Troubleshooting the upgrade

The following table lists common issues you might encounter during the 6.3.0 upgrade, along with their causes and fixes.

Symptom	Cause	Fix
Upgrade fails immediately with an EUA error.	global.acceptEUA isn't set to true.	Add --set global.acceptEUA=true to the upgrade command.
preupgrade-dbmigrate job fails, upgrade aborts.	Usually a database connectivity issue, or an incorrect initDBUserName or initDBUserPassword.	Check the job pod logs, which are retained on failure.
Hook reports 0 rows migrated.	Either global.domainSuffix does not match the suffix used at install time, or the stored JSON format differs from what REPLACE() expects.	With no MySQL client on the host, use a throwaway pod with the toolkit image to confirm the rows the hook will migrate. Refer to the baseline database check below: kubectl run dbcheck --rm -i --restart=Never --image=<registry>/apim-le-toolkit:v6.3.0 --image-pull-policy=Never --env="MYSQL_PWD=<dml-password>" -- "mysql -h <db-host> -P<db-port> -u<dml-user> -D<schema> --ssl=OFF -e "SELECT domain FROM system_domain_endpoints WHERE domain LIKE '%:7000';""
Trafficmanager or configui pods never become ready after the upgrade.	The full loader job has not run.	Run the full data load step.
New pods show ImagePullBackOff.	The 6.3.0 images are not present, or the apimImageRegistry or imagePullPolicy is incorrect.	Verify the image registry, pull policy, and that 6.3.0 images are available.