This post is more than 5 years old
4 Posts
0
460
August 10th, 2015 08:00
Unique exlusions for numerous servers
I was wondering how people handle exclusions for servers when you have a number of servers that all have unique exclusion requirements. So in my environment I've got 100 or so physical Linux servers; all of those share some basic Linux exclusions, but then probably 30 of those are running different applications that all have their own unique exclusions that aren't related to other servers, so one server may exclude /etc/customapp1 and another /opt/custom app2 and some have multiple directories that need excluding that are only unique to that server.
So, is there a best practice or just personal preference? I know technically you can add all of those exclusions to one dataset, the only thing I've seen I don't like about that is all the exclusions are in one big list and it ends up being hard to see which exclusions belong to which server and with 30 servers with multiple exclusions, you end up with one big list. The other way would be to create separate datasets for each app needing custom exclusions, which you end up with a lot of data sets. The other way I've seen mentioned is using an avtar.cmd file, which I guess you'd only know what's there by checking each server, though with TSM we're using now that's how it is I suppose.
Anyways, any tips/thoughts would be greatly appreciated.
ionthegeek
2 Intern
•
2K Posts
0
August 10th, 2015 13:00
This mainly comes down to personal preference.
For my part, I'd recommend using multiple datasets. I had an experience once where a customer was maintaining a huge exclusion list and managed to accidentally exclude all his volumes, which he discovered this during an important restore. I'd recommend against huge exclusion lists since it's possible to generate conflicting excludes and give yourself this kind of a nasty surprise.
The avtar.cmd file is a reasonable approach for small numbers of systems but it gets unwieldy fast.
gdFrank00
4 Posts
0
August 11th, 2015 10:00
Thanks Ian, that was the way I was leaning but just wanted to get some feedback from the community first if that was the way to go.