What should I watch out for when taking a device in the cluster into maintenance?

Short answer

Before a device in the cluster is taken into maintenance, its traffic is transferred to the peer device in a planned way. First, it is verified that the Cluster Status row shows Cluster connection is OK, that the Sync. Summary row shows Synced, and that the peer device can carry the vServices. Then, on the Network → Interfaces screen, the VIPs of the maintenance device are turned off with Disable for <device name>; the VIPs stay on the peer device and do not return during the maintenance. When the maintenance is over, the VIPs are turned on at a planned, low-traffic moment with Enable for <device name> in the same menu. The Interface IP in the VIP Communication pair must not be turned off; otherwise, the VIPs on that interface do not become active on either device (best practices).

Problem: Performing VIP Transitions in a Planned Way During Maintenance

A device in the cluster is to be taken out of service for an update, a hardware change or a restart. If the device is shut down without planning, the VIP transitions happen at unplanned moments, and when the device is turned back on, the VIPs change places again; an operation that affects both devices at once (e.g. a Down that turns off the interface on both devices), on the other hand, stops the service. When the transitions are performed in a planned way, the traffic stays on the peer device during the maintenance.

Solution: Three Methods

Your situation	Suitable method
Transferring the traffic in a controlled way (recommended)	Method 1: Disable for <device name> on the VIPs
Turning off only one interface on that device	Method 2: Details → Down
Restarting or shutting down the device	Method 3: Local Services

Method 1: Turning Off the VIPs for the Maintenance Device Recommended

• How it works The VIP is removed from the maintenance device and runs fixed on the peer device; even if the maintenance device restarts, the VIP does not return to it.
• Where it is openedNetwork → Interfaces → right-click on the VIP → Disable for <device name>
• Requirement The cluster connection must be up
• vService type All vService types

Step-by-Step Configuration

1

Adım 1 Check the cluster health

On the Settings → Device Management → HA Clustering & Sync screen, Cluster connection is OK must be seen in the Cluster Status row and Synced in the Sync. Summary row. In the vService list, the half circle of the peer device must be green; if it is yellow or red, those vServices are affected when the VIP moves (What does a status circle split in two mean in the vService list in a cluster?). The VIPs that appear with the (M) suffix on the Network → Interfaces screen are the ones that will be moved from the maintenance device.

2

Adım 2 Turn off the VIPs

On the Network → Interfaces screen, the VIP address is right-clicked and Disable for <maintenance device name> is selected (Right-clicking an IP address). For multiple VIPs, the interface row is clicked, the VIPs are selected in the IP tab of the preview panel, and Actions → Disable for <maintenance device name> is used. This is repeated for each interface. The Interface IPs in the VIP Communication pair are not turned off.

3

Adım 3 Perform the maintenance

If necessary, the device is restarted or shut down with Reboot the Device or Shut Down the Device in its own column on the Settings → Device Management → Local Services screen. No configuration change is made during the maintenance.

4

Adım 4 Verify the cluster and synchronization

When the device is turned on, Cluster connection is OK and Synced are expected on the HA Clustering & Sync screen. If synchronization is needed, Synchronize is run on the device that has the current configuration; running it on the device returning from maintenance writes the old configuration over the peer (Is the configuration automatically synchronized between cluster members?).

5

Adım 5 Turn on the VIPs at a planned moment

The VIPs are turned on from the same menu with Enable for <device name>. If the device is Master on these VIPs, the VIP returns to it at this moment; for this reason, a low-traffic moment is chosen for the return.

Verification

1During maintenanceTR7 side

On the Network → Interfaces screen, it is seen that the VIP is Disabled in the maintenance device column and active in the peer device column; in the vService list, it is seen that the half of the maintenance device is gray.

ExpectedThe VIP is on the peer device

2After maintenanceTR7 side

After the VIPs are turned on, the two columns and the vService half circles are checked.

ExpectedMaster VIP is active · Slave VIP · both halves green

Method 2: Turning Off the Interface Only on the Maintenance Device

• How it works When the interface is turned off on only one device, the VIPs on that interface move to the peer device.
• Where it is openedNetwork → Interfaces → interface → Details tab → Down
• Requirement The cluster connection must be up
• vService type All vService types

Step-by-Step Configuration

1

Adım 1 Check the cluster health

On the Settings → Device Management → HA Clustering & Sync screen, Cluster connection is OK must be seen in the Cluster Status row and Synced in the Sync. Summary row. In the vService list, the half circle of the peer device must be green; if it is yellow or red, those vServices are affected when the VIP moves (What does a status circle split in two mean in the vService list in a cluster?). The VIPs that appear with the (M) suffix on the Network → Interfaces screen are the ones that will be moved from the maintenance device.

2

Adım 2 Turn off the interface

The interface row is clicked; in the Details tab of the preview panel, Down above the column of the maintenance device is used. Up/Down in the right-click menu and in the panel header affect both devices at once; they are not used during maintenance.

3

Adım 3 Turn it back on

When the maintenance is over, Up is used from the same place. Down is not permanent: TR7 turns the down interface back on at the next application of the network configuration (Up/Down). The interface that carries the management access IP and the Sync Interface cannot be turned off.

Verification

1VIP statusTR7 side

On the Network → Interfaces screen, the VIPs of the interface are checked.

ExpectedSlave VIP is active

Method 3: Restarting or Shutting Down the Device

• How it works When the device shuts down, the VRRP advertisements stop and the VIPs on which it is Master move to the peer device.
• Where it is openedSettings → Device Management → Local Services
• Requirement The cluster connection must be up
• vService type All vService types

Step-by-Step Configuration

1

Adım 1 Check the cluster health

On the Settings → Device Management → HA Clustering & Sync screen, Cluster connection is OK must be seen in the Cluster Status row and Synced in the Sync. Summary row. In the vService list, the half circle of the peer device must be green; if it is yellow or red, those vServices are affected when the VIP moves (What does a status circle split in two mean in the vService list in a cluster?). The VIPs that appear with the (M) suffix on the Network → Interfaces screen are the ones that will be moved from the maintenance device.

2

Adım 2 Restart or shut down

The Local Services screen shows the two devices in separate columns; Reboot the Device or Shut Down the Device in the column of the maintenance device is confirmed. The VIPs on which it is Master move to the peer device according to the VIP Transition Duration (s) in the Edit Interface form.

3

Adım 3 Take the return into account

In this method, when the device is turned on, the VIPs on which it is Master return to it on their own; this is also a VIP transition (When a failed device comes back, does the VIP automatically move back to it?). If the moment of return needs to be controlled, Method 1 is used.

Verification

1VIP statusTR7 side

While the device is off, the VIP status is checked on the Network → Interfaces screen.

ExpectedSlave VIP is active

Common Mistakes

• Turning off the Interface IP in the VIP Communication pair. If one address of the pair is off, the VIPs on that interface do not become active on either device.
• Using the Up/Down that affects both devices. Down in the right-click and panel header menus turns off the interface on both devices at once.
• Disabling the vService. The vService configuration is shared on both devices; Disable turns off the service, not a device.
• Confusing it with the Backend Service maintenance mode. Enable maintenance takes a Backend Service out of distribution, not the TR7 device (What does Maintenance Mode do?).
• Running Synchronize on the device returning from maintenance. Synchronize sends the configuration from this device to the peer; the old configuration is written over the peer.

How it is done in the documentation

• Interfaces · Best Practices
• Right-clicking an IP address
• Local Services · Reboot the Device

Related questions

• When a failed device comes back, does the VIP automatically move back to it?
• What does a status circle split in two mean in the vService list in a cluster?
• Is the configuration automatically synchronized between cluster members?