Archives of the TeradataForum

Message Posted: Sat, 04 Apr 2015 @ 09:18:47 GMT

	<Prev	Next>		<<First	<Prev	Next>	Last>>

HI Tomas, Steven,

Thanks a lot for your direction. I could understand the reason for slow response and it's interesting now. :-)

There is nothing in IO variation, But, I see a big difference in (Max in Use Max) of Resusage data for that particular hour and it's going into flow control at 18:10.

What exactly these parameters indicate? Max(In use Max) and Avg(In Use Max)

AWT Info:

     Hr   Min  Max In Use Max  Avg In Use Max  Sum of Flow cntrl Vproc Flow cntrl
     18   0    65                              39              0                0
     18   10   93                              64              2                2
     18   20   93                              60              0                0
     18   30   63                              41              0                0
     18   40   84                              54              0                0
     18   50   88                              59              0                0

Based on this info, I have looked at my DBQL again.

Noticed that, one of the user's sessions are going into DELAYED STATE as we have a throttle limit (Med -2 and Long -2) . So, 3 were in ACTIVE and 3 were in DELAYED STATE for 18:00 till 18:20.

There were 3 sessions started by this user at 18:04, 18:05, and they took more than 1hr to complete, and whatever the sessions came newly, they went into DELAYED.

The maximum Delayed time was 940 second (15mins) for one of the session during 18:00 to 19:00.

     Username     Session id  Start time              First step time         First resp time         Delay       Elapse      Total
     ETL_XYZ_USR  6.319.604   26-03-2015 18:03:10,20  26-03-2015 18:03:10,20  26-03-2015 19:21:07,19  0,00    1:17:56.99  1:17:56.99
     ETL_XYZ_USR  6.319.653   26-03-2015 18:04:06,84  26-03-2015 18:04:06,85  26-03-2015 19:19:39,30  0,00    1:15:32.45  1:15:32.46
     ETL_XYZ_USR  6.319.653   26-03-2015 18:05:52,17  26-03-2015 18:19:12,76  26-03-2015 19:15:07,34  779,42  0:55:54.58  1:09:15.17

This might be the reason that my Canary response at 18:15 was showing 47K ms (50 sec). But, Our System is quite huge 6750 with 216 amps, surprized to see this behaviour.

Few Questions:

1. The total active sessions during this time was only around 50. (just to add, the cpu consumption was also normal). Why does it still go to Flow control and slow response time?

2. System will hold AWT's for the sessions which are in Delayed Queue?

Also, I cannot analyze my CANARY QUERIES from DBQL as I have disabled Query logging for VIEWPOINT user as it populates huge data.

3. What is the recommendation for DBQL logging for VIEWPOINT user?

Thank you & Happy Weekend!

Sravan

	<Prev	Next>		<<First	<Prev	Next>	Last>>