ADSM-L

dsmmigrate and dsmrecall help needed

1998-09-22 12:12:55
Subject: dsmmigrate and dsmrecall help needed
From: Chuck Mattern <Chuck_Mattern AT HOMEDEPOT DOT COM>
Date: Tue, 22 Sep 1998 12:12:55 -0400
Greetings,

I am working on some Perl scripts that call dsmmigrate and dsmrecall to
manage storage of some digitized images which reside on an AIX 4.3.1 client
and are migrated off to an MVS server dsmmigrate and dsmrecall both report
that they are 3.1.0, the MVS server version is 3.1.1.3.  I'm running into a
few bumps in the road and would appreciate feedback from anyone with
experience in using these programs.  I've got several questions here:

1. With both commands I read from a data source and when I have what I
believe to be a suitable length string I fork another process to run the
migrate or recall command as appropriate  passing the string as an argument
then go back to my source and continue to read until ready to fork again.
I've not found a reference for the maximum command line length, IBM states
that it can be at least 1024 chars, perhaps more but not certain.  Any one
know for sure?

2. When several dsm[recall|migrate] commands kick off nearly simultaneously
I will frequently get ANS1017E errors.  To deal with this I have inserted a
random sleep in each forked off process and also set up a loop that will
retry the command up to 10 times if the command exits with a 1, I've noted
that dsmmigrate seems to return a 2 if the file was too small to migrate, 4
if the file is already migrated and 1024 if it encounters a bunch of
errors, I say seems since I can't seem to get a listing of the exit codes
for either of these apps.  Can anyone assist with that?

3. Is there any reference as to how many concurrent recall sessions or
migrate processes a server should be able to handle?

4. When I run a large number of concurrent dsmmigrate commands I sometimes
find a few left in something akin to zombie state.  I can see the
dsmmigrate process in the output of ps -ef but the server shows no
processes running.  The dsmmigrate command gets no CPU time and does not
exit, it just sits there.  I have left these to see how long they will sit
there and have never seen them die until killed, they have idled there for
several days.  It would seem that some sort of a timer mechanism should
exit so that if the server does not respond in a reasonable amount of time
the process will attempt to do something useful; i.e. restart or exit with
a meaningful return value.  Any thoughts here?

Thanks in advance,
Chuck
<Prev in Thread] Current Thread [Next in Thread>