John R Pierce wrote: > On 3/2/2015 11:56 AM, m.roth at 5-cent.us wrote: >> Well, we got it working. However, the issue we're now worried about is >> users creating files and subdirectories. Do we need to worry, and if so, >> is there some way to reserve inodes < 32k table, other than creating >> tens of thousands of dummy files now? >> >> We don't want, a year or two down the road, for this system to be >> running, and suddenly everything's broken, because all lower inodes have been >> used. > > inodes of individual files on the file system behind NFS don't matter. > the only issue is, NFS defaults to using the folder's inode as the fsid. > > my solution to this is to just generate an arbitrary unique integer fsid > on each share in /etc/exports on the server... just add ,fsid=# in the > (options) field, where # is 1,2,3,4... > When I first created it, I tried that, and it complained. I have another system, a new one, which we'll be using for backups; I can try it on that (the original one's *really* not somewhere to play, since my manager's manager has a team all over it, working up to a very high profile demo a week from now....) Thanks. I'll look at this. mark