Note: This bug is displayed in read-only format because the product is no longer active in Red Hat Bugzilla.

Bug 2024575

Summary:	[RHV 4.4] In RHV with SAP HANA, Make NUMA work with hugepages with vdsm 4.30
Product:	Red Hat Enterprise Virtualization Manager	Reporter:	Chetan Nagarkar <cnagarka>
Component:	vdsm	Assignee:	Milan Zamazal <mzamazal>
Status:	CLOSED ERRATA	QA Contact:	Polina <pagranat>
Severity:	high	Docs Contact:
Priority:	urgent
Version:	4.4.3	CC:	ahadas, emarcus, ljelinko, lsurette, lsvaty, mavital, michal.skrivanek, mtessun, mzamazal, sfroemer, srevivo, ycui
Target Milestone:	ovirt-4.3.21	Keywords:	ZStream
Target Release:	---
Hardware:	x86_64
OS:	Linux
Whiteboard:
Fixed In Version:	vdsm-4.30.52	Doc Type:	Enhancement
Doc Text:	With this enhancement, when scheduling a virtual machine with pinned NUMA nodes, memory requirements are calculated correctly by taking into account the available memory as well as hugepages allocated on NUMA nodes. This enhancement applies only to Red Hat Virtuaization 4.3 hosts running VDSM; a Manager with version 4.4 or later is needed to take advantage of this enhancement.	Story Points:	---
Clone Of:		Environment:
Last Closed:	2021-12-16 20:03:20 UTC	Type:	Bug
Regression:	---	Mount Type:	---
Documentation:	---	CRM:
Verified Versions:		Category:	---
oVirt Team:	Virt	RHEL 7.3 requirements from Atomic Host:
Cloudforms Team:	---	Target Upstream Version:
Embargoed:
Bug Depends On:	1812316
Bug Blocks:

Description Chetan Nagarkar 2021-11-18 11:38:54 UTC

Description of problem:

Currently, the engine is unaware of hugepages per numa on RHVH 4.3 running under RHV 4.4 manager [Cluster compatibility 4.3]. 
VDSM 4.30 is not reporting hugepages per NUMA node.

It seems we already have a patch that makes 4.4 engine support hugepages per NUMA node, and perform validation with it: https://gerrit.ovirt.org/c/108190/ 
hence this works fine with vdsm 4.40.

Version-Release number of selected component (if applicable):
RHV 4.4.3
vdsm 4.30

Actual results:
VM refuses to start with a "General command validation error."


Expected results:


Additional info:

Workaround:

1] On RHV Portal, Go to Administration -> Configure.

2] In the scheduling policies subtab, select the 'none' scheduling policy and click on 'Copy'

3] In the Clone scheduling policy window, remove/disable the NUMA from the first window (filters), save with another name. 

4] then configure the cluster of that host to use that new policy.

Comment 13 Sandro Bonazzola 2021-12-02 08:33:12 UTC

Not ready for 4.3.20 inclusion, please re-schedule.

Comment 18 Polina 2021-12-15 15:58:29 UTC

the setup for verification  
vdsm version vdsm-4.30.52-1.el7ev.x86_64
engine version ovirt-engine-4.4.9.5-0.1.el8ev.noarch

In engine 4.4: create DC with compatibility version 4.3, cluster 4.3,  installed host4.3 (vdsm-4.30.52-1).

Steps:

1. Configure VM_1 with the memory close to the host memory to create load. Start VM_1 on the host supporting NUMA.
2. Add two vNuma nodes to another VM_0 and configure this VM with memory more than max free memory remaining on the host for scheduling new VMs. Check that this VM could not be started on the same host
3. Configure VM_0 with hugepages and start again. The VM is started on the same host successfully since the host has free hugepages to accommodate the VM.

the verification was based on the existing automation test (according to https://polarion.engineering.redhat.com/polarion/#/project/RHEVM3/workitem?id=RHEVM-27430).

Comment 20 errata-xmlrpc 2021-12-16 20:03:20 UTC

Since the problem described in this bug report should be
resolved in a recent advisory, it has been closed with a
resolution of ERRATA.

For information on the advisory (vdsm bug fix and enhancement update), and where to find the updated
files, follow the link below.

If the solution does not work for you, open a new bug report.

https://access.redhat.com/errata/RHBA-2021:5203