Note: This bug is displayed in read-only format because the product is no longer active in Red Hat Bugzilla.

Bug 1431667

Summary:	Pod metrics should display memory cache information
Product:	OpenShift Container Platform	Reporter:	Matthew Robson <mrobson>
Component:	Hawkular	Assignee:	Ruben Vargas Palma <rvargasp>
Status:	CLOSED WONTFIX	QA Contact:	Peng Li <penli>
Severity:	medium	Docs Contact:
Priority:	medium
Version:	3.4.1	CC:	adeshpan, amuller, aos-bugs, arnaud, dapark, dluong, erich, erjones, gekis, jaliang, jokerman, maszulik, mmccomas, mwringe, nstielau, pweil, rhowe, rludva, ruben.vargasjr, rvargasp, spadgett, spasquie, surbania, tkatarki, travi
Target Milestone:	---	Keywords:	Reopened
Target Release:	3.11.z	Flags:	nstielau: needinfo+ jaliang: needinfo-
Hardware:	Unspecified
OS:	Unspecified
Whiteboard:
Fixed In Version:		Doc Type:	If docs needed, set a value
Doc Text:		Story Points:	---
Clone Of:		Environment:
Last Closed:	2021-07-06 19:19:32 UTC	Type:	Bug
Regression:	---	Mount Type:	---
Documentation:	---	CRM:
Verified Versions:		Category:	---
oVirt Team:	---	RHEL 7.3 requirements from Atomic Host:
Cloudforms Team:	---	Target Upstream Version:
Embargoed:

Description Matthew Robson 2017-03-13 14:47:22 UTC

Description of problem:

See: https://github.com/openshift/origin-web-console/issues/1315

Version-Release number of selected component (if applicable):

3.4

How reproducible:

Always

Steps to Reproduce:
1. View usage in the console

Actual results:

Usage is higher than anticipated

Expected results:

Accurate usage

Comment 2 Samuel Padgett 2017-03-13 15:32:49 UTC

Matt, is this data available to us from Heapster and Hawkular?

Comment 3 Matt Wringe 2017-03-14 19:23:21 UTC

This information is not currently available from the current Heapster version that we are using (1.2.0). For Heapster 1.3.0 I believe this information will be made available, but its currently not yet released.

Comment 4 Samuel Padgett 2017-03-17 17:04:59 UTC

Transitioning to the Metrics component for now. We can make the web console updates when the data we need is in Hawkular.

Comment 5 Matt Wringe 2017-04-25 20:17:21 UTC

@sross: do you know if the summary endpoint is going to expose the memory cache information? Heapster can currently gather it from the 'kubernetes' source but we are currently using the 'kubernetes.summary_api' one instead (where this metric is not exposed).

Comment 6 Matt Wringe 2017-08-03 16:13:07 UTC

I am closing this issue as deferred as Heapster is being deprecated.

Comment 7 Eric Jones 2018-03-27 18:12:15 UTC

I am reopening this bug as heapster/hawkular is still the primary metrics tool, and the replacement for it has not been officially GA'd.

As such, if any of the applications in a cluster use a lot of cached memory then the memory being reported by metrics is not reliable enough.

@Engineering, I understand that we plan to deprecate heapster/hawkular at a later point, but as it is still the primary metrics tooling, we really need to try and address this issue.

Comment 11 Ryan Howe 2018-04-17 21:34:29 UTC

Same as this bug right:  https://bugzilla.redhat.com/show_bug.cgi?id=1457933

Comment 15 Radomir Ludva 2019-06-22 15:57:43 UTC

Hello, customers still meet this issue. 

Can you please advice any workaround or hint: How to determine the actual usage for the purpose of daily memory usage monitoring and for memory based autoscaling capability?

Comment 21 Jace Liang 2021-06-28 06:48:20 UTC

Hi, is theres any plan to fix this? My CU at OCP 3.11 have this issue.
Or provide any workaround method for this?
Thank you.

Comment 26 Matt Wringe 2021-07-06 19:19:32 UTC

This is not a bug but a feature request to add additional metrics that are not currently being collected or exposed to users. The metrics value we use is the total memory which is the combined value of the cache and rss memory.

OpenShift 3 is in maintenance and we will not be adding support for new cache and rss memory metrics.