FOG 1.5.2 on all servers/nodes.
Main server: Ubuntu 14.04.5 Server
Storage Nodes: Ubuntu 16.04 Desktop; Debian 9.4
I have a replication issue that I have not been able to overcome, and I’m sure that I’m missing something minor. 1 default site, and 3 remote sites with a storage node at each remote site. Each remote site is set as its own Storage Group. I have 2 images that I want to replicate to each remote node, 2 images that I want to replicate to 1 remote node, and 2 images that I want to replicate to the other 2 nodes. The initial replication worked well, everything replicated where it was supposed to. I have the location plugin installed and it seems to be working correctly. Images are only uploaded to the Default storage group.
My problem is that when I update the image at the Default node, the FOG Server, I can see where the files are updated on the remote node, but when I image a computer from the remote storage node, the image is not the same one as I uploaded at the default location.
Here are some screenshots to show my setup:
Storage Group section for one of the images I want to be replicated to all nodes
Storage Group section for one image that I want to replicate to a specific node:
From the remote storage node:
If I delete the image folder from the remote storage node, replication kicks in, and then everything is up to date.
What am I missing?
Wayne Workman last edited by
The master node in each storage group must be a full fog server (this one I’m 80% sure of).
Question about this condition, how would you handle the databases of each full fog server if each master node is a full fog server? I’m sure there is some tricks that have to occur but there has to be an easier way. The way I have my storage nodes is I have 18 sites and one full fog server. I have set various nodes as a master to their respective storage groups and have storage nodes within those storage groups that are not masters. Images will replicate within the storage groups as defined in the image settings with only one full fog server and its database. With the location plugin, I use the inits and kernels from the storage nodes that are masters to that particular storage group. I think it works great to be honest. The only dependency is when registering a client, you would have to make sure the image for that particular machine exists in the storage group for that location.
Would having full fog servers for the masters nodes be easier? At least in the sense that ipxe doesn’t occur across the WAN?
You have a couple of issues here.
- Replication only happens from Full fog servers to FOG storage nodes.
- The master node in each storage group must be a full fog server (this one I’m 80% sure of).
- Each storage group can only have 1 master node and many storage nodes.
- Only the master node in a storage group can capture files (the master node in a storage group doesn’t have to be a full fog server)
- Replication is either on or off for both overall replication and for images. i.e. you can’t have 2 to replicate to storage node A, 1 to B and 3 to C in the same storage group. All members of the storage group will get all images from the master node.
If you want images to flow from the root fog server the storage nodes must be in the same storage group as the root fog server. What you have setup will not replicate anywhere. Each storage group has one master and not storage nodes, so there is nothing to do.