On new installations kickstart fails because the system can't NFS mount the directory containing the kickstart file and the RedHat pinstripe CD contents. Error is "I can't mount the directory". Try manually typing in the NFS source mount path using both IP address and hostname with the same result. After installing the system using the http method and rebooting, the same NFS source mount path has no problem. As well we are getting random "Abnormal exits with Signal 11" at various points in the fresh installations using either the http or the ftp method.
Did you copy over the contents of both cds to the nfs directory? This error message is also given I believe when the NFS mounted directory does not contain RedHat/instimage (which comes from the second CD, thus if you are nfs mounting a loopback of CD1, it's not going to work). Also, did you see any errors in the logs on the NFS server regarding the failure to mount?
CD data is copied onto the NFS volume. ie this is not a loopback mount of an iso image. This is an NFS volume that contains the contents of both cd1 and cd2. The NFS server is a Network Appliance 720. There are no error messages on the NFS server side.
hmmm ... We are not having those problems here in test ... could you attach a copy of your kickstart file ...?
Here is the kickstart file we have been using. It is unmodified from the version we are successfully using to deliver RedHat 6.2 to our machines. --------------------------------------------------- lang en_US keyboard us # DHCP network --bootproto dhcp nfs --server cimnfs --dir /vol/vol0/linux # Ethernet and others # device ethernet 3c59x # device ethernet 3c509 # device ethernet eepro100 # device ethernet rtl8139 # device ethernet ne2k-pci # Partitions zerombr yesclearpart --all part /boot --size 25 part swap --size 100 #part / --size 750 part / --size 2000 # install # # mouse ps/2 --emulthree mouse generic3ps/2 # timezone MST7MDT # xconfig --server "SVGA" --monitor "generic multisync" # rootpw byte1me auth --enablenis --nisdomain wns_cim_prod --useshadow --enablemd5 # #lilo --append 'anything-needed' lilo --location mbr %packages @ Base # @ Printer Support # @ X Window System @ Multimedia Support @ Networked Workstation @ NFS Server @ SMB (Samba) Server @ Authoring/Publishing @ Utilities gdm autofs xntp3 # XFree86-100dpi-fonts pdksh wu-ftpd %post /bin/mkdir /mnt/RPMS /bin/mount 47.193.1.11:/vol/vol0/linux /mnt/RPMS/Cim/hpevectra.sh /bin/umount /mnt/RPMS
Please test this Brock.
after removing xntp3 from the package list, the above kickstart file works in our test lab using the pinstripe public beta (as well as later RC-1) ... Can you try a recopy, or otherwise try to verify that the data from CD was copied properly ... ? Iffy, flaky or otherwise marginal CD drives have been observed to do similar things in our test lab ... :(
Please reopen this if you are still seeing the problem and have more information on how to reproduce the issue ... we haven't seen it in testing here ... thanks for your report!
I work with Dwayne. We have just done this experiment with RedHat 7.0. We are kickstarting off of a bootable CDROM and using an NFS server for the install. The NFS install fails right away with "Can't mount <path to RedHat install>". This was when using our Network Appliance Filer as the NFS server. For fun, I moved the RH7 image to our HP-UX 11 servers and tried it again. Same result. Then for kicks, we moved the image to one of our other Linux boxes. It worked!! It was then I noticed that our Linux box was running mountd with the flag "--no-nfs-version 3". For even bigger kicks, I turned off the flag and tried kickstart again against our Linux box. Failed the same as before! It appears the NFS install fails if the NFS server is providing NFSv3 support. Forcing NFSv2 only seems to fix it. This did not happen with RH6.2. Hope that helps! We'd really like this to work!
NFS bug consolidation
*** This bug has been marked as a duplicate of 17069 ***