First time application-apply stx-openstack failed on vbox due to timeout from unbalanced CPU load
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
StarlingX |
Fix Released
|
High
|
David Sullivan |
Bug Description
This issue started to appear around April 15.
Env: VBOX
Configuration: AIO-SX
Problem: system application-apply fails during the neutron phase (compute-kit charts).
Diagnosis: None of the pods show any indication of Crashing. The pods (nova, neutron, openvswitch, nova-api-proxy, libvirt) eventually come up. The problem is that the timeout occurs. openvswitch times out after 15 minutes because not all those pods came up fast enough.
Troubleshooting:
running top and pressing "1" will show the cpu load distribution.
The platform processes should be running across CPU 0 and 1. However they are all affined to only one CPU.
This leads to a load average of over 50, and eventually a timeout.
Theories:
cat /proc/cmdline shows empty value for isolcpus
Additional Notes:
I didn't use the bug reporting template.
Changed in starlingx: | |
status: | New → Triaged |
Changed in starlingx: | |
assignee: | Jim Gauld (jgauld) → David Sullivan (dsullivanwr) |
tags: | added: stx.config |
Marking as release gating. Issue seen on virtual environments and appears related to https:/ /review. openstack. org/#/c/ 648511/