HPE Community read-only access December 15, 2018
This is a maintenance upgrade. You will be able to read articles and posts, but not post or reply.
Hours:
Dec 15, 4:00 am to 10:00 am UTC
Dec 14, 10:00 pm CST to Dec 15, 4:00 am CST
Dec 14, 8:00 pm PST to Dec 15, 2:00 am PST
Operating System - OpenVMS
cancel
Showing results for 
Search instead for 
Did you mean: 

SYSGEN - GBLSECTIONS parameter

 
Brian Reiter
Valued Contributor

SYSGEN - GBLSECTIONS parameter

Hi Folks,

 

Just run AUTOGEN to TESTFILES on one of our RX2800s (VMS 8-4, Update 7), some of the processes were struggling. The report indicates that the  calculated value of GBLSECTIONS is 78640. Looking at the output from SYSGEN the upper limit seems to be 65535.

 

(CATA2)SYSTEM>mc sysgen sho gblse
Parameter Name Current Default Min. Max. Unit Dynamic
-------------- ------- ------- ------- ------- ---- -------
GBLSECTIONS 65535 1024 80 65535 Sections

 

 

Have things changed and SYSGEN hasn't caught up? This is a test system so I've been able to spread the load a round the cluster but I am curious as to whether or not this would cause problems if the system were to be autogenned.

 

cheers

 

Brian Reiter

 

2 REPLIES
abrsvc
Respected Contributor

Re: SYSGEN - GBLSECTIONS parameter

The few times that I have seen similar calculations in the past resulted in the parameter being set to the maximum value.  If you have a system that can be used for a test, let Autogen complete.  Check the result.  If not set to the MAX, post the value that is set.  At the very least, there are work-arounds for this, but knowing what Autogen actually creates will allow for this to be fixed (if needed).

 

Dan

Hoff
Honored Contributor

Re: SYSGEN - GBLSECTIONS parameter

Some of the processes are "struggling"?  Could you elaborate on that?  Symptoms, messages, log entries, etc?

 

I've encountered configurations that blow out the GBLSECTIONS limit due to section leakage, and feedback can pick that up and run with it.   There are other ways that sections can leak, particularly on a test system, and also sometimes in cases where there's an inexplicable and often entirely intractable resistance to just rebooting the box to clear it out.

 

Put another way, how many sections do you expect to have in use under normal load?  Usually, configure that, plus ~25% or so.

 

Could also be a bug in AUTOGEN.  Wouldn't be the first the that's happened.

 

As for overhead, having 65536 sections configured requires 256 pages of physical memory for the overhead, plus however much memory that the sections themselves are using.

 

How much memory is configured in this rx2800 box?  Even more than Alpha, Itanium is quite fond of physical memory.