Details
-
Bug
-
Resolution: Unresolved
-
Undefined
-
None
-
4.10.0
-
None
-
Moderate
-
No
-
False
-
Description
Description of problem: Related to closed bugzilla; https://bugzilla.redhat.com/show_bug.cgi?id=1899321
Per documentation: https://docs.openshift.com/container-platform/4.14/scalability_and_performance/cnf-low-latency-tuning.html
"userLevelNetworking is true, the queue count is set to the reserved CPU count for all supported devices. The default is false."
When tested in the CU environment, the queue count would not update to the reserved count until the interfaceName was specified,
"devices is an optional field specifying a list of devices that will have the queues set to the reserved CPU count. If the device list is empty, the configuration applies to all network devices. The configuration is as follows:"
Version-Release number of selected component (if applicable): 4.12.29
How reproducible: Consistent/Easy
Steps to Reproduce:{code:none} 1. Set reserved CPU count via Performance profile CU Example: cpu: isolated: 1-23,25-47,49-71,73-95 reserved: 0,48,24,72 net: userLevelNetworking: true 2. Apply the profile, then review the queue count for the expected interface; CU Example: # ethtool -l ens2f0 Channel parameters for ens2f0: Pre-set maximums: RX: n/a TX: n/a Other: 1 Combined: 96 Current hardware settings: RX: n/a TX: n/a Other: 1 Combined: 96 3. Actual results: No change (queue count not == to reserved)
Expected results: queue count == reserved cpu count (4)
Additional info:
Driver user on host;
- ethtool -i ens2f0
driver: i40e
version: 4.18.0-372.69.1.el8_6.x86_64
firmware-version: 8.60 0x8000be69 1.3140.0
expansion-rom-version:
bus-info: 0000:37:00.0
supports-statistics: yes
supports-test: yes
supports-eeprom-access: yes
supports-register-dump: yes
supports-priv-flags: yes
Works once the interface name is set, which is opposite of the documentation.
net:
userLevelNetworking: true
devices:
- interfaceName: "ens*"