[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: LDM 6.2.1 possible memory issue?

This archive contains answers to questions sent to Unidata support through mid-2025. Note that the archive is no longer being updated. We provide the archive for reference; many of the answers presented here remain technically correct, even if somewhat outdated. For the most up-to-date information on the use of NSF Unidata software and data services, please consult the Software Documentation first.

  • Subject: Re: LDM 6.2.1 possible memory issue?
  • Date: Tue, 08 Mar 2005 13:11:19 -0700
Steven,

>Date: Tue, 08 Mar 2005 13:55:07 -0600
>From: "Steven Danz" <address@hidden>
>Organization: Aviation Weather Center
>To: Steve Emmerson <address@hidden>
>Subject: Re: LDM 6.2.1 possible memory issue?

The above message contained the following:

> Sure thing, but the VSS doesn't change much.  However, the RSS grows to 
> near the VSS size over time.

The LDM doesn't have any control over that.  The size of the RSS is
completely up to the operating-system.

You might try Google-ing for RHEL3, virtual size, RSS, etc.

> Would it help to run the script on a 6.0.14 system as well as the 6.2.1 
> system?

The purpose of the script is to gather memory-usage statistics so that a
visual display of memory usage will indicate whether or not a problem
exists.  If the LDM 6.0.14 doesn't have a memory problem, then I'm not
concerned about its memory-usage statistics.  Presumably, you've been
running LDM 6.0.14 for some time now and haven't noticed anything
amiss (no one else has either).

> Well, that's what I'm sorta getting at (though probably not very
> well).  The OS is the exact same in both cases, so the memory behavior
> (VSS, RSS, or whatever) should be the same unless there is something
> significantly different between 6.0.14 and 6.2.1, true?  With 6.0.14
> the VSS is ~27Meg, with 6.2.1, the VSS is ~790Meg.

I don't know why the virtual size of an LDM 6.2.1 process would be so
different from a similar LDM 6.0.14 process.  It would seem that one
system is including the size of the memory-mapped product-queue and the
other isn't.  Why this should be, I don't know (it does seem odd).  The
size of the product-queue can't change, however, so the absolute value
isn't meaningful -- only changes are.

As long as the virtual size doesn't grow, I'm not concerned about
the reporting difference.

Regards,
Steve Emmerson