kriso@lombard.dartmouth.edu (Kris Olander) (03/16/91)
I know this is probably a shot in the dark, but... I'm getting a bit desparate. We have a cluster of HP9000 model 340s running HP-UX 7.0. One is a file server and the rest are all diskless clients. We are running under HP's CNODE environment -- I don't believe there is any other method (if this is an incorrect assumption, please let me know!). Here's my dilemma. We have been using the Andrew File System (AFS) for years. AFS generally likes to have local disks on each workstation for it's file cacheing system. Unfortunately, when we spec'd out our HP systems, we chose configurations which would give us more CPUs. -BIG MISTAKE- We now have more CPUs, and less useful systems. AFS has a new feature that will allow us to use RAM as a local cache area. This will probably work for us, however when we try to start up the AFS daemon on our diskless HPs, we get: ---------------------------------------------------------- Starting AFS cache scan...setting chunk size to 2^13... initializing memory cache...using 8192 block size... allocating 512 dcache entries... panic: diskless_mem_get: Request greater than max. allowable message. ---tons of register values...--- ---------------------------------------------------------- 512 dcache entries == about 4 Meg of RAM. We have 12 Meg or RAM available on this particular diskless 340. Now I know that the afs deamon is trying to do a mount of it's file system on /afs. What are the implications of a diskless cnode attempting to mount a file system? Can anyone tell me what 'diskless_mem_get: Request greater than max. allowable message.' means? -- Kris Olander --------------kris.olander@dartmouth.edu----------------