Memory usage performance test [15785]

[jparisu]

Merge after:

• 175

• 178

• 172

• 174

• 177

• 176

[codecov[bot]]

Codecov Report

Base: 58.42% // Head: 58.42% // No change to project coverage :thumbsup:

Coverage data is based on head (d7bd7bf) compared to base (74f5197). Patch has no changes to coverable lines.

Additional details and impacted files

```diff @@ Coverage Diff @@ ## main #170 +/- ## ======================================= Coverage 58.42% 58.42% ======================================= Files 33 33 Lines 4421 4421 Branches 2352 2352 ======================================= Hits 2583 2583 Misses 54 54 Partials 1784 1784 ``` Help us with your feedback. Take ten seconds to tell us [how you rate us](https://about.codecov.io/nps?utm_medium=referral&utm_source=github&utm_content=comment&utm_campaign=pr+comments&utm_term=eProsima). Have a feature suggestion? [Share it here.](https://app.codecov.io/gh/feedback/?utm_medium=referral&utm_source=github&utm_content=comment&utm_campaign=pr+comments&utm_term=eProsima)

:umbrella: View full report at Codecov.
:loudspeaker: Do you have feedback about the report comment? Let us know in this issue.

[MiguelCompany]

@EduPonz I think I don't have enough knowledge about bash scripting to review this one ...

[MiguelCompany]

@jparisu Please rebase this one

[EduPonz]

I have rebased the PR while looking at it. I think we ought to have a discussion about this; right now, it's a draft as it is far from being ready to be incorporated in CI and merged. The pain points I see with the current for merging as a test are:

• [ ] The test creates a dependency on a specific Fast DDS example, which may not be present
• [ ] CMakeLists.txt to declare the test for ctest
• [ ] Discussion on passing/failing criteria
• [ ] Script documentation & help guide
• [ ] CSV documentation

Another possibility would be to add the script as a resource from which users can benefit, but I'm really attracted to the idea of having this kind of stress tests.

[jparisu]

Another possibility would be to add the script as a resource from which users can benefit, but I'm really attracted to the idea of having this kind of stress tests.

This test was never meant to be added in the CI. This was intended to be a manual test to have a partial idea of what is the memory usage of an execution. In order to automatize it and have a failing criteria I think it could not be done as results highly depend on the architecture, OS and state of the machine where tests are running. I think it would be better to have it as a helper script rather than an actual test (does Fast DDS have memory usage automatized tests with failing criteria?).

PD: I personally think we have bigger problems that having a stress test.

[jparisu]

This is an image of the results obtained by running this test:

[EduPonz]

I completely agree. I'd leave this as a draft for now and we'll see in the future

[JLBuenoLopez]

@jparisu which is the status of this Draft? Does it make sense to try to merge it? I think we should change the status and assign a milestone so the discussion is had at some point.