Date:         Wed, 29 Sep 2010 05:29:43 -0400
Reply-To:     Proc Me <procme@CONCEPT-DELIVERY.COM>
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From:         Proc Me <procme@CONCEPT-DELIVERY.COM>
Subject:      Re: Monitor a spooling file
Comments: To: Tom Abernathy <tom.abernathy@GMAIL.COM>

Tom,

Thanks, yes, I think so, the original post was in September week 1, and you
may be able to see it here:

http://www.listserv.uga.edu/cgi-bin/wa?A2=ind1009a&L=sas-l&F=&S=&P=4592

What I would like is a half-way house between parallel running (both kicked
off at the same time) and sequential running (second Proc SQL step starts
after first has finished). I want the execution of the second job to
coincide with the spooling/data transfer phase of the first job. Each phase
of the first job takes around two hours.

The problem I've got is that SAS doesn't provide a trigger* to tell me when
the database changes from executing the query to spooling the data, because
this occurs within the step boundary of the first Proc SQL statement.

To work around this I want to automate our current manual process of
watching the output SAS dataset. This is created right at the start of the
Proc SQL step with a nominal size (1kb); when the database spools data this
file starts to grow, and we know it is safe to kick off the second job.

I hope this makes sense!

Proc Me
* It may be that unbuffering the log as Data _Null_ suggests and parsing the
log may provide such a trigger.