It wouldn't be too difficult to find customers who would like to configure dataservices in a much more user friendly and simpler way. Come 3.2, the latest offering from Sun for Solaris Cluster, we have what is called in software jargon - Dataservices Configuration Wizards. Personally I really like the term Wizard, which would relate to something magical - And that is really what these wizards can do for configuring dataservices on Solaris Cluster.
So what exactly is the advantage? Well, let us assume an NFS Data Service needs to be configured on a given cluster on the Solaris OS. The typical sequence of steps would be,
- Create a Failover resource group.
- Create a logical hostname resource inside the resource group.
- Put all the filesystems that have directories to be shared, under one or more HAStoragePlus resource in the same resource group. Creating an HAStoragePlus resource would also require an updation of the /etc/vfstab file on Solaris.
- Create a <PathPrefix>/SUNW.nfs/ directory. Care should be taken that this directory itself should be HA.
- Create a dfstab.<NFS Resource Name> file under this directory.
- Dump all the share commands corresponding to whatever directories to be shared in this file
- Create an NFS resource with dependencies on above logical host and HAStoragePlus resource.
- Bring the resource group online.
- Care should be taken to provide a unique name to the resource as a given cluster configuration cannot have duplicate resource names.
- All this and much more if we encounter any error message while configuration AND finally, the need to remember various commands involved, and their sub-options.
Now consider configuring the dataservice through the NFS Data Service wizard. The steps NOW would be,
- Invoke the wizard
- Follow the sequence of steps giving your inputs (ex : nodelist, pathprefix, mountpoints, hostname) OR just accept the default options.
- Finally review the configuration to be created, select create to get the NFS datservice up and running !!
So when you carry out the above three simple steps, in background the wizard would
- automatically discover available options for you, say available logical hostnames, HA mountpoints, share options...
- generate unique resource names.
- create/update required configuration files.
- validate input given at each step, thereby reducing any fault in configuration.
- automatically generate and execute Solaris Cluster commands.
- even rollback all the changes thus done, if the newly created resources fail to come up.
- and, finally displays list of all commands it executed, during the above steps (for those ever-curious types who would love to know how the wizard does all that it does), success/failure status, error messages if any, including rollback commands in such a case.
Each configuration wizard has been designed to efficiently and effectively handle the configuration of the corresponding Data Service. It covers most of the use cases related to different ways through which a dataservice can be configured.
Using a terminal access on any of the cluster nodes the wizards can be invoked using the command,
and then option - 3 (Data Services) is to be selected to get a menu of all those Data Services that can be configured through these wizards. This list has been carefully selected, taking into account customers’ feedback through market surveys. These are easily the most used Dataservices on Solaris Cluster.
If a browser is all that you have at present to access the cluster in question, these wizards could be invoked from the Task Page of Solaris Cluster Manager. The url -
is used to gain access to the browser user interface to manage the cluster (Solaris Cluster Manager).
Let us select NFS from the displayed list of dataservices.
The first screen that comes up is a little introduction and a list of prerequisites which are to be met to configure the dataservice. Verify prerequisites and press ENTER (Click Next - in case of BUI) to continue.
The next screen would show a list of available nodes and zones present on the cluster. Select the desired nodelist or just press ENTER to accept the default options (all available nodes/zones). A comprehensive set of validations would be performed by the wizard before proceeding to the next step. Some of them are – if input list contain zones then validating whether the dataservice support zones, verifying Dataservice package installation on the given nodes, etc.
The wizard steps are quite straight forward and one can easily continue further till the configuration results are shown through the Results panel. Validations performed at each step would ensure the correctness of the inputs given. The Result panel is of much importance in a way that it shows all the commands executed, and as I have already mentioned, it would show errors and rollback commands if any.
A very minimal knowledge of Solaris Cluster is needed to configure a dataservice through these wizards. Again, as I have already mentioned, the wizards would also update various configuration files needed for the dataservice to get configured. Wizard algorithms would always try to discover data for user inputs before it asks the user for the same, thereby reducing user interaction to a bare minimum.
Another important feature is the comprehensive help provided at each step of these wizards. For the command line based version just press '?' and ENTER to get the related help. For Browser user interface - click help tab on the left pane of the wizard screens. Our technical writers have spent good time framing these help sections to cover most of the aspects which a user might need at that time.
There is much more to these wizards and their magical ways than what I can discuss here, but yes, using them would definitely make an administrator's job much simpler. We would be coming out with more of such wizards in future releases supporting more and more of dataservices, and till then, more HAppier times would follow.
Member Technical Staff,
Solaris Cluster Engineering