[prev in list] [next in list] [prev in thread] [next in thread] 

List:       mumps-l
Subject:    Re: More explicit
From:       don groves <dtgroves () shentel ! net>
Date:       1999-09-16 11:06:46
[Download RAW message or body]


Open-m supports implicit directories so that when the original database is
getting full you can create a new data base and put some of the globals in
it and implicitly link them back to the original database, You can have
several implicit databases linked to the original database. Each implicit
database can be up to 16gb. Cache supports splitting a single global over
multiple databases.  do not recommend using anything less than open-m 6.x

The basic process is:

create new database and make active.

use ^NAMESP in the managers directory to setup a configuration file (use
default dataset). set configuration to replicate the global desired.

use ^SYSMGR to edit the msql startup to include the configuration.
stop/start open-m and verify what is written to the global in the original
database is also in the new database.

write a drill down routine that starts at the top of the global being
replicated and re-writes the data to its self.  ex: s
a=$o(^TMP(a)),b=$g(^TMP(a)),@a=b this will put existing data into the new
database.

once the drill down is complete use ^NAMESP to remove the replication
stop/start open-m.

verify what is written to the global in the original database does NOT
appear in the new database.

Kill the global from the original database, this will take some time
depending on the size of the global being killed (garbage collector and
global buffers)

stop/start open-m use ^NAMESP to set up the implicit reference from the
original database to the new database. stop/start open-m

the global should now be visible to the original database and the original
database now has free space.

Don

dtgroves@shentel.net

----- Original Message -----
From: Matt Conley <conleym@exempla.org>
To: <MUMPS-L@LISTSERV.UGA.EDU>
Sent: Wednesday, September 15, 1999 3:43 PM
Subject: Re: More explicit


> VLS is "Very Large Systems". It is an Epic Systems solution for IS[M] when
a volume group (MUMPS.DAT and it's subsequent MUMPS.EXT's) run out of room.
You can only have 7 MUMPS.EXT files per each parent MUMPS.DAT file....so you
VLS the INI's that contain/use the 'out of room' directory and spread it out
over several filesystems. The globals that get split are considered a VLS'd
global. :-)
>
> >>> <HulseyBruceB@exchange.uams.edu> 09/15/1999 1:18:24 PM >>>
> Hello,
>
> I have not heard of VLS globals before.  What implementation of M are you
> using?
>
> ---
> Bruce Hulsey
> Arkansas Cancer Research Center
> University of Arkansas for Medical Sciences
>
> > -----Original Message-----
> > From: Bronwen Hunt 50 2443 [SMTP:Bronwen.Hunt@kp.ORG]
> > Sent: Wednesday, September 15, 1999 11:43 AM
> > To:   MUMPS-L@LISTSERV.UGA.EDU
> > Subject:      More explicit
> >
> >    RE my last email question:
> >
> >    VLS globals look like one file to M but are actually spread over
> > several
> >    disks and directories. %G only gives what is in current directory, as
> >    far as I can tell. I'd like to be able to look at the entire global
> > with
> >    a %G type utility, but no luck finding one (or trying to write one).
> >
> >    Does anyone have any ideas?
> >
> >    Bronwen
>

[prev in list] [next in list] [prev in thread] [next in thread]