Hi All, I've got my autofs maps setup to NFS mount home directories on my mail server. This works fine when the clients with home directories are up and running. But if a NFS/automount client machine goes down, this causes disasterous results on my mail server. sendmail, procmail, and spamd processes get stuck waiting to access files on these NFS filesystems. Or sometimes if the directory was mounted while the client computer crashes, other processes can get stuck. How can we tell these processes to time out and continue without these files? When a processes gets stuck waiting for automount and NFS to respond, it is virtually impossible to kill that process. And as each processes gets stuck, the load is driven higher and higher until finally sendmail refuses conections because the "RefuseLA" value in sendmail is reached. Also, we cannot "umount -f <directory>" to force NFS to unmount that directory. We also cannot kill or kill -9 the autofs processes because they refuse to die. Our /etc/auto.master file looks like this: /home yp:auto_home rw,nosuid,soft Our yp auto_home file looks like this: client1 &:/export/home/& client2 &:/export/home/& If I run the mount command, these home direcotires are mounted with these values: automount(pid804) on /home type autofs (rw,fd=5,pgrp=804,minproto=2,maxproto=3) client1:/export/home/client1 on /home/client1 type nfs (rw,nosuid,soft,addr=168.192.211.202) client2:/export/home/client2 on /home/client2 type nfs (rw,nosuid,soft,addr=168.192.211.142) So does anyone know what we are doing wrong in our autofs/NFS setup? We are running RedHat 8.0, nfs-utils-1.0.1-2.80, autofs-3.1.7-33, kernel-2.4.18-24.8.0, and sendmail-8.12.8-9.80. Thanks, Cheryl -- Cheryl Southard cld@xxxxxxxxxxxxxxxxx -- Psyche-list mailing list Psyche-list@xxxxxxxxxx https://www.redhat.com/mailman/listinfo/psyche-list
