This section explains the range of support of PRIMECLUSTER in an FJcloud-O environment.
Number of cluster nodes: 1 to 2 nodes
Operation mode of the cluster system: 1:1 Standby operation, Mutual standby, Single-node cluster, Scalable operation (only when using Symfoware Server (Native) Hot Standby feature)
Network configurations
The virtual servers and the management client in the cluster system must belong to the same availability zone and subnet.
For the cluster interconnect, its network must be independent from the network used with the administrative LAN, the public LAN, and the mirroring among the servers of GDS. This setting is not necessary in a single-node cluster.
The virtual servers in the cluster system must communicate with the API endpoints. This setting is not necessary in a single-node cluster.
Security groups
One security group must be set among the virtual servers in the cluster system for a security reason.
Another security group must be set between the virtual servers and the management client in the cluster system.
The security group for the cluster interconnect must be set so that the node cannot communicate with any node outside of the cluster system. This setting is not necessary in a single-node cluster.
Error of OS on the virtual server and the cluster interconnect
The cyclic monitoring of the cluster interconnect (LAN) detects a hang-up of OS and the service is switched to the standby system.
Error of the shared disk and the disk access path
By combining the volume management function (GDS), a failure of the disk access and the disk access path can be detected (monitored by the Gds resource), and the service is switched to the standby system when the disk access is disabled or a failure of the whole system of the disk access path occurs.
Error of the cluster application
When a resource error of the cluster application occurs, the service is switched to the standby system.
Note
PRIMECLUSTER cannot be used in the virtual server built in a separate availability zone.
Set each virtual server so as to start it on each individual physical host.
To take over the data from one virtual server to the other virtual server, setting the mirroring among the servers of GDS is required.
To take over the IP address of the virtual server, setting the virtual NIC mode for Global Link Services (hereinafter GLS) is required.
The snapshot to the virtual server can be acquired only when OS is stopped.
The following functions for PRIMECLUSTER are not available:
Global File Services (hereinafter GFS)
GDS Snapshot
Root class and local class of GDS
Single volume of GDS, and disk groups of GDS other than the netmirror type (mirror, stripe, concatenation, and switch)
Cluster application using the takeover network resource
In addition to the above functions, the following functions for PRIMECLUSTER are not available in a single-node cluster.
GLS
GDS
The auto-scaling function for FJcloud-O is not available.
Can be used in conjunction with automatic failover function for FJcloud-O.
If an error occurs on the physical host on which the operational virtual server is running, the operation does not switch automatically. Switching requires manual intervention. If an error occurs on a physical host, you must re-create the virtual server on the physical host on which the error occurred, regardless of whether it is on the operational or standby system.
If an error occurs on the physical host on which the operational virtual server is running, the operation is switched after the automatic failover operation (about 30 minutes to 60 minutes).
For details on the automatic failover function, refer to the official FUJITSU Hybrid IT Service FJcloud documentation.
Duplicate virtual server names cannot be used in the project on FJcloud-O.
The console cannot be used in the virtual server in an FJcloud-O environment. Do not set the single user mode.
