TSM_User,

I wasn't picking on your use of processed, just highlighting it to
distinguish it from the act of reading the file.

The processing has to be confusing to those who are used to other methods.
Some options like Domain can be specified multiple times and the arguments
are additive.  Most options can be specified multiple times and dsmc only
uses the last argument, as in the ones we're discussing.  Then there is the
continuous confusion over the includes and excludes of files and
directories.

 I'd someday like to understand why the original developers chose to process
the include/excludes of files from the bottom up.  But you have to pick
something.

Bill Smoldt
STORServer, Inc.

-----Original Message-----
From: ADSM: Dist Stor Manager [mailto:[EMAIL PROTECTED] On Behalf Of
TSM_User
Sent: Wednesday, September 01, 2004 9:31 PM
To: [EMAIL PROTECTED]
Subject: Re: Include/Exclude Question

Backup, wait a minute and read what I said again.

I did not say that it is read from bottom up!  As a matter of fact I said I
didn't know how it was read!  I said it was processed from bottom up.  If
you what to pick on my use of processed OK.  Well technically it may be that
each line is read one at a time from top to bottom.  Then repeated values
later in the file are layed over top of values that were previousely read
in.  I was only trying to point out that when a human looks at the file they
need to be sure that if there any repeated values the ones on the bottom
will win.  Of course you shouldn't have any option listed more than once in
a file but if you do this was my observation.

Yes, you are correct there are other factors at work with includes and
excludes like exclude.dir.

Further I pointed out what is done with the set opt command on the TSM
server.


Bill Smoldt <[EMAIL PROTECTED]> wrote:
TSM_User,

I don't understand how your observation supports your conclusion that the
.opt file is read in from the bottom. I believe it simply means that TSM
uses the last instance of each option that has a unique value in the .opt
file.

The include options are processed in a very specific manner. Exclude.dir
statements are evaluated first and files in those directories are excluded.
Then, the includes and excludes are "processed" from the bottom up, not read
from the bottom up.

I'd be glad to be corrected by someone who really knows, but the .opt file
is a sequential text file and is likely read sequentially from beginning to
end. The options are evaluated and stored in structures in the running
instance of dsmc. Why the developers chose to use a bottom up
reconciliation of the includes and excludes is independent of everything
else in the file and the way it is "processed".

Andy, my experience the dsm.sys file is the opposite of Becky's and
consistent with your expectation. The first stanza on my UNIX systems is
used by default, and I must specify any other stanza that I want a client to
use. The only change to this would be if I have a dsm.opt somewhere in my
path that specifies a servername option.

Bill Smoldt
STORServer, Inc.

-----Original Message-----
From: ADSM: Dist Stor Manager [mailto:[EMAIL PROTECTED] On Behalf Of
TSM_User
Sent: Wednesday, September 01, 2004 5:10 PM
To: [EMAIL PROTECTED]
Subject: Re: Include/Exclude Question

I have always seen that the dsm.opt file IS processed from bottom up. How
it is read I don't know but everything not just the include/exclude list
appears to be processed from teh bottom up.

Take a windows client and put 2 "NODE" statements in it. Notice that it
will always use the bottom one to connect.

We have seen this with other things like TCPWidowSize. Someone had
accidently put that entry in two places in the dsm.opt. One at the top and
the other at the bottom. We adjusted the one at the top not realizing there
was another entry. Of course nothing changed. Once we removed the entry
from the bottom all was well.

I just ran the test above again using V5.2.3 on Windows XP.

Further, with TSM servers dsmserv.opt the same thing happens. When you run
the set opt command it adds an entry to the bottom of the dsmserv.opt file
so that it is read in first. This way if the default option was set above
it will use the one at the bottom of the file.

Andrew Raibeck wrote:
Interesting.... I'll have to look into this. If you don't specify the
DEAULTSERVERNAME option in dsm.sys, the default stanza should be the
*first* one in the list, not the last.

Regards,

Andy

Andy Raibeck
IBM Software Group
Tivoli Storage Manager Client Development
Internal Notes e-mail: Andrew Raibeck/Tucson/[EMAIL PROTECTED]
Internet e-mail: [EMAIL PROTECTED]

The only dumb question is the one that goes unasked.
The command line is your friend.
"Good enough" is the enemy of excellence.

"ADSM: Dist Stor Manager" wrote on 09/01/2004
11:49:16:

> My finding is that if you have several stanzas
> Server A
> ...
>
> Server B
>
> And you do a dsmc it will automatically connect to server B and not
server
> A. That is important to us because if we have an oracle database
connecting
> with one node name and the client files being backed up as a different
node
> name we use the dsmc -se=servera if we want to use servera or vice
versa.
> If we just typed dsmc we would get serverb. We had some issues with
some
> backup software that wouldn't let us specify a node name or a server
name so
> we had to make it the bottom most stanza so it would connect with the
node
> name that we preferred and then just used the -se option for everything
> else.
>
> Becky


---------------------------------
Do you Yahoo!?
New and Improved Yahoo! Mail - Send 10MB messages!


---------------------------------
Do you Yahoo!?
Win 1 of 4,000 free domain names from Yahoo! Enter now.

Reply via email to