Nfs problem

[Anonymous]

I have a renderfarm that is mounted to a number of OSX servers via nfs, they render maya, shake, and mental Ray using the rush render que. I am automounting the mounts at /Volumes/ which for some reason is a link to /automount/static/Volumes/ which is a link to /private/automount/Volumes/.

My problem is, when a program ( for example, shake) calls a script that is in /Volumes/mount/project/x, it can’t find it, It only sees things as /private/automount/Volumes/mount/project/x. So even if shake finds the project script, it can’t load all the macros that we keep on the server, because the app is looking for them at /Volumes/mount/project/macros, and even though the cmdline can navigate to that path, it would only be able to find them at /private/automount/Volumes/mount/project/marcos. Why is this? Am I mounting things wrong? Below is how I am currently mounting things. Is there any way to just mount things straight to /Volumes? and not worry about the whole /private/automount thing?

rd-019:~ root# nidump -r /mounts .
{
“name” = ( “mounts” );
CHILDREN = (

{
“name” = ( “100.100.102.4:/Shared Items/rushlog” );
“vfstype” = ( “nfs” );
“opts” = ( “-P”, “-i” );
“dir” = ( “/Volumes/rushLog” );
}
)
}

[andrina]

I’ve got something similar going on here – however, we’re not using static mounts, we have dynamic mounts instead – not sure that would really make a difference to the issue you’re seeing at the moment though.

FYI though – this is what one of our mounts looks like:

sudo nidump -r /mounts .
{
  "name" = ( "mounts" );
  CHILDREN = (
    {
      "name" = ( "servername:/mount/point" );
      "vfstype" = ( "nfs" );
      "opts" = ( "net" );
      "dir" = ( "/Network/Servers" );
    }, 

Your clients that are creating the scripts though – how do they have the NFS drives mounted? Are they being mounted by the client in the finder, or are the auto-mounted also?

We have found that Shake does tend to keep the whole /private/var/automount… in the script, even if your compositors select their file-in nodes from a simpler path. I.e. we have anything that’s automounted into /Network/Servers linked with symbolic links to /mnt/mountname and all of our paths are then called from /mnt – due to Shake and it’s want to include the whole path we wrote an in-house script that then got included in the render submit script to parse the Shake script, clean out the /private/automount mess and replace it with the correct paths… I don’t know if this sort of scripting may solve your issues.

[Anonymous]

Does anyone know how to bypass automounter completly? Is it possible? If I want something to be mounted in /Volumes/mountpoint can I just mount it there without having to worry about the automounter /private/automount stuff?

[Author]

Posts