On Wed, 5 May 1999, Dick Nieuwenhuizen wrote:
> There is nothing wrong with the dsmsched.log file. The problem is the
> running proces dsmc sched. When I'm looking with ps -elf the core image is
> growing on several clients. After a restart of the scheduler on one client
> the proces took only 1612 units instaed of the 50456 before the restart. Why
> is it growing and what can I do to prevent it.

I have observed this on the Solaris client also.  When a backups starts,
the server's view of the client as of the last backup is downloaded to the
client, and stored in memory.  This structure is then used to see what
has changed, and what needs to be backed up.  IMHO, the memory allocated
for this structure is not freed when the backup completes, perhaps on
purpose, perhaps not.  I see this as either a bug, or poor design,
depending on whether or not the retention of the memory was done on
purpose, perhaps so that allocation for the next backup is not required?

I have some clients that download upwards of 200M for a backup, so this is
not an insignificant amount of memory.

IBM, is my take on what is happening accurate?  If so, could you please
change the clients so that the memory is freed up when it is no longer
needed?

Thanks,

Steve (unVMix Systems Programmer/Dude) Roder
(tkssteve AT ubvm.cc.buffalo DOT edu | tkssteve AT acsu.buffalo DOT edu | 
(716)645-3564 ,
   | http://ubvm.cc.buffalo.edu/~tkssteve)