I have a few questions regarding Linux and Windows masters servers.
Currently, there is one windows master server on-site, and the previous admin team purchased a veritas appliance located at the DR site. However, the appliance at the DR site is running the show. It is the master server with the Catalog and what all clients consider the master server.
I don't understand what the original intent was with this configuration.
I have gathered that the plan was to duplicate all disk backups to the DR site (Currently done using SLP's). In the event of a failure, the data could be restored and cover a loss of the on-site backup system. The configuration at the DR site would then take over the backups. It appears that this configuration was not successful, and the appliance at the DR site is now the main and only server.
Is it possible to have multiple Master Servers sharing a catalog?
They could have planned to set up an HA Master Server and failed. But is this even possible using an onsite Windows and offsite Veritas appliance?
A mystery indeed. To answer your questions first.
As for the windows master - does it run any backups? How many media servers do you have? Which master do they work with?
I can imagine, that they may have considered using AIR to replicate backups from the windows master to the DR site (that is the a way one would set up an environment like yours). How it ended up as it is though is a mystery.
Depending on what else you have in the environment you may be able to put it right (so to speak) - mostly by setting up AIR, and start backing up the clients using the windows master. How complicated this is though is hard to say with the limited information to hand.
The current Master is not doing anything, but some clients have both masters listed on the host properties. There are three media servers, all located on-site.
The tape library is configured to one of the media servers, and half a petabyte of disk storage both on-site and offsite are configured to the DR-Site Master and the other two media servers.
Everything is set to work with the Master at the DR-Site. Nothing is going to the on-site Master server, and I can just shut down the on-site windows server as it serves no function.
All the backups are written to the on-site disk storage and then duplicate via SLP to the DR site.
In the DR site, there is the 5240 appliance with a 20TB MDSP pool for some reason that has never been used.
What is the target for the SLPs that duplicate backups to the DR-site if not the MSDP pool on the appliance? Are there other media servers/diskpools/tape at the DR-site?
From what you have said it may be challenging to make use of the Windows master. It's doable with thought and planning - basically stop using a media server move the backups from it (or let them expire) and then reinstall using the master as the WIndows master. Once the first is done, repeat with the next one (it sounds simple, but needs careful planning).
Alternately, just leave everything as it is and repurpose the WIndows master for solmething else.
Or relocate the appliance to on-site, and the Windows master to the DR-site and reconfigure things that way.
Lots of options.
There is another non-veritas appliance that NetBackup duplicates the backup to using SLP. I am not sure why the MSDP pool was purchased as 20TB is too low of a capacity for this environment, and there is no licensing available even to use it.
It looks like leaving everything with the master server at the DR site was the solution ultimately taken. However, today this is causing issues with consolidation.
Moving the master server back onsite was my initial idea, but I ran into issues with the way NetBackup was configured. However, it looks as infuriating as the process will be; this will be the best way forward.
Licensing shouldn't be an issue as it is normally based on what you are protecting not the amount of backend storage allocated to MSDP pools (there is an exception for virtual appliances, but not relvant here).
There may be other issues in your environment, but changing the IP address of the master server to bring to change where it resides is not that complicated (unless you are using static hosts files everywhere).
There is no way for two masters to live-share the same catalog. You can't even have a media servers that works with two Master Servers. But clients can have one or multiple master servers. That said the idea must have been to run backups on the local-windows master and then replicate them to the remote-appliance master server. In the event that the local site was down you could activate the policies on the remote site and resume critical backups through the WAN. If client side dedupe was enabled this could mean some policies would not need to send all data across. Maybe a COB test was done and they left it that way or issues with the local-windows master forced them to run this way and decided not to go back.
If the goal is to have all backups available at both sites this is the least efficient method, if the goal is to have data offsite as quickly as possible this method works until but you have to transfer the data twice across the WAN and backups might take longer to run.