Skip Navigational Links
LISTSERV email list manager
LISTSERV - COMMUNITY.EMAILOGY.COM
LISTSERV Menu
Log In
Log In
LISTSERV 17.5 Help - LSTSRV-L Archives
LISTSERV Archives
LISTSERV Archives
Search Archives
Search Archives
Register
Register
Log In
Log In

LSTSRV-L Archives

LISTSERV Site Administrators' Forum

LSTSRV-L

Menu
LISTSERV Archives LISTSERV Archives
LSTSRV-L Home LSTSRV-L Home

Log In Log In
Register Register

Subscribe or Unsubscribe Subscribe or Unsubscribe

Search Archives Search Archives
Options: Use Forum View

Use Monospaced Font
Show Text Part by Default
Show All Mail Headers

Message: [<< First] [< Prev] [Next >] [Last >>]
Topic: [<< First] [< Prev] [Next >] [Last >>]
Author: [<< First] [< Prev] [Next >] [Last >>]

Print Reply
Subject:
Re: SET RELPAGE OFF
From:
Brent Stilley <[log in to unmask]>
Reply To:
Forum on LISTSERV release 1.7
Date:
Tue, 26 Jan 1993 16:39:45 CST
Content-Type:
text/plain
Parts/Attachments:
text/plain (23 lines)
On Tue, 26 Jan 1993 18:29:32 +0100 Eric Thomas said:
>I have just noticed that many LISTSERV functions will perform much better
>if you add SET RELPAGE OFF to LISTSERV's profile. For instance, a refresh
>of the LISTS database takes 24.8 seconds of VCPU on SEARN with the normal
>RELPAGE ON setting and 19.8 (20% off) with RELPAGE OFF. The only drawback
>is that, if LISTSERV needs a lot of storage to perform a certain function
>and then  releases it,  it will not  notify CP that  the contents  of the
>pages are no longer important. This forces  CP to keep page slots on DASD
>for the  corresponding virtual memory  locations, and it may  take longer
>for the real memory frames to be reallocated to another user since CP has
>to assume  they are still needed  until they have been  unreferenced long
>enough. The paging DASD allocation issue is negligible on anything bigger
>than a  9370 with 5  users whereas the  recycling time problem  will only
>impact small to medium page-bound systems  (on a large system LISTSERV is
>just a  user among a thousand  and it doesn't make  any difference). This
>command will be included in the startup code in the next release.
 
This is a know problem, and has been fixed in VM/ESA R2 with the pending
page release enhancements.  So you'll probably want to make the command
conditional on the release of CP & CMS.
 
Brent Stilley,  Oklahoma State University, 113 Math Sciences, Stillwater, 74078

ATOM RSS1 RSS2

COMMUNITY.EMAILOGY.COM CataList Email List Search Powered by LISTSERV