[HCI-VN] Migrate router error: Currently in migration state, operation not Allow

Problem Description

Migrate or operate the router. Error: The router is currently in migration state and operation is not Allow!

Warning Information

Possible warning messages:

Effective troubleshooting steps

1. Check the vn-manager-service-rpc service log of the HCI Cluster Controller node (path: /sf/log/today/vn/vn-manager-service-rpc.log). You can see that there is an event decoding error log repeated printing, as shown in the following figure. There are three key error information points:

• Error type and error description are: UnicodeDecodeError: 'utf8' codec can't decode byte … * The error code appears around line 132 of vn_manager_service.rpc_app.v1_0.event.processor.event_base * The operation that directly caused the error was the logger.error statement
  

Root Cause

When using the oslo_log package to print serialized event information, there is a possibility of decoding errors, which ultimately causes the management plane to be unable to process the router role events reported by the data plane.
Solution
Temporary Solution:

1. Connect to the backend of all HCI node in sequence; 2. Delete the alarm node in the vxlan alarm directory on the node;
   rm /tmp/vnet_event/vn-reporter/vxlan_alert/``*
   node the vn-agent container on the node and restart the vn-reporter service;
   # Enter the container
container_exec -n vn-agent
# Restart the service
/sf/vn/etc/init``.d``/vn-reporter restart
2. Enter the database, query the name of the router in the migration state, and delete the migration flag;
   # Connect to the database node
ssh mysql-master
# Enter the container where the database service is located
container_exec -n plat
# Connect to database
mysql --login-path=root -D network
# Query the name of the router being migrated
select id``, name, migrate_host from router where migrate_host != ""``;
# Remove migration flag
update router set migrate_host=``""``;
3. In the network topology of HCI, search by the router name queried in step 4, and edit the **primary Run Location by one to a new node node other than the current primary and backup Run Location). Then, you can observe that the Run Location is restored to one primary and one backup within seconds.
   Permanent Solution:
   680 currently has a technical support patch package, it is recommended to upgrade to 6.8.0R1 and above

Operation Impact Scope

Loss of some vxlan alarm events

Is this a temporary solution?

no

Original Link https://support.sangfor.com.cn/cases/list?product_id=33&type=1&category_id=28257&isOpen=true