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

Conference vaxaxp::vmsnotes

Title:VAX and Alpha VMS
Notice:This is a new VMSnotes, please read note 2.1
Moderator:VAXAXP::BERNARDO
Created:Wed Jan 22 1997
Last Modified:Fri Jun 06 1997
Last Successful Update:Fri Jun 06 1997
Number of topics:703
Total number of notes:3722

291.0. "Generic batch queue logic" by COMICS::MILLSS ("Jump! Jump now!" ...Kosh) Thu Mar 06 1997 12:19

Hello.

I know that generic queues don't do any prioritising when submitting their jobs
to execution queues apart from "first available in the list", but I have a
customer who is seeing jobs going to the second queue in the list even though
the first one is empty and idle.

Can anyone shed any light on why this might happen ?

Many thanks for any help or pointers.

Simon R. Mills
OpenVMS Specialist, UK CSC
T.RTitleUserPersonal
Name
DateLines
291.1AUSS::GARSONDECcharity Program OfficeThu Mar 06 1997 20:048
    re .0
    
    Selection can depend on form and characteristics (and who knows what
    else).
    
    At the very least you need SHOW QUEUE/FULL from all the relevant
    queues. Even better would be a log file showing exactly what the queues
    look like and what the user is doing and what the results are.
291.2COMICS::MILLSS"Jump! Jump now!" ...KoshMon Mar 10 1997 05:0542
re .1

>    Selection can depend on form and characteristics (and who knows what
>    else).

They're batch queues, so form and characteristics not relevant in this case.

>    At the very least you need SHOW QUEUE/FULL from all the relevant
>    queues. Even better would be a log file showing exactly what the queues
>    look like and what the user is doing and what the results are.

Yep. Good point. I'll get the hang of this Noting lark, yet !

> $ sh que ransysq/ful
>
> Generic batch queue RANSYSQ
> /GENERIC=(RANSYSQ_MIL2,RANSYSQ_MIL1) /OWNER=[SYSTEM,SYSTEM_MGR]
> /PROTECTION=(S:M,O:D,G:R,W:S)
>
> $ sh que ransysq_mil1/ful
> Batch queue RANSYSQ_MIL1, idle, on MIL1::
>   /AUTOSTART_ON=(MIL1::) /BASE_PRIORITY=4 /JOB_LIMIT=2
> /OWNER=[SYSTEM,SYSTEM_MGR] /PROTECTION=(S:M,O:D,G:R,W:S) /WSEXTENT=15000
>
> $ sh que ransysq_mil2/ful
> Batch queue RANSYSQ_MIL2, idle, on MIL2::
>   /AUTOSTART_ON=(MIL2::) /BASE_PRIORITY=4 /JOB_LIMIT=2
> /OWNER=[SYSTEM,SYSTEM_MGR] /PROTECTION=(S:M,O:D,G:R,W:S) /WSEXTENT=15000

I don't have a log of what happened to the customer, but it was simple. He
submitted a job to RANSYSQ whilst both queues it pointed to were empty. The job
went to RANSYSQ_MIL1 even though RANSYSQ_MIL2 is listed first and was empty and
idle.

It happened on (VAX, I think) VMS V6.2. I've tried reproducing it myself but it
all works as I would expect.

Any ideas ?

Thanks and regards,

Simon.
291.3AUSS::GARSONDECcharity Program OfficeMon Mar 10 1997 20:3722
re .2
    
>They're batch queues, so form and characteristics not relevant in this case.
    
    Characteristics are applicable to batch queues but if the queues don't
    show characteristics set then they are not relevant to this particular
    topic.

    Actually what I thought the customer was complaining was that jobs
    didn't go to an idle execution queue but instead remained pending on
    the generic queue in which case SHOW QUEUE/FULL might give clues as to why.
    
    If the situation is simply that they expect that there is ordering
    within the list of execution queues then perhaps they need to point to
    some doco that covers this.
    
    I have no idea what controls allocation to execution queues all other
    things being equal.
    
    If the customer can reproduce the behaviour then a log would help to
    make sure we have a common understanding of the scenario and would be
    highly desirable if they want to escalate formally.
291.4COMICS::MILLSS"Jump! Jump now!" ...KoshTue Mar 11 1997 09:0711
re .3

Thanks for your input. I did ask the customer the other day to supply me with
some hard evidence but his only response was "I have the accounting files". I
have asked him again to supply me with evidence.

As an aside, he also claims not to have installed any patches.

Thanks and regards,

Simon.