Using Ignite for DR

[Bruce Owen Lee]

I have a question concerning Ignite/UX and DR. I make weekly images of my servers using make_net_recovery. This works well if I intend to recover systems in the data center from which they were made. However, I recently attempted to recover several servers in a DR site simply to test the concept. I created an Ignite tape of the Ignite server and restored it onto a server at the DR site. It contained all of the server images and config files but I had to give it a different IP address. The DR site did not have the same IP address ranges as my data center and the clients failed the install because the NFS (Ignite) server was at a different address than what was put in the archive.cgf file. I hacked the archive.cfg file for the client and got a little further but still ran into numerous errors when the client tried to mount the image directory.

Should I have used make_sys_image and clone the servers? This seem like it would take for time to create a config file for each image ( I have dozens of servers).

If I can't create a network at the DR site that has the exact IP addresses as my primary site, how do I use Ignite to recover systems at a DR site.

Please help.

[Ganesan R]

Hi,

Yes. ofcourse, you should have used make_sys_image to install DR clients. It needs some modifications to restore a client image on some other servers. Because ignite is storing the client images with reference to client's MAC address, client host names and client's configurations. So you cannot directly restore these images to some other clients.

As you said, NFS also come into picture. You can consider to have seperate ignite server on DR site to have DR site client images.

[Doug O'Leary]

Hey;

Most DR exercises *use* the same IP addresses as their production counterparts. Changing the IP addresses can certainly be done but, if necessary, should be step 2.

1. clone systems.
2. update IP addresses.

There will be any number of things affected by the changed IP addresses... NFS, licensing, routing, end user application configurations, etc, etc.

For our DR exercises, we use make_net_recovery images and clone the systems. Regardless of which way you go, realize that any non-vg00 volume group/lvs will have to be restored in some other manner.

HTH;

Doug O'Leary

[Rita C Workman]

We use our ignite tapes for DR. But we don't use our Ignite Server for DR. You can, but as you say you now have a some things you have to address first to get it to work. And what happens if the IUX tape copy of your server is bad. [Then everything stops-you put all eggs in one basket-ouch].

** One thing that helped us was our DR Site, we use HP's (not the other big DR vendor), so we can actually rebuild same network (i.e. internal subnets) down there on that network. **

I took a different approach. Although I have an Ignite server here...I felt keeping DR 'vanilla and simple' might be a good thing.
So ignite tapes for individual servers are made and rotated to our DR site. In the event of a disaster, we are not dependant on ONE box to get us ALL boxes. Each box is brought up on it's own tape. Resolution is key!!! We have an 8 node cluster, and for resolution we require that EVERYTHING resolve to the pkg name (example-Oracle tnsname so users station, jobs, etc. never require attention.). Only minimal, if anything, to change. We take our 8 node cluster in DR, recreate each server to single node cluster. It takes maybe less than minutes or so per box to do this.

Get the backup server rolling and we start restoring data. We daily scp our Data Protector database down to a server at the HP DR site. So we are always ready to restore once the tapes get there.

Take a look at your DR plan. If you want to go with restoring your Ignite Server than you need to test this out and document every step to do this. And weigh the value of everything in one basket point. Or, if this is not working well for you - create a new plan.

Just a thought,
Rita