mload active for more than an hour however is not captured in log

Database

mload active for more than an hour however is not captured in log

Hi,

 

We use informatica to load data into Teradata using MLOAD or Fastload utilities , Everytime informatica job starts i can see multiple spawned sessions in viewpoint.Once the job completed I tried to map sessions in DBQLOgTbl with DBQLObjTbl  , I did not find the sessions which were active in Viewpoint for more than an hour. I could only see 1 or 2 Mload session in DBQLOgTbl. 

and  the sesion which got captured in DBQLOgTbl shows elapsed time of 1 or 2 seconds , however the Mload was active for more than 1 hour as per the viewpoint data. My question is ,What  those sessions were doing for moe than 1 hour , Is there any way to find out? 

 

1 REPLY
Senior Apprentice

Re: mload active for more than an hour however is not captured in log

Hi,

 

Trying to track Multiload/Fastload activity via DBQL is not going to be easy - especially if you're looking at the 'obj' table. That is typically only going to record the results of sql commands and the bulk of (for instance) ML commands are not sql. (and in this context 'sql' is the 'dbc/sql' partition - that you'll see in dbc.sessioninfov - as opposed to the 'mload' partition).

 

What you probably see in there are the commands that ML will use to create (and then drop) the work tables and the restart log and possibly not much else. These are standard SQL commands.

 

What you won't see are all the commands to load data from the client platform into the work table(s), then sort it and then load/merge the data from the work table(s) into the target table(s). It is these commands which will 'take an hour' - or however long the ML job takes.

 

If you're on TD15.0 or higher you could use the 'utilityinfo' dbql option which will capture a lot of info about how long individual stages in the utility last, how much data is transferred etc.

 

Does that help/explain what is going on?

 

Cheers,

Dave

 

Ward Analytics Ltd - information in motion
www: http://www.ward-analytics.com