I have configured various SLED12 machines to automount users home directories and some software packages using NFS. I am seeing on seemingly random machines the following error in /var/log/messages when the automount fails:
2015-09-03T11:45:19.732322+01:00 this-machine rpc.gssd[7911]: ERROR: gssd_refresh_krb5_machine_credential: no usable keytab entry found in keytab /etc/krb5.keytab for connection with host remote-machine
2015-09-03T11:45:19.732663+01:00 this-machine rpc.gssd[7911]: ERROR: No credentials found for connection to server remote-machine
2015-09-03T11:45:19.737695+01:00 this-machine rpc.gssd[7912]: ERROR: gssd_refresh_krb5_machine_credential: no usable keytab entry found in keytab /etc/krb5.keytab for connection with host remote-machine
2015-09-03T11:45:19.737938+01:00 this machine rpc.gssd[7912]: ERROR: No credentials found for connection to server remote-machine
Other SLed12 machines can connect without issue.
I do not use Kerberos authentication and never have. In the Yast2 setup page for NFS Client on the NFS Settings tab, Enable NFSv4 is checked (I have tried unchecking it but it made no difference), Enable GSS Security is not checked.
Any insights would be much appreciated as I a starting to tear my hair out (and I don’t have much to spare!). Never had this issue with SLED11.
The errors you quoted are somewhat cosmetic, and shouldn’t actually result in failed mounts. Are you sure that you aren’t seeing these messages on both successful and unsuccessful mount attempts?
Changes have been made to prevent those messages, i.e. to avoid loading rpc.gssd if there is no krb5.keytab file. You could update your SLED to current maintenance, to get those changes (which were made public just this week.) The specific package involve is nfs-client, but I would always update nfs-kernel-server at the same time. And you might chose to update nfs-doc while you’re at it… all those packages come from nfs-util source code.