Multiple Storage Node Problems

  • I am trying to set up a FOG system with 3 storage nodes to help image multiple computers at a time, and I am having problems getting them to sync the images over from the main server. All of the servers are using a minimal install of Ubuntu 14.04.4. The server is running FOG 7202 and the nodes are running 7218, just because they were downloaded at different times. Here is the story of all my troubles…

    I set up the main server and got 4 images uploaded to it without much trouble after I got all the DHCP issues sorted out. I built the first storage node and got it set up without any problems at all, and the images started to sync over. The other two nodes took a little more to get them going because of problems with the static ip configuration, but I got them up and running after multiple attempts. In the web interface I have all of the nodes added and I can see the hardware specs, so I know they are communicating with each other, and the passwords are correct because I copied them from the ssh session directly into the web interface. But I can not get nodes 2 and 3 to sync with the server.

    I read that it may not start to sync until I added another image, so I did, but that image did not sync with anything and is just on the fog server. I am out of ideas, and wanted to know if anyone else has done something like this. I have a project coming up soon where I will need to image 600 computers, so that is why I built so many nodes to help cut down on the time needed.

  • The FTP credentials issue is likely the issue with replication as well.

  • Moderator

  • I appreciate the quick reply, I did not know about the log in the web interface, and had been going by the log file in the /opt/fog/log/fogreplicator.log. The web interface is giving the error “/var/www/fog/lib/fog/fogftp.class.php, Line: 106, Message: ftp_login(): Login incorrect”

    Where would I correct this?

  • @chogue23 There are logs for the FOGImageReplicator inside of here, let us know what you find:
    Web Interface -> FOG Configuration -> Log Viewer -> Image Replicator

    also - even though there is such a small difference in version, things change all the time and it’s best to have the main and all nodes on the exact same version. For all I know, it might fix the issue for you.

    Also - a single FOG Trunk server is stupid fast. You can push a 40GB image to a machine over 1Gbps network in about 6-7 minutes give or take, assuming HDDs on both ends can keep up. Something to think about… if your image was 40GB, with one fog trunk server unicasting to 600 machines, with the maximum connections to the node set to 2, and without a single moment of interruption, you could probably get done in around 60 hours flat.

    Now, adding nodes can increase performance, but only if the nodes are strategically placed. An example of how to do it wrong would be to put all nodes on the same core-switch that the main is on, which has a uplink of 1Gbps to the router, which then distributes traffic to satilite campuses. Because of the 1Gbps up-link bottleneck, your stuck with 1Gbps throughput even though you have a dozen nodes on the switch. They must be strategically placed on your network or it won’t accomplish anything - and you are the only person who would know where the strategic spots are.