All the nodes of a NetScaler cluster must be running the same software version. Therefore, to upgrade or downgrade the cluster, you must upgrade or downgrade each NetScaler appliance of the cluster, one node at a time.
A node that is being upgraded or downgraded is not removed from the cluster. The node remains a part of the cluster and serves traffic uninterrupted, except for the downtime when the node reboots after it is upgraded or downgraded.
When you upgrade or downgrade the first cluster node, configuration propagation is disabled automatically when there is a cluster version mismatch among the cluster nodes. Configuration propagation is enabled only after you upgrade or downgrade all the cluster nodes. You cannot perform any configuration through the cluster IP address when configuration propagation is disabled.
The following table explains when the configuration propagation is disabled during the upgrade or downgrade of the first cluster node.
| Upgrade/Downgrade |
From version |
To version |
Description |
| Upgrade |
13.1 Build 21.50 or earlier |
13.1 Build 24.38 or later |
Configuration propagation is disabled on the cluster after the cluster node reboots and comes up. Note: You must not perform any configurations through the cluster IP address while the node reboots. |
| Upgrade |
13.1 Build 24.38 or later |
13.1 Build 27.59 or later |
Configuration propagation is disabled on the cluster just before the node reboots. |
| Downgrade |
13.1 Build 24.38 or later |
13.1 Build 21.50 or earlier |
Configuration propagation is disabled on the cluster after the cluster node reboots and comes up. Note: You must not perform any configurations through the cluster IP address while the node reboots. |
| Downgrade |
13.1 Build 27.59 or later |
13.1 Build 24.38 or later |
Configuration propagation is disabled on the cluster just before the node reboots. |
You can also verify the status of command propagation using the following command:
show cluster instance
-
In a cluster setup with the maximum connection (maxConn) global parameter set to a non-zero value, CLIP connections might fail if any of the following conditions is met:
Workarounds:
-
Before upgrading a cluster setup from NetScaler 13.0 76.x build to NetScaler 13.0 79.x build, the maximum connection (maxConn) global parameter must be set to zero. After upgrading the setup, you can set the maxConn parameter to a desired value and then save the configuration.
-
NetScaler 13.0 76.x build is not suitable for cluster setups. Citrix recommends not to use the NetScaler 13.0 76.x build for a cluster setup.
-
In a cluster setup, a NetScaler appliance might crash, when:
-
upgrading the setup from NetScaler 13.0 47.x or 13.0 52.x build to a later build, or
-
upgrading the setup to NetScaler 13.0 47.x or 13.0 52.x build
Workaround: During the upgrade process, perform the following steps: