Bug 2076100 - Two VMs with summary CPU more then host CPU are allowed to run on the same host.
Summary: Two VMs with summary CPU more then host CPU are allowed to run on the same host.
Keywords:
Status: CLOSED CURRENTRELEASE
Alias: None
Product: ovirt-engine
Classification: oVirt
Component: BLL.Virt
Version: 4.5.0
Hardware: x86_64
OS: Linux
unspecified
medium
Target Milestone: ovirt-4.5.1
: ---
Assignee: Lucia Jelinkova
QA Contact: Polina
URL:
Whiteboard:
Depends On:
Blocks:
TreeView+ depends on / blocked
 
Reported: 2022-04-17 19:35 UTC by Polina
Modified: 2022-06-23 05:54 UTC (History)
3 users (show)

Fixed In Version: ovirt-engine-4.5.1.2
Doc Type: If docs needed, set a value
Doc Text:
Clone Of:
Environment:
Last Closed: 2022-06-23 05:54:58 UTC
oVirt Team: Virt
Embargoed:
pm-rhel: ovirt-4.5?


Attachments (Terms of Use)
engine log dump xmls (497.25 KB, application/gzip)
2022-04-17 19:35 UTC, Polina
no flags Details


Links
System ID Private Priority Status Summary Last Updated
Github oVirt ovirt-engine pull 299 0 None open core: Refactoring of CPUPolicyUnit 2022-04-22 12:37:12 UTC
Red Hat Issue Tracker RHV-45789 0 None None None 2022-04-17 19:38:29 UTC

Description Polina 2022-04-17 19:35:48 UTC
Created attachment 1873121 [details]
engine log dump xmls

Description of problem:
HE VM with topology 1:4:1 and the dedicated VM 1:15:2 are not allowed to run on the same host 1:16:2 in general. In the described scenario they run together on such host.

Version-Release number of selected component (if applicable):
ovirt-engine-4.5.0.2-0.7.el8ev.noarch

How reproducible: 100%

Steps to Reproduce:
setup with three hosts of topology
CPU(s):              32
On-line CPU(s) list: 0-31
Thread(s) per core:  2
Core(s) per socket:  16
Socket(s):           1
NUMA node(s):        1


1. Run on host3 dedicated VM1 with 30 CPUs 1:15:2 while HE VM with topology 1:4:1 runs on host2 . If we would try to run such dedicated VM1 on the same host2 it is not allowed and this is right . 
2. Configure another dedicated VM1 with the same topology 1:15:2. 
3. Migrate the HE VM. for the migration  it chooses automatically host1 since VM1 runs on the host3 and they are not allowed together. 
4. while migration is processing , try to run this VM2 on host1 - runs successfully
5. wait until the end of migration process.

Actual results:
Migration is completed successfully . HE VM is migrated to the host1 and in this way HE VM with topology 1:4:1 and the dedicated VM with topology 1:15:2 are running on the same host 1:16:2 while it must not be allowed.
If we stop VM2 , it won't be allowed to start on host1.

Expected results:
I think maybe migration must be cancelled in such case to not allow running 34 CPUs on host with 32 CPUS


Additional info: xml after migration and engine log are attached

Comment 1 Arik 2022-04-19 14:57:20 UTC
Lucia, as you are looking at the handling of the pending resources - this one is related but a bit different
I believe that we only count on the reported value from the host when scheduling the dedicated VMs, without taking into consideration pending resources

Comment 2 Lucia Jelinkova 2022-04-22 12:37:13 UTC
Yes, the pending resources have not been taken into account during the filter phase of the scheduling.

Comment 3 Polina 2022-06-19 19:55:15 UTC
verified on ovirt-engine-4.5.1.2-0.11.el8ev.noarch according to the described scenario .
Running while migration process in Step4 is not allowed now: The host host_mixed_1 did not satisfy internal filter CPU because it has an insufficient amount of CPU cores to run the VM.

Comment 4 Sandro Bonazzola 2022-06-23 05:54:58 UTC
This bugzilla is included in oVirt 4.5.1 release, published on June 22nd 2022.
Since the problem described in this bug report should be resolved in oVirt 4.5.1 release, it has been closed with a resolution of CURRENT RELEASE.
If the solution does not work for you, please open a new bug report.


Note You need to log in before you can comment on or make changes to this bug.