I'm not sure if this is a bug in the kernel, pxeboot, tftp or system-config-netboot. The symptom is that diskless clients panic during boot with this error message: RAMDISK: Compressed image found at block 0 EXT2-fs: blocksize too small for device. The initrd image on the server looks good: it's a valid gzip file containing a valid ext2 image that mounts with loopback. The ramdisk_size kernel parameter matches the uncompressed image size. The client kernel is 2.6.17-1.2600.fc6 (x86_64) The pxeboot command line is: label 2 kernel fedora-x86_64/vmlinuz append initrd=fedora-x86_64/initrd.img root=/dev/ram0 init=disklessrc NFSROOT=10.3.3.1:/diskless/x86_64/fedora ramdisk_size=28086 ETHERNET=eth0
append ramdisk_blocksize=1024 in your pxe cfg see also https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=210774
I've updated the system-config-netboot scripts to use initramfs and cpio. Anyone wants the patches?
Created attachment 146763 [details] switch to initramfs This patch changes updateDiskless to build a new style cpio archive for initramfs instead of the old initrd filesystem image. Lots of changes (actually simplifications) are needed in the init script to use switch_root instead of pivot_root.
Thanks for patch, it's was applied in version 0.1.42.