On 1999/12/06, Eddy Thilleman wrote to Peter French on message number 330;
  Hello Eddy
ET> 03 Dec 99 17:27, Peter French wrote to Eddy Thilleman:
ET> PF> 3 Lazy write worker(s) are enabled.
ET>
ET> So I assume you don't know why 3 lazy write workers are (or
ET> should be) faster than 2 or 1 lazy write workers?

  No, I assumed that Denis selected 3 because of my memory size.  At the time,
he was helping me to trace the burst of disk activity - at a high priority -
that seemed to occur every couple of minutes.  This burst was really bugging
me
as it seemed that some people have this "effect" while others do not complain
about it.  I realise that it (in all probability) was the system flushing
cache
to disk - however the priority was so high that it would effect use of
something like EPM.  If I was typing - for instance - the burst would mean
that
the characters appearing on the screen would STOP until the burst stopped and
the system caught up to me.  This just seemed wrong on a OS that was designed
to do many things at the same time.  (we did confirm it was the USER/SYSTEM
INI
update)

  My understanding is, that each worker is a thread which means that the write
to disk is achieved by a worker, and a second one is activated should parallel
"stuff" in memory is required to be written at the same time!  Unless you have
a very memory intensive application it is unlikely that the volume to be
written out of cache would require more than the 3.

  Our co-operation was not conclusive - the problem was helped by cleaning up
my INI files, and I assume it is something to do with my system, my setup,
whatever.  I have found a very good work-around though - one I can recommend
to
anyone seeing the disk burst on their system.

REM -------------RAM INI Backup----------------------------
ifs=d:\util\ramfs64\ramfs64.ifs
REM SET USER_INI=C:\OS2\OS2.INI
REM SET SYSTEM_INI=C:\OS2\OS2SYS.INI
set user_ini=z:\inis\os2.ini
set system_ini=z:\inis\os2sys.ini
run=d:\util\ramfs64\ramdisk.exe z: ramdisk
run=D:\util\updini12\updini.exe c:\os2\ 10
REM ----------------RAM INI--------------------------------

  This change to the CONFIG.SYS sets up a RAM disk using RAMFS64 and changes
the systems standard INI positioning to the new RAM location.  Then it runs
the
UPDINI.EXE by Peter Engels.  This application copies the INI's - if they have
changed, to the normal location in idle and low priority time.  The "burst"
disappears, the system goes quiet, the operation does not effect any user
operations, the same caveats apply to operational data held in memory - should
the system abend abnormally.  OS/2 works (for me) now like I assume it should
and for over a year of use I have not had any (known) ill effects.


___
 X KWQ/2 1.2i X There is no right way to do a wrong thing

--- Maximus/2 2.02