[Search for users] [Overall Top Noters] [List of all Conferences] [Download this site]

Conference ssdevo::hsj40_product

Title:	HSJ30/40 Product Conference

Moderator:	SSDEVO::EDMONDS

Created:	Tue Jul 13 1993
Last Modified:	Fri Jun 06 1997
Last Successful Update:	Fri Jun 06 1997
Number of topics:	1264
Total number of notes:	4958

1213.0. "HSJ40 throughput performance" by CSC32::L_SOBCZYNSKI () Wed Feb 19 1997 15:58

This is a general question at this point, additional information can be 
gotten when requested.


    ?Can cycling power on an HSJ40 affect throughput performance?
			   v2.7 firmware

The situation in brief is as follows; a number of nightly batch jobs 
are run that perform various record/file manipulation.  The jobs
running are consistent, however the files affected and record sizes
being manipulated, are relative to the type of transactions that have 
taken place during the day.  Therefore to say that there is  day to 
day consistency in processing then the answer is no.  We can safely 
say that over a period of time however the type of process, duration 
would even out, so some general observations, relative to elapsed time, 
can be made based on a working knowledge of the day to day operations.  

Ok, it has been noted that apparently the time to complete overall 
processing each night increases.  The time frame being measured is 
approximately seven calendar days.  If the HSJ40 is power cycled, not
necessarily in an orderly fashion, the cycle time for a job tends to 
decrease, back to what is considered as normal processing time, point 
zero.  Then gradually time increases to job completion will be realized, 
first night, possibly the second night the job will run great, hack to 
time frame zero.  Then the times start extending out, longer and longer
each day.  Have seen the length of time increase to over 45 minutes in 
a weeks time frame.

Any ideas appreciated!

Regards
Leonard

T.R	Title	User	Personal Name	Date	Lines
1213.1	Cache flush timer???	SSDEVO::RMCLEAN		`Thu Feb 20 1997 12:36`	1
	What is the value you have as the cache flush timer set to?
1213.2	using default of 10 seconds	CSC32::L_SOBCZYNSKI		`Thu Feb 20 1997 13:06`	9
	Hello Ronald, The default value, 10 seconds. Regards Leonard
1213.3	!!!! HELP !!!!	CSC32::L_SOBCZYNSKI		`Tue Feb 25 1997 09:32`	7
	Hopefully this is not a dead note, I do need assist on this. Thank you. Leonard
1213.4	Not a clue without being able to debug it...	SSDEVO::RMCLEAN		`Tue Feb 25 1997 12:06`	2
	Haven't got a clue but... I would suggest you wait a couple of weeks and try V3.1. I guarantee that the results will be different ;-.]
1213.5	Release Notes available yet??	KERNEL::LOANE	Comfortably numb!!	`Wed Feb 26 1997 02:03`	4
	>I would suggest you wait a couple of weeks and try V3.1. ...are the V3.1 release notes available yet so we can see what major features'n'functions are available??
1213.6	Ok, then	CSC32::L_SOBCZYNSKI		`Wed Feb 26 1997 09:00`	14
	Ronald, Thank you for reply in .4. As questioned in .5, if the released version of the release notes are not available as yet, can an advance copy be had? Cheers Leonard [EOB]
1213.7	Not yet	SSDEVO::RMCLEAN		`Wed Feb 26 1997 12:34`	2
	No release notes yet. It is not a released product yet. (read this it isn't written yet ;-.])
1213.8	What info can be gathered to troubleshoot?	CSC32::L_SOBCZYNSKI		`Thu Feb 27 1997 09:14`	25
	Hello Ronald, Hopefully my attempts to further the inquiry on the situation that has been encounterd at my customer's site is not being viewed as, its back again? As yet I have not mentioned the upgrading to version 3.1 as a possibility, I'll wait until its released and I have a copy in hand. In a discussion with the customer it was stated that when the situation was experienced originally it was the HSJ that was powered cycled and a decrease in job process time was observed. The latter experience involved a number of other hardware entities shall we say not interacting together properly that ultimately led to the HSJ being rebooted, so no factual information could be gathered from that incident. What information could be gathered from the HSJ should the situation be observed again to aid in getting to the root cause of the increased job cycle time? Regards Leonard