9929979
to
85a0304
Compare
1c6848f
to
c859c87
Compare
e9255c6
to
2c7b3a2
Compare
Signed-off-by: Simon A. F. Lund <simon.lund@samsung.com>
Signed-off-by: Simon A. F. Lund <simon.lund@samsung.com>
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.
In general very good, however there are some things related to batching that should be reconsidered and redone.
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.
Many things related to the documentation and performance report
Signed-off-by: Simon A. F. Lund <simon.lund@samsung.com> Co-authored-by: Karl Bonde Torp <karlowich@gmail.com>
This is inherited from "cijoe-pkg-linux", and extended with options utilized for interface efficiency evaluation, as done in the fio repository at "t/one-core-peak.sh". Signed-off-by: Simon A. F. Lund <simon.lund@samsung.com>
* configs/bench-*.toml - These configurations are equivalent - They introduce a new section "duts", "duts" have the information needed to produce bdev-configs - The configurations specifically describe the machines utilized for automating the performance evaluation, this will be covered by documentation on the xNVMe CI environment * scripts/spdk_bdev_confs_generator.py - Emits bdev-configurations on the target-system - It utilizes the list of [['duts']] aka "device-under-test" * scripts/bdevperf.py - Wraps the execution of the SPDK benchmark tool bdevperf - Names output-files by a convention utilized by the 'bench_plotter.py' * scripts/bench_plotter.py - Produce plots from the output of 'bdevperf.py' - Auxiliary files 'auxiliary/plot-*.yaml', these define limits, legends on data-sets etc. * scripts/bench_reporter.py - Produce a .pdf report using the graphs produced by bench_plotter.py - templates/perf_report - This is a report template - The files are specific to the tool rst2pdf * {fio,liburing,spdk}_build.py - Scripts to build and install fio, liburing, and SPDK from source Signed-off-by: Simon A. F. Lund <simon.lund@samsung.com> Co-authored-by: Karl Bonde Torp <karlowich@gmail.com> Signed-off-by: Simon A. F. Lund <simon.lund@samsung.com>
Currently, the 'runs-on' label just uses 'self-hosted', since we only utilize a single "type" of self-hosted runner. However, with the introduction of self-hosted runners for perf. eval, then they need to be distinguishable. This is done by adding descriptive labels 'verify'. Also, renaming the job from 'build-and-test' to 'verify'. Thus, this change, to ensure that the jobs currently using self-hosted runners, won't start using the perf-runners. Signed-off-by: Simon A. F. Lund <simon.lund@samsung.com>
Signed-off-by: Simon A. F. Lund <simon.lund@samsung.com>
This gives an overview of the xNVMe CI environment, that is the jobs that it performs and the resoucres it utilizes to do so. This documentation is not complete, however, it is vastly better than having no documentation, which is what we came from. The file named "xnvme-ci-overview.drawio.png" is a png-file with the diagram embedded inside. This means, that you can view it as a png-file, however you can also edit it draw.io / Diagrams.net Co-authored-by: Karl Bonde Torp <karlowich@gmail.com> Signed-off-by: Simon A. F. Lund <simon.lund@samsung.com>
@karlowich Thanks for all of your feedback. Is there anything I have overlooked / not addressed? |
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 did a final pass and I only have a single comment - which might be irrelevant, in which case feel free to ignore it.
Otherwise everything is looking good.
Items remaining
Add cijoe script patching SPDK to use xNVMe from local repository as defined in cijoe config (xnvme.repository.path
)bench-provision.yaml
andbench-run.yaml
to GitHUB Workflownext
bench
Note: the performance report is available in
artifacts/perf_report/perf.pdf
inside of thebench-results
artifact.