FOG Storage Nodes reaching out to each other instead of just the Master Node?
First of all, the FOG Project has helped my team achieve what we thought was impossible before and I first want to thank everyone involved in creating and maintaining it.
We currently have FOG ver.1.5.9 setup as the following:
Master Node - Virtual - Main Site - 1
Storage Node with DHCP - Physical - Off-Site Location (Segregated/Silo’d from other sites other than Master Node) - 170
Host to be Imaged - Physical - Approx 10 per site - 1700+
This is actually working perfectly for our use-case in being able to remotely image a host at the click of a button utilizing locations but I did notice something that I wasn’t expecting yesterday when adding a new off-site storage node before segregating it with the ASA.
I noticed (using NHRP) the Storage Node attempting to reach out to seemingly all other Storage Nodes (X.X.X.0/24) and taking up a large number of traffic through tunnels from the router. Once the ASA kicked in and blocked the traffic from getting there, the tunnel issue resolved but now i’m wondering what the intended purpose of this is and how can i turn it off so that there isn’t unnecessary more work from the ASA if we aren’t using this functionality?
Thanks in advance!
** Extra Context of Environment **
Hosts are all connected to FOG Storage Node via same switch and subnet/vlan
That switch then routes to another MAIN Switch at site where it goes through ASA then Router to Main Site where Master Node resides.
Everything is currently working as intended… just curious about this particular function of Storage Nodes attempting to reach each other and how to disable it.
Wayne Workman last edited by Wayne Workman
Are you using the location plugin? If not, that could explain a lot, as FOG will image from whatever storage node with open slots rather than a specific one in the same location as the host.
@stealthlegend AFAIK storage node A could care less about storage node B-X. The only thing a storage node cares about is the master node, because that is where the database is. Replication only happens between the master node in the storage group and the slave storage nodes.
With that said, it would be interesting to know what traffic was trying to flow between the storage nodes. What network ports were in play when this was happening. Using tcpdump on one of the storage nodes that are unblocked might give a bit more detail.
Now computers with the FOG client installed need to have access to the master node it find out which is their defined storage node, I can see that for off site traffic. As long as you are using the location plugin and have the storage nodes and workstations assigned to the location very minimal traffic will go across your WAN.