ERROR: Engine allows only sequential access while running Proc NLMIXED in HADOOP

[righcoastmike]

Hi All, 

I am attempting to run PROC NLMIXED in a HADOOP environment, and it keeps throwing up the error: 

ERROR: FILE SARV.FUD.DATA is sequential. This task requires reading observations in a random order, but the engine allows only sequential access.

I don't think it's a coding issue because my program worked fine when I was trying it on a local machine. The issue only seems to have arrived when I shifted to the HADOOP server. 

Any thoughts one why things might have changed and what a potential solution might be? 

Thanks so much. 

Mike

[righcoastmike · Accepted solution]

HI All 

Managed o figure this one out with a pretty simple solution, sorting the data before running proc nlmixed seems to have resolved the issue. 

Hope that helps folks moving forward. 

Mike 

[Rick_SAS]

Are you using PROC SORT and a BY statement? If so, read p. 5-6 of the paper by Mike Rhoads (2014)