On Fri, Jan 4, 2019 at 5:31 PM Daniel Wagner wagi@monom.org wrote:

Hi Chase,

...

...

I figure all those tests relegated to preempt-rt need some love since the parameters used for cyclictest are not really recommended. I've started to change a few things [1] and would like to contribute them back, but I don't know what the correct way is.

Thanks for contributing back. I believe you are the best person to determine what params should be used for these tests.

I plan to review the existing test configuration and then expand the -rt testing. In order to verify the -rt code is working correctly is also to create various loads and check if we see any spikes in cyclictest. One thing I couldn't figure out so far, if it is possible to combine tests. That is I would like to run stress-ng and at the same time cyclictest.

It is ok to run some workloads in the background while running cyclictest. However, all the background workloads should be stopped or killed properly when the main test script exists, otherwise the background processes will block the whole LAVA test action from exiting until timeout reached.

I tried to run 'stress-ng hdd' when running network stress testing. See https://git.linaro.org/qa/test-definitions.git/tree/automated/linux/24h-stre.... When I look at it now, I feel it is actually not the best practice. A cleanup function should be defined to kill the background process when trap captured abnormal exit signals.

...

...

Should I send patches here, or do I need to do PR ong it.lavasoftware.org?

test-definitions is not part of lavasoftware. It is not hosted on the gitlab instance either. https://git.linaro.org/qa/test-definitions.git is the upstream and we use gerrit to manage patches.

Thanks, I started to setup my account and uploaded an ssh key. Let's see if I get my head around gerrit again. It's a while since I had to work with it :)

Thanks for doing so. Dan proposed to move the project to github. A ticket has been raised. If we managed to move the collaboration to github, a notification will be sent to this ML.

Thanks, Chase

...

You should be able to register as a community contributor from here https://register.linaro.org/. Once your account is active and your SSH key added to https://login.linaro.org/, you should be able to upload patches to review.linaro.org with 'git review' command. If you have difficult in going through these steps, please just send your patches here. We will get them reviewed/merged.

Great!

Thanks, Daniel

On Sat, Jan 5, 2019 at 9:13 PM Novello Giampiero novellogp64@gmail.com wrote:

SORRY , I would like to use zephyr and to test it via lava. There is an application note that could help me ?

Off the topic? Here is the help page https://master.lavasoftware.org/static/docs/v2/monitors.html?highlight=zephy...

Best Regards Novello G.

Il giorno ven 4 gen 2019 alle ore 15:16 Chase Qi chase.qi@linaro.org ha scritto:

...

On Fri, Jan 4, 2019 at 5:31 PM Daniel Wagner wagi@monom.org wrote:

...

Hi Chase,

...

...

I figure all those tests relegated to preempt-rt need some love since the parameters used for cyclictest are not really recommended. I've started to change a few things [1] and would like to contribute them back, but I don't know what the correct way is.

Thanks for contributing back. I believe you are the best person to determine what params should be used for these tests.

I plan to review the existing test configuration and then expand the -rt testing. In order to verify the -rt code is working correctly is also to create various loads and check if we see any spikes in cyclictest. One thing I couldn't figure out so far, if it is possible to combine tests. That is I would like to run stress-ng and at the same time cyclictest.

It is ok to run some workloads in the background while running cyclictest. However, all the background workloads should be stopped or killed properly when the main test script exists, otherwise the background processes will block the whole LAVA test action from exiting until timeout reached.

I tried to run 'stress-ng hdd' when running network stress testing. See https://git.linaro.org/qa/test-definitions.git/tree/automated/linux/24h-stre.... When I look at it now, I feel it is actually not the best practice. A cleanup function should be defined to kill the background process when trap captured abnormal exit signals.

...

...

...

Should I send patches here, or do I need to do PR ong it.lavasoftware.org?

test-definitions is not part of lavasoftware. It is not hosted on the gitlab instance either. https://git.linaro.org/qa/test-definitions.git is the upstream and we use gerrit to manage patches.

Thanks, I started to setup my account and uploaded an ssh key. Let's see if I get my head around gerrit again. It's a while since I had to work with it :)

Thanks for doing so. Dan proposed to move the project to github. A ticket has been raised. If we managed to move the collaboration to github, a notification will be sent to this ML.

We have moved the project to https://github.com/Linaro/test-definitions

- Chase

...

Thanks, Chase

...

...

You should be able to register as a community contributor from here https://register.linaro.org/. Once your account is active and your SSH key added to https://login.linaro.org/, you should be able to upload patches to review.linaro.org with 'git review' command. If you have difficult in going through these steps, please just send your patches here. We will get them reviewed/merged.

Great!

Thanks, Daniel

---

linaro-validation mailing list linaro-validation@lists.linaro.org https://lists.linaro.org/mailman/listinfo/linaro-validation

---

linaro-validation mailing list linaro-validation@lists.linaro.org https://lists.linaro.org/mailman/listinfo/linaro-validation

It is ok to run some workloads in the background while running cyclictest. However, all the background workloads should be stopped or killed properly when the main test script exists, otherwise the background processes will block the whole LAVA test action from exiting until timeout reached.

I tried to run 'stress-ng hdd' when running network stress testing. See https://git.linaro.org/qa/test-definitions.git/tree/automated/linux/24h-stre.... When I look at it now, I feel it is actually not the best practice. A cleanup function should be defined to kill the background process when trap captured abnormal exit signals.

Thanks for the pointer. I'll give this try tomorrow and try to work on your idea.

Thanks, Daniel