Storage node setup?
Im having to rebuild my master server and am curoius on how to go about this without loosing my all my images.
Currently I have a storage node that has all the backups/synced images on it. So here is what I am looking at doing.
Question 1: Can I rebuild my server, set the storage node as master, wait for the images to sync/transfer to the main server, then switch the main server back to master.
Does this work? has anyone tried it?
Question 2: When I transfer the server/node over to my corporate network, they will be on different subnets. I read the Wiki and it stated they had to be on the same. Is this still an issue and why do they need to be on the same?
Question 3: The benefits of a storage node are not very clear, can someone please clarify what the benefits are besides data redundancy and increased queue size by 10? Has anyone really tested the load balancing?
Thanks in advance!
We wanted to know what your server version was b/c there are may things you must do if you are moving from 0.32 to 1.x.x. So many things there is a nice wiki for it.
- I guess I haven’t as of yet. I keep rebuilding my images yearly and deploy soon after they are finished. You could always ftp them over to the “master” server before connecting the storage nodes. then just create the images in the web gui. as long as you drop the images in /images and name them exactly the same then you should be ok.
You may also want to export your hosts and export you database. This will get you back in action faster.
I am running Fog v1.2.0 on Ubuntu 12.04LTS.
The reason I am rebuilding the master server is I need to switch it to a raid configuration with a swap drive. Currently it is set up in an LVM.
- Ok, I’ll try it. how else do you replace the images if your main server goes down?
- Got it. Thanks!
- Ill have to look into that. Currently we have about 6 TB of storage on the master and about 13 TB on the node. not too worried about storage just yet, just more concerned about the load balancing.
Why are you rebuilding the “master server”? Is this due to a hardware change or upgrade?
Also what versions of FOG are we looking at? Old server vs. New server?
- You can increase the queue size as far as you wish but the more you add the slower the output of that node. If you queue and connect 100 hosts your speed may be 100mb/min but if you decrease to 20 then it may be 1GB/min (environment variable dependent). You can also have the node do all the nfs work and the server only house the mysql decreasing the demand of a large harddrive for the images to be stored.