Architecting, installing and maintaining your SAS environment

TKGrid Process

Reply
Regular Contributor
Posts: 212

TKGrid Process

Hi everyone.

 

There is a few days that we are having troubles with the LASR Servers of SAS Visual Analytics.

The problem is:

 

Lasr Servers Stopped:

 

LASR Analytic Server - MyServer              

Public LASR Analytic Server - MyServer

LASR Server DBM

 

Causes:

 

We could not find them yet. But there are somethings we've found:

 

  1. There are some TKGrid Processes Stopped during days in all nodes into the VA Server
  2. These processes take more than 100gb
  3. LASR Server DBM is in 74% reached
  4. We've made the Stop/Start, but the scripts did not killed these processes
  5. We run some table to load from the Guide directly to LASR server (Without the HDFS process

I'd like to know what are the possible causes when the LASR Servers Stops.

And why these TKGrids processes are still running during days, even when we restart the sas.servers ?

 

Thanks in Advance

 

Occasional Learner
Posts: 1

Re: TKGrid Process

Hello,

 

If your LASR Server(s) are stopped check the SAS Visual Analytics Administrator page where you can see their status. Right click then select LASR Action log to see if it has a recent log. You can also use this utilitity to monitor your LASR servers for things like table memory usage - Usage Note 58922: Using the LASRInfo.jar utility to see information about your SAS® LASR™ Analytic S... . If you have password-less ssh setup you can launch the TKGrid/bin/gridmon.sh script with X Windows running to see their memory usage, jobs, and activity. For more details see Paper 3275-2015 Great Performances: SAS® Visual Analytics Performance Monitoring and Enhancement.

 

 

  1. There are some TKGrid Processes Stopped during days in all nodes into the VA Server
    1. Are they stopped, or are these for LASR servers that have tables loaded in its memory? 
  2. These processes take more than 100gb
    1. LASR is meant to hold large amounts of data in memory, it is normal for it to consume a significant amount of memory, that is its purpose. It is also normal for the data to expand from the size on disk depending on the data itself.
  3. LASR Server DBM is in 74% reached
    1. If you mean it has reached 74% memory consumption that might not be good. There is a hard limit set on the LASR servers properties to stop loading when when LASR reaches 80% of memory used.
  4. We've made the Stop/Start, but the scripts did not killed these processes
    1. sas.servers scripts do not start/stop LASR, you manually start/stop them form SAS Visual Analytics Administrator web ui, or autoload can be configured to start them as well.
  5. We run some table to load from the Guide directly to LASR server (Without the HDFS processI'd like to know what are the possible causes when the LASR Servers Stops.And why these TKGrids processes are still running during days, even when we restart the sas.servers ?

 

The primary cause of LASR servers stopping, or at least not loading data is either you are out of memory on one of your node (80% of RAM usage), you reached a set limit of some sort such as ulimit number of files, number of processes, of ssh sessions. The second likely reason is communication failures between TKGrid nodes, SAS, and Hadoop cluster if you have it. The processes for a LASR server will continue to run as long as their is data loaded into LASR memory, which is why you see the memory continue to be used. The sas.servers does not start/stop LASR servers, you typically do this from SAS Visual Analytics Administrator, or autoload can possible start them automatically if configured.

 

 

Ask a Question
Discussion stats
  • 1 reply
  • 365 views
  • 4 likes
  • 2 in conversation