search

feedhenry / fh-openshift-templates

FeedHenry MBaaS OpenShift Templates
http://feedhenry.org
Apache License 2.0
1 stars 24 forks source link

⚡ Revert requests & limits to previous defaults #305

Closed grdryn closed 5 years ago

grdryn commented 5 years ago

JIRA: https://issues.jboss.org/browse/RHMAP-21646

grdryn commented 5 years ago

@camilamacedo86

The JIRA is missing the link for the PR

I was making another PR in the fh-core-openshift-templates, and was going to add both at the same time. I have done that now.

What is the issue?

The resource limits and requests were reduced in the templates, and probably some more thought is needed before reducing them.

What is the image used to check the issue?

The productized image from CR2. It's not a problem with the image though.

What test was performed in order to ensure that this PR reverted is the root cause of the issue?

I'm not reverting a PR, I'm reverting some changes from a PR. The 'test' was basically this:

  1. When QE deployed CR2 this morning, it had the changes before this PR (the reduced limits)
  2. It failed
  3. Resetting back to the old defaults (as re-introduced in this PR) fixed it (the deployment was able to succeed).

What test was performed in order to verify that revert this PR will solve the issue?

I don't fully understand this question? Is it the same as the last one? It mentions reverting this PR...I don't think we should revert this PR.

grdryn commented 5 years ago

@jessesarn Thanks for the review!

I'm hesitant to leave the lower values for anything right now, but given that they're configurable (thanks for that!), you can set them to be whatever you want for your use cases. Those numbers would have originally come from load testing of the platform (and yes, the values could probably even be considered conservative after that). I don't believe we've gotten any other requests to change them, but if it's a common ask, then we should definitely consider it in future (and maybe do further load testing at that point to see what we can bring them down to).

camilamacedo86 commented 5 years ago

Hi @grdryn.

According to the comment in the JIRA shows that @austincunningham have some chat internally and get the required info and could reproduce and verify it. Then, I don't think is required another professional retest it and I am ok with the solution.

However, I believe that if still required the need to reduce this values than the original task should be re-opened or create another JIRA for we are able to check it.

c/c @jessesarn