----- Original Message ----- | Ole Holm Nielsen wrote: | > We have CentOS 6 manual installation working by PXE booting from a | > RHEL5.6 | > PXE/TFTP server. However, when we add a Kickstart file in the PXE | > configuration: | > | > kernel CentOS-6-i386/vmlinuz | > append load_ramdisk=1 initrd=CentOS-6-i386/initrd.img network | > ks=nfs:130.226.86.4:/u/rpm/kickstart/ks-centos-6-clean-i386.cfg | > | > then the CentOS 6 client install reports "Unable to download the | > kickstart file". | > The console 3 reports "failed to mount nfs source". | | This problem has been resolved. A silly editing error replacing 5->6 | also | changed the IP-address :-( With the correct IP-address Kickstart works | correctly with an NFSv3 server as shown above. No need to upgrade to | NFSv4 | and Kerberos :-) | | For the record, it is in fact possible to add NFS mount options to the | PXE | APPEND line, as documented in | http://fedoraproject.org/wiki/Anaconda_Boot_Options. | To explicitly force an NFSv3 mount you may add the following NFS mount | option: | ks=nfs:nfsvers=3:servername:filename | | Thanks again for everybody's help. | /Ole | _______________________________________________ | CentOS mailing list | CentOS@centos.org | http://lists.centos.org/mailman/listinfo/centos
I would still push for a move to HTTP. I manage about 1500 GNU/Linux, CentOS 5 workstations and about 3 months ago moved away from NFS and to HTTP. This was for additional security, NFSv3 being less secure than NFSv4, as well as for scalability reasons.
After making the move from NFS to HTTP I found that my installations went from 20 minutes for a @core installation to ~5 minutes for the same install. I am also now able to do more 'interesting' things like heavy HTTP caching, load balancing as well as other things to "tune" my installation path. You really should consider it if you've got the resources. ;)