Configuring EMI Framework in SAS Grid Environment

[Madhan_cog1]

Hi All, We have installed and configured SAS Grid 9.4 M5 in a Linux Environment. we have installed and configured meta and web tier in local mount points where as we have installed Application server in a shared Mount point (IBM Spectrum scale).

Since we have installed Application server in a shared directory we have installed it only on the grid controller node and only configured the grid compute nodes for object spawner and SAS Environment manager agent. 

Now we are trying to configure EMI framework using the below link.

https://support.sas.com/rnd/emi/SASEnvMgr/EVSAF/SAS_Environment_Manager_Service_Architecture_Quickst...

when we are trying to configure EMI Framework it is configuring and initialising only for the Grid control server and it is not considering the grid compute nodes.

 Is this the right approach or is there any other process or steps to be followed to make it work even for the grid compute nodes.

Please suggest.

Thanks,

Madhan M.

[gwootton]

Yes, that is correct. The grid nodes use the same configuration directory as the grid control server, so they do not need to participate as well. The per-host ETL process is log collection. As these are stored in the configuration directory, the grid controller can pick up the logs for servers run on the nodes.

[Madhan_cog1]

Hi Greg,

Thanks for the response. 

In the document we have an appendix section 

Appendix C: Methods of collecting logs in a multiple machine
environment.

Attaching the link for pdf again.

"https://support.sas.com/rnd/emi/SASEnvMgr/EVSAF/SAS_Environment_Manager_Service_Architecture_Quickst...

Could you please let us know what is this section for. 

Thanks,

Madhan M.

[gwootton]

The EMI Framework scripts are:
Log Collection - Runs on each host (except grid nodes), gathering log files into a single location in the configuration directory for that host.
Log Centralize - Runs on the compute server (grid controller), this uses the SAS Deployment Agent to gather the logs from each configuration directory on the various hosts into a single directory in the compute tier's configuration directory.
Master ACM ETL - Runs on the compute server (grid controller) to update the data sets used by the ACM reports.
Master APM ETL - Runs on the compute server (grid controller) to update the data sets used by the APM reports.

The Appendix C you mentioned gives alternatives to the Log Centralize process if it cannot be used for some reason.