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

Conference humane::scheduler

Title:	SCHEDULER
Notice:	Welcome to the Scheduler Conference on node HUMANEril
Moderator:	RUMOR::FALEK

Created:	Sat Mar 20 1993
Last Modified:	Tue Jun 03 1997
Last Successful Update:	Fri Jun 06 1997
Number of topics:	1240
Total number of notes:	5017

1230.0. "V2.1 ECO9 timing problems - any clues ?" by ATZIS3::PIEBER (chaos has many faces) Tue Mar 25 1997 03:13

    Hi folks,
    
    we received ECO 9 for Scheduler V2.1B in conjunction with ABS V2.1.
    This ECO is a prerequisite for ABS 2.1 due to the API for ABS, that 
    comes with this ECO.
    
       Now we see timing issues with this version, we did not see with
    prior versions (V2.1B-1). Some of those are:
    
    * job sits in 'scheduled' state, with start time already passed with
      no action. The Debug file tells, that the scheduler woke up and decided 
      not to do anything about this and falls asleep again.
    
    * some jobs must be sent a RUN plus a RELEASE command to get it on the
      road, although they are in scheduled state.
    
    
      Sometimes the SCHED CHECK command corrects this mis-behavior, but not
    always.
    
    
    -->  Is there a fix to be expected soon ?
    -->  Can we provide any informations to track down these issues ?
    -->  Are these known issues yet ?
    -->  Need an IPMT ?
    
    
    Ewald.

T.R Title User Personal
Name Date Lines

1230.1 Tell some more... HLFS00::ERIC_S Eric Sonneveld MCS - B.O. IS Holland Wed Mar 26 1997 01:48 13

T.R	Title	User	Personal Name	Date	Lines
1230.1	Tell some more...	HLFS00::ERIC_S	Eric Sonneveld MCS - B.O. IS Holland	`Wed Mar 26 1997 01:48`	13
	> --> Are these known issues yet ? I'm running V2.1B-9 since almost a year (? - just got the kits before the scheduler people wnet to CA) in a mixed cluster with around 700 active jobs and have not seen this behaviour. As you stated that in the debug logging nothing about time is mention it looks very strange. How about timestamps in the logfile ? how did you conclude that scheduler found something and decided to do nothing ? Are you on VAX or Alpha ? Is DTSS implemented ? Is it a cluster, is so is loadbalancing used ? Are these detached or batch jobs ? Eric

>    -->  Are these known issues yet ?
I'm running V2.1B-9 since almost a year (? - just got the kits before the
scheduler people wnet to CA) in a mixed cluster with around 700 active
jobs and have not seen this behaviour.
As you stated that in the debug logging nothing about time is mention it looks
very strange.

How about timestamps in the logfile ? how did you conclude that scheduler found
something and decided to do nothing ?
Are you on VAX or Alpha ? Is DTSS implemented ? Is it a cluster, is so is
loadbalancing used ? Are these detached or batch jobs ?

Eric