I have a centralized job system & several (45) render nodes that pull files from that central job system. Some files need to be read continuously, so I've made a symbolic link in the job system to each machine's /tmp dir (`ln -s /tmp/foo /job/foo` where /job is an nfs share). This does the trick pretty well, but all the requests to that symbolic link load up my file server... I'm seeing a load average of about 1.3 and nfs tops at about 70% processor usage.

I can live with that, but if it's possible to cache that link so it doesn't pull the link from the server each time a node tries to read from the server, it'd make my overheating server happier.

Is that possible?