Thanks Scott.
We are setup using POLLING, and our Randomization is set to 10% with a
Client Action Duration of 5 days. Perhaps this is why it ran so much
later. The startup window is 720 minutes (12 hours). I can change my
Client Action Duration to 1 day, that will get me a start window of 144
minutes, run the schedule, then change it back. Are there any other
gotchas that might hit me there?
I placed commas without spaces in between the objects because that's what
the "Help" indicated to do. I am guessing I would need the double quotes
around the list of drives if I am using spaces. Correct?
As far as why I need to do this...
I thought I needed to do a selective backup of all directories on all
servers so that the ones that are stored on tape could be backed up again,
and then stored on disk only, eventually allowing the ones on tape to
expire off. We do have a few Netware servers, and a large number of NT/W2K
servers. So, in order to make sure we have everything moved to a the new
management class, I thought the selective backup would be warranted.
|--------+------------------------>
| | Scott McCambly|
| | <mccambly@ATTC|
| | ANADA.CA> |
| | Sent by: |
| | "ADSM: Dist |
| | Stor Manager" |
| | <ADSM-L AT VM DOT MAR|
| | IST.EDU> |
| | |
| | |
| | 03/18/2003 |
| | 10:00 AM |
| | Please respond|
| | to "ADSM: Dist|
| | Stor Manager" |
| | |
|--------+------------------------>
>-----------------------------------------------------------------------------------------------------------------------------|
|
|
| To: ADSM-L AT VM.MARIST DOT EDU
|
| cc:
|
| Fax to:
|
| Subject: Re: Selective backups
|
>-----------------------------------------------------------------------------------------------------------------------------|
I suspect your immediate action ran 7 hours later because your node is
using POLLING schedmode and you have randomization set to 50%. Immediate
action schedules are defined with a huge starting window, so although I
don't typically use POLLING, I've never understood how it would ever
work for those nodes (should be called "eventual" action schedule :-) ).
The job failed because of the commas in your objects list. Just type it
exactly as you would on the command line. And all you need is "C:\ D:\
G:\" - the asterix is not necessary, but shouldn't hurt.
Most importantly however is why you think you need to do this in the
first place. Unless you have excessively large ACL's on your
directories, they are stored entirely in the TSM database. They didn't
used to many versions ago, so what you are setting up used to be valid.
If this is your situation, then great!
I'm assuming that most people today only use DIRMC to avoid binding to a
management class that has NOLIMIT for object retonly.
Does anyone else out there use DIRMC?
Todd Lundstedt wrote:
>I have recently setup a Directory Management Class, some server-side
Client
>Option Sets pointing DIRMC to that management class, and assigned those
>option sets to nodes. So now, all new/changed directories should be being
>managed by that management class (disk based storage, no migration, etc.).
>I can see that going to each node and issuing a
>dsmc selective -dirsonly -subdir=yes c:\* g:\*
>and any other drives that exist will re-backup versions of all directories
>on the listed disk drives.
>
>But, that's a lot of work.
>So I tried an immediate action schedule on one of the servers.
>Action = Selective
>Options = -dirsonly -subdir=yes
>Objects = c:\*,d:\*,g:\*
>
>and that immediate action schedule finally tried to run 7 hours later, and
>didn't work.
>
>Is there anything I can do from the server to kick off a -dirsonly
>-subdir=yes backup of ALL existing drives in a list of nodes? From what I
>am gathering, the selective backup requires a filespace and I have not
been
>able to get *:\* to work yet. Or, do I need to even worry about this? My
>intention was to (eventually) get all backup versions of directory objects
>off of tape and stored only on disk.
>
>Thanks in advance.
>
>
>
>
|
