I'm working with various Solaris based distributions (OmniOS, SmartOS, Nexenta) and I have a highly compressible and deduplicable data set on a very small amount of high performance storage.
The problem is that the space free announced by NFS to the clients is the space seen as physically free. But since I can deduplicate and compress the incoming data the actual space requirements have little or no bearing on the source data file sizes. Where this causes problems is when I want to copy a large file or group of large files, the client checks to see if there's enough space free and it sees that there isn't, even though I'm going to deduplicate over 90% of the file's contents.
So in this type of environment, is there any way to override the value that nfsd presents to its clients?