"Bruce Arden" writes:

I have just had to reboot our filer for the second time in about 3 months. Both times the reboot failed with the error:

A CIFS I/O operation may be hung. Shutdown the system using halt

-d and report this incident to NetApp.

But I don't want to do a "halt -d". This causes a core dump which means that the filer may be down for about 10 mins. Is there a way of forcing a halt without using the -d switch (halt on its own just gives the CIFS error message again). I'm currently running 5.3.1D5.

• Bruce

PS The problem that caused me to reboot was that I was getting the error "Cannot find logon server" from PCs trying to connect to the filer. The reboot fixed this. Any ideas?

A suggestion which won't solve your problems, but may help the boot time: move savecore to the end of /etc/rc; it speeds up availability after a boot at the potential expense of corrupting your coredump.

PS: I have seen the ``can't halt because filer is wedged on CIFS users'' problem on older 5.x releases (can't remember exact numbers), but I think we've been OK since 5.3 (now running combination of 5.3D19 and 5.3.2D4, with the latter on our more loaded systems). I know this contradicts your experience (since you're running 5.3.1D5), but you never know; 5.3.2Dx may have fixed something related to this. :-/