Forcing a VM to Boot
You can force a VM to boot using the Force Boot button on the VIRTUAL MACHINES page. However, the Force Boot button is active only when the everRun Availability Console reports that the partner node is powered off or otherwise unreachable. When you use Force Boot to bring a VM online, you manually by-pass the system's safety checks to protect data, so you must use Force Boot with extreme caution and with full understanding of the conditions and consequences of using it.
Caution:
Before using Force Boot, read this entire topic and consult with your authorized Stratus service representative. The service representative can review your system, including the date of the last volume synchronization, and can then discuss with you the full impact of using Force Boot. Then, you can decide, with your service representative, whether or not to force a VM to boot.
When you force a VM online with Force Boot, you select a node (that is, the node that is reachable) on which to force the VM to boot. All data on that node is marked as valid, regardless of the actual condition of the data (for example, the data's state, the last synchronization, the condition of the volume, etc.).
During the Force Boot process, the VM's volumes are tagged with the date and time that the force-boot process was initiated. The VM's AX components (that is, the VM's AX pair) use the data on the VM's volumes and communicate the status of that data to determine which AX contains the up-to-date volume information. The Force Boot process overrides the built-in logic that protects a VM from running in a split-brain condition. If the AX pair cannot communicate, a split-brain condition occurs and damages data integrity (for information on the split-brain condition, see Creating a SplitSite Configuration).
Warnings: Do not use
Force Boot in the following situations:
- One or more volumes is the target of an unfinished mirror copy on the node where you will perform Force Boot.
- A target of an unfinished mirror copy is not good and will not be available even with Force Boot.
- The volumes are not synchronized. The following two situations are examples:
- Both of the VM's AXs must have access to all of the VM's data volumes.
- On a system with multiple volumes, the VM needs both AXs to be running in order for the VM to have access to all of its volumes because each node has a green-checked copy of a different volume, and the volume's mirror copy on the opposite node is not green-checked.
- Both nodes are required because multiple VMs are degraded, yet are green-checked on opposite nodes (for example, Node0 has a good boot volume but a bad data volume, while Node1 has a bad boot volume but a good data volume).
- The system is licensed for one node.
If you perform a Force Boot on a system with outdated volumes, contact your authorized Stratus service representative immediately. If both nodes are powered on and have started to synchronize data, the system uses data from the VM that you forced to boot, and you cannot recover the data on the node that was unreachable.
In some circumstances, however, you might be able to recover data after you use Force Boot on a system with outdated volumes:
-
If the unreachable node is still powered off, do not power it on.
-
If the unreachable node was powered off before you clicked Force Boot, then the VM's AX on the powered-off node is preserved and you can reverse the Force Boot without data loss under the following conditions:
- The VM that you forced to boot does not have new data (that is, the VM has not been put in production).
- Before you forced the VM to boot, the VM's AX on the unreachable node did not exchange status with the AX of the VM that you will force to boot.
- The issue preventing the VM's AX on the unreachable node from booting is resolved.
- All VM data between the two nodes is accurately synchronized. The system has no VMs where, of each VM's two AX components, the data of the VM's AX on one node is in a different state from the data of the VM's AX on the other node.
If your system meets all of the conditions above, contact your authorized Stratus service representative to advise you on a recovery process.
If you have decided to force a VM to boot, be sure to prepare for it by performing the prerequisite procedures.
Prerequisites:
-
Manually check all volumes to ensure that you can safely override them. For example, the volume state should be green-checked, and disk synchronization should be finished.
-
Determine if both AX components of the VM can communicate and can allow the system processes to determine the state of each volume. To prevent a split-brain condition, you must ensure that the two AX components of the VM can communicate status and can determine which AX has good data and good boot volumes.
-
Ensure that the system is licensed for two nodes.
- Contact your authorized Stratus service representative.
To force a VM to boot
After you have consulted with your authorized Stratus service representative, and you have decided to force a VM to boot, perform the following procedure. In the examples, node0 is offline, node1 is the primary, and VM-1 is stopped.
- In the everRun Availability Console of a system licensed for two nodes, click Virtual Machines in the left panel.
- Navigate to the Virtual Machines page.
- On the Virtual Machines page, select the VM that is stopped and that you want to force to boot (for example, VM-1).
-
In the bottom panel, click the Start button.
The VM begins to boot. It continues booting until the time-out limit is reached, possibly as long as 5 minutes. When the time-out limit is reached, the Force Boot button becomes active.
-
To force the VM to boot, click Force Boot.
A warning appears, asking you if you are positive that you know which node has the most up-to-date VM data. The warning also tells you to be aware that data loss is possible. In addition, a message tells you the node on which you can force the VM to boot.
Caution:
If you select the wrong node during Force Boot, data is damaged.
You must type the node (node0 or node1) as indicated in the message. The following message is an example:
| |
|
|
|
| |
|
Force Boot VM-1
|
|
| |
|
DO NOT PROCEED UNLESS YOU ARE POSITIVE YOU KNOW
WHICH NODE HAS YOUR MOST UP TO DATE VM DATA. BE
AWARE THAT DATA LOSS IS POSSIBLE. |
|
| |
|
|
|
| |
|
Only node1 can be force-booted. |
|
| |
|
If you would like to boot the VM on node1, type node1: |
|
| |
|
node1 |
|
| |
|
 [ OK button ][ Cancel button ] |
|
| |
|
|
|
-
Click OK to force the node (for example, node1) to boot. (Click Cancel to cancel the procedure.) As the force-boot process begins and continues, additional confirmation messages appear before the VM starts and the data is marked as valid to the system.
The VM begins to run. On the Virtual Machines page, the VM is listed with a warning because the node (for example, node0) is still offline.
Once the secondary node is brought back in to the system, all data synchronizes from the node running the VM. In this example, all data synchronizes from node1 to node0.
Product Support and Downloads