Search <book_title>...

InfoScale™ Cluster Server 9.0 Bundled Agents Reference Guide - Linux

Last Published: 2025-04-14

Product(s): InfoScale & Storage Foundation (9.0)

Platform: Linux

GoogleDisk agent

The GoogleDisk agent lets you configure data disks for HA in GCP, so that the disks can also be failed over during an application failover between instances.

To store the data of the applications that you configure for HA, InfoScale supports the following storage types with GCP VM instance:

Zonal persistent disks - Standard and SSD
Regional persistent disks - Standard and SSD

The GoogleDisk agent brings the GCP data disks online, monitors their status, and takes them offline. The GoogleDisk agent uses GCP Python SDK to determine whether the data disks are attached to the virtual machines or not.

Prerequisites

Your Google service accounts must have the following permissions, at the least:
- compute.projects.get (at the project level, to fetch the project metadata)
- compute.instances.get (at the instance level, to fetch the instance metadata)
- compute.instances.attachDisk (at the instance level, to attach a disk to the instance)
- compute.instances.detachDisk (at the instance level, to detach a disk from the instance)
- compute.zoneOperations.get (to check the status of operations in the zone)
- compute.disks.get (to fetch the list of the current users of a disk)
- iam.serviceAccountUser (lets a principal attach a service account to a resource)
To access Google APIs, the google-api-python-client Python module must be present on the InfoScale cluster nodes.
You can install Python SDK for GCP on each of the cluster nodes by running the following command:
# /opt/VRTSpython/bin/pip install google-api-python-client
You must first create a disk before you configure a resource for it. Otherwise, the online operation fails.
Add dependencies in the service group as depicted in Figure: Sample service group dependency

Figure: Sample service group dependency

Agent functions

Open	Fetches the current GCP VM instance name and the zone and project to which the instance belongs, using its metadata.
Online	Attaches the GCP disk to the VM instance.
Offline	Detaches the GCP disk from the VM instance.
Monitor	Determines whether the GCP data disk is attached to the current VM instance or to any other VM instance within or outside the cluster. It also identifies the mode in which the disk is attached and accordingly returns a status value.
Clean	Detaches the GCP disk from the VM instance.

State definitions

ONLINE	Indicates that the GCP disk is attached to the current VM instance.
OFFLINE	Indicates one of the following: The GCP disk is not in use by any VM instance. The disk is detached from the current VM instance, but it is attached to another VM instance within the cluster in the read-write mode. The disk is attached to any other VM instance within or outside the cluster in the read-only mode
FAULTED	Indicates that the GCP disk was unexpectedly detached from the VM instance outside of VCS control.
UNKNOWN	Returns UNKNOWN if: The required Google Python SDKs are not installed on the current VM instance. It fails to fetch the required attributes values. The value of the ReadOnly attribute does not match the mode in which the GCP disk is attached to the VM instance. Such a situation may cause issues for the applications that use the disk. Roles are not configured correctly or they are not correctly associated with the VM instance so as to fetch the disk information. An invalid DiskName is provided. The specified DiskName of a zonal disk is not found in the zone in which the VM instance exists. The zonal disk exists in some other zone. The zonal disk is attached to a VM instance outside the cluster in the read-write mode. The regional disk is attached to a VM instance outside the cluster in either the read-write or the read-only mode.

Attributes

Table: Required attributes

Attribute	Description
DiskName	Name of the disk. Type and dimension: string-scalar
ReadOnly	Mode in which to attach the disk: 0: READ_WRITE 1: READ_ONLY Default value: 0 (READ_WRITE) Type and dimension: boolean-scalar
RegionalDisk	Type of the disk: 0: zonal disk Zonal disks can be failed over to cluster nodes within the same zone. 1: regional disk Regional disks can be failed over to cluster nodes that may exist across different zones but within a region. Default value: 0 (zonal disk) Type and dimension: boolean-scalar

Attribute

Description

DiskName

Name of the disk.

Type and dimension: string-scalar

ReadOnly

Mode in which to attach the disk:

0: READ_WRITE
1: READ_ONLY

Default value: 0 (READ_WRITE)

Type and dimension: boolean-scalar

RegionalDisk

Type of the disk:

0: zonal disk
Zonal disks can be failed over to cluster nodes within the same zone.
1: regional disk
Regional disks can be failed over to cluster nodes that may exist across different zones but within a region.

Default value: 0 (zonal disk)

Type and dimension: boolean-scalar

Table: Optional attribute

Attribute	Description
ProjectID	Project to which the VM instance belongs. Type and dimension: string-scalar
Zone	Zone to which the instance belongs. Type and dimension: string-scalar
VMName	Name of the GCP VM instance on which the agent runs. Type and dimension: string-scalar

Attribute

Description

ProjectID

Project to which the VM instance belongs.

Type and dimension: string-scalar

Zone

Zone to which the instance belongs.

Type and dimension: string-scalar

VMName

Name of the GCP VM instance on which the agent runs.

Type and dimension: string-scalar

Resource type definition

type GoogleDisk (
	   static int FaultOnMonitorTimeouts = 0
   	static str ArgList[] = { DiskName, ReadOnly, 
      RegionalDisk, ProjectID, Zone, VMName, 
      tempProjectName, tempVMName, tempZone }
   	boolean ReadOnly = 0
   	boolean RegionalDisk = 0
   	str DiskName
   	str ProjectID
   	str Zone
   	str VMName
   	temp str tempProjectName
   	temp str tempZone
   	temp str tempVMName
)

Sample configuration

GoogleDisk VVRSG-GoogleDisk (
   	DiskName = sql-disk
)

Debug log levels

The GoogleDisk agent uses the DBG_1 and the DBG_2 debug log levels.