On upgrade from version 20.1.7 onwards to patch versions 22.1.2-2px, the SE upgrade on Azure cloud fails because certain set of instance types are not supported by Azure anymore.

Products

• VMware Avi Load Balancer

Issue/Introduction

On upgrade from version 20.1.7 onwards to patch versions 22.1.2-2px, the SE upgrade on Azure cloud fails because certain set of instance types are not supported by Azure anymore.

 

 Additionally, the SE upgrade fails with the error "Error while performing SE startup" and issues arise during the calculation of SE Flavor Bandwidth. The following errors are observed in the se_supervisor.log:

[2023-01-11 05:37:05,335] ERROR [se_supervisor.run:2342] ^[[31mError while performing SE startup^[[0m
Traceback (most recent call last):
  File "/opt/avi/scripts/se_supervisor.py", line 2339, in run
    se_startup(cluster_state.leader, cluster_state.local_ip)
  File "/opt/avi/scripts/se_supervisor.py", line 1711, in se_startup
    init_hw_properties(g_se_group.connection_memory_percentage,
  File "/opt/avi/scripts/se_supervisor.py", line 1609, in init_hw_properties
    myfile.write('SE Flavor Bandwidth: %d\n' %(2**math.ceil(math.log2(nw_bandwidth_mbps/1000.0))))
ValueError: math domain error
[2023-01-11 05:37:05,336] INFO [se_supervisor.safe_exit:2382]
[2023-01-11 05:37:05,336] INFO [se_supervisor.safe_exit:2384] Sigkill self, forcibly terminating
[2023-01-11 05:37:06,855] INFO [se_supervisor.<module>:2729]

Environment

Azure Cloud.

Cause

• One of the unsupported instance types was configured in the SE group during the setup.

 

• The instance type configured in the SE group was previously supported in an earlier release,but it is no longer supported in the latest patch release. The issue arose because the corresponding flavor type was not included in flavours.json (maintained by the CC) or in cloud_instances.json (used by the SE for auto-RSS).

 

• As a result, the missing entry in cloud_instances.json caused the se_supervisor to fail during startup.

Resolution

Temporary Workaround:

To address the issue, the instance_type: bandwidth entry must be manually added to the cloud_instances.json file on the controller. This action should be performed with support assistance to ensure proper configuration and avoid any potential misconfigurations

Permanent Fix:

Perform the upgrade to the following versions where the fix has been applied:

 

22.1.3
22.1.2-2p5
30.1.1