While working on setting up my software deliveries, I have run into an issue (maybe two) that I cannot figure out cause of.
Here is the setup..
- Core server is located at my home office with remote locations connected via MPLS (1.5 - 3mb pipes...low bandwidth). I am attempting to utilize UNC path preferred servers at my remote locations.
- Software delivery setup using a Policy method "Required Installation" modified to "Require high speed connection" with Peer Download, Attempt Preferred server, and Allow source selected.
Outcome..
Software deliveries fail to the remote locations due to "insufficient bandwidth". If I set the policy to just require "Non-Ras connectivity" then they succeed. I am guessing this is because the clients are not utilizing the local preferred server. I also tried installing the software on a machine (through a different non bandwidth aware policy) at the remote site so the client would utilize a peer download, however when looking at the task information on a subsequent failed install on another machine at that location, I do not see where it even attempted to download from a peer or preferred server (though I might be reading it incorrectly). There are two things I have noticed while looking into this that I believe may indicate where the issue is. Sometimes the preferredserver.dat file can be located on the clients, and does have the correct server listed in the file, however it looks like the preferredserver.dat is removed when the client attempts to utilize it. And when looking at the proxyhost.log I can see where the client pulls the preferred server.dat file down...but then it looks like it tries to connect to the server via http which fails because IIS is not installed on the preferred server since I am attempting to only utilize UNC paths in my deployments.
Sidenotes..
- Testing the credentials on the preferred server properties works and data replications succeed.
- I have attempted to rehash the Distribution Packages (and tried multiple different packages with the same result)
- Clients can ping the preferred server (and the firewall on the server is disabled at the moment)
- The UNC share paths are identical on the core and preferred servers.
- I am using FQDNs in my paths on the distribution packages and preferred server setups to avoid potential DNS resolution issues (\\servername.domainname\packages)
I'm not sure if the peer downloads failing have the same root cause as the preferred servers not being utilized, but I would like to correct both issues if they are separate. Attached are some of the log files that are relevant. I'm probably just missing something simple in my configurations as this is my first deployment, I just can't figure what I have missed. Any assistance is greatly appreciated.