Some of the virtual services go down during an upgrade, leading to traffic disruptions.
book
Article ID: 405629
calendar_today
Updated On:
Products
VMware Avi Load Balancer
Issue/Introduction
VS placement may fail for some VSs with the error "Failed to add VS to the interface".
The VSs affected by this will be down and not processing traffic.
Environment
Avi NSX-T deployments with SEs created on versions lesser than 30.2.1.
Cause
The customer environments, where the "nsxt_no_hotplug" is set as False in the SE data vNics, might experience issues like this because of a bug in Cloud Connector.
All SEs created before v30.2.1 will have the "nsxt_no_hotplug" set to False by default. Hence, they are vulnerable to this.
The cloud connector ends up setting a valid NIC as delete pending due to non-deterministic behavior of the encoding json library and removing it in the subsequent delete vNic request from its runtime, causing traffic disruption after an upgrade.
Resolution
Bug ID: AV-237208
Details: Some of the virtual services go down during an upgrade, leading to traffic disruptions.
Workaround: Run the script fix_sevm_runtime.py shipped with the latest version, 31.2.1+, of the controller BEFORE upgrading the controller.
To execute the script, SSH into the controller.
The script can be run as using the command below:
$ sudo -i
# cd /opt/avi/scripts
# python3 fix_sevm_runtime.py -h
Usage: fix_sevm_runtime.py [options] arguments
Options:
-h, --help show this help message and exit
--se-name=SE_NAME Name of the Service Engine you want to fix.
--fix Fix the SE VM Runtime configuration.
--dry-run Do a dry run and list the fixed SE VM Runtime without
saving. (default: True)
# python3 fix_sevm_runtime.py --se-name=<se-name> [--fix]
Note: If you are running into this issue on a version less than 31.2.1, please reach out to support and we can assist you accordingly.