home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: comp.sys.sgi
- Path: sparky!uunet!cs.utexas.edu!hellgate.utah.edu!cs.utah.edu!ldl
- From: ldl@cs.utah.edu (Louis Langholtz)
- Subject: Problem running amd on multiprocessor IRIX host
- Date: 25 Jan 93 13:07:06 MST
- Message-ID: <1993Jan25.130707.10015@hellgate.utah.edu>
- Organization: University of Utah Computer Science
- Lines: 27
-
- We've been seeing a problem with amd running on a 4-processor sgi IRIX host.
- I'm wondering if anyone else has seen this problem and if anyone has a fix for
- it that they will send me. If you respond please email me directly as I don't
- often get the chance to read this newsgroup.
-
- Basically after running smoothly for several weeks the load average on
- the host will start to climb without stopping. Amd gets into a blocking
- situation where amd kids come about and are unable to resolve requests.
- The kids just sit invincible in the kernel (kill -9 as root doesn't kill'em)
- and load down the machine. We've seen this problem with several versions
- of amd, most recently:
-
- amd 5.2.2.2 of 1992/05/31 16:53:21 bsd44-alpha #0: Sun Dec 27 22:12:57 MST 1992
- Built by ldl@lotus.cs.utah.edu for a mips running irix4 (big-endian).
-
- Interestingly the same versions of amd have always worked fine on single
- processor sgi IRIX hosts which points strongly at there being a bug in
- IRIX's handling of multiprocessing that causes amd's troubles. So why doesn't
- sgi's automounter have the same problem? Did sgi hack in a work around to
- their own automounter? Can someone from sgi confirm any of this?
-
- Thanks much.
- --
- Lou
- --
- Lou
- unixsys
-