In article <199409091433.AA17252 at cvx12.inet.dkfz-heidelberg.de> D.Wolf at dkfz-heidelberg.de (Detlef Wolf) writes:
>Hi,
>did you also experience this:
>if the database directory is on an NFS-mounted file system
>and the NFS-server machine has high load than the ACEDB process
>goes into an eternal disk wait.
>This was between SunOS -- SunOS,
>SunOS -- DEC alpha, SunOS -- (some mainframe NFS server)
>moral: run the ACEDB process on a CPU local to the database disk.
This is doubtlessly my fault.
Try it again without TRUST_LOCKD defined (if this constant is still
in the 3.3 make, otherwise make the switch from the F_SETLK fcntl
manually), and see if it doesn't crap out.
If that clears it up (at some loss of distributed performance),
then the code will need to be reworked to distrust the remote lockd daemon.
I neglected to test for high loading against the vendors supplied
daemon, but I'm not with the project anymore, and so can only wish you'll
the best of luck here.
--
Henry J. Cobb hcobb at fly2.berkeley.edu
All items Copyright (c) 1994, by their respective authors, permission is
granted to redistribute as long as proper credit is given.
