How Metadata Quorum works in a Clustered grid environment also the configuration details info needed

[DevaCh]

Hi All,

This would be a great help if any one could help me with "How Metadata Quorum works in a Clustered grid environment"  also in which configuration files the server details will be configured.

Thanks,

Deva.

[AnandVyas]

Hi @DevaCh 

IMO, clustering metadata is same irrespective of the environment of being grid/non-grid. It depends more on the topology of the environment deployed at your site. The sync happens using a shared file system where the metadata configuration is stored as well as backups. As long as this is available, you can have your SASHOME/SASCONFIG locally on each server or even on a shared file system depending on the available hardware.

If you have SASCONFIG on local FS then you will have to make changes on each server locally. In-case of a shared file system single change would do. Changes related to metadata (using SMC) is only required once and it will be synced automatically.

You can check this video for more information: https://www.youtube.com/watch?v=7O8clfMn88s

[gwootton]

As @AnandVyas says, when the environment is grid (compute tier) has no bearing on how Metadata clustering works.

While a grid environment typically uses a shared configuration path for the compute tier, metadata and middle tier clustering do not, so each metadata host would have it's own configuration directory. The configuration directory for the metadata servers contain their own copy of metadata. Quorum exists when a majority of the metadata servers have the same metadata contents. In the event of an even number of metadata servers, exactly half would only be considered in quorum if one of those hosts is the initially configured metadata server. This is described in more detail here:

How Metadata Server Clustering Works
https://go.documentation.sas.com/doc/en/bicdc/9.4/bisag/n1w2q4quib18udn1h8oj0wxfmgx6.htm

[DevaCh]

Hi Anand,
Thank you for your reply. I saw this URL (https://go.documentation.sas.com/doc/en/bicdc/9.4/bisag/n1w2q4quib18udn1h8oj0wxfmgx6.htm), Could you please help me with the Metadata configuration files of how Clustered and Non Clustered environment and is differentiated that would be very much helpful.

Thanks,
Deva.

[gwootton]

Are you asking how to use configuration files to tell if the environment you are using is a clustered or non-clustered environment? For the most part this information exists in Metadata rather than configuration files, so you could log in to SAS Management Console and see if there is more than one Metadata Server defined. As part of the process the metadataConfig.xml file is supposed to be updated on the compute tier (Lev1/ObjectSpawner/metadatConfig.xml) to list all the cluster hosts, but because things still work when this step is omitted I would not consider it a reliable way to tell if you are working with a clustered metadata server.

The documentation on configuring a cluster can be found in the link below.

Configuring a Metadata Server Cluster
https://go.documentation.sas.com/doc/en/bicdc/9.4/bisag/n1odroiqzm0fkhn16sakutai9qr5.htm