Skip to content
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

scaling: add used memory to PDF report table #279

Open
wants to merge 1 commit into
base: master
Choose a base branch
from

Conversation

askervin
Copy link
Contributor

Currently the PDF report shows how much scaling up consumes free
memory. This number is not comparable between cluster nodes or even
test runs because RAM used for OS caches/buffers/slab is counted as
consumed. As a consequence, consumed free memory depends heavily on
initial memory conditions of a node, instead of used memory by the k8s
and pods. This patch adds "memory used" to the report in order to have
less node-dependent and more reproducible memory figure.

Using /proc/meminfo "MemAvailable" was also tried out, but it varies
almost like "MemFree" that is currently reported.

Signed-off-by: Antti Kervinen antti.kervinen@intel.com

@grahamwhaley
Copy link

Looking good @askervin . Code looks good, and it ran and made a table for me... a couple of questions:

  • should we also plot the used data on the graph? I suspect so, as having both used and and free in the table but not in the graph does not feel correct.
  • we probably need to update the test text in the top level .Rmd file as well to match the changes (to say we are measuring not just free memory any more).

I also suspect once we have settled on which result is 'best' (most representative of theoretical maximum pod density), we should also note that in the .Rmd file text to make it clear which number to grab from the report...

@askervin
Copy link
Contributor Author

Good points, thanks @grahamwhaley. I added the memory used plot. Now the page looks more complete, and the title offered a nice place to explain what "used" actually means. (That is, why $\delta$Free > $\delta$Used.)

A note on grabbing used memory stats added to Rmd as well.

Copy link

@grahamwhaley grahamwhaley left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

lgtm
I see the graphs and data. Tested with 1x1k and 2x2k datasets.
I'm hoping once we merge/meld along with #280 then it will look even nice :-)

Currently the PDF report shows how much scaling up consumes free
memory. This number is not comparable between cluster nodes or even
test runs because RAM used for OS caches/buffers/slab is counted as
consumed. As a consequence, consumed free memory depends heavily on
initial memory conditions of a node, instead of used memory by the k8s
and pods. This patch adds "memory used" to the report in order to have
less node-dependent and more reproducible memory figure.

Using /proc/meminfo "MemAvailable" was also tried out, but it varies
almost like "MemFree" that is currently reported.

Signed-off-by: Antti Kervinen <antti.kervinen@intel.com>
@askervin
Copy link
Contributor Author

askervin commented Dec 5, 2019

rebased, commits squashed, tested, ready for merge.

@grahamwhaley
Copy link

@marcemq - PTAL

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
Projects
None yet
Development

Successfully merging this pull request may close these issues.

2 participants