I have a PXE Genkernel squash image that was created using the instructions found on the Gentoo wiki (http://gentoo-wiki.com/HOWTO_build_a_LiveCD_from_scratch). This has been up and running for several years without a problem and has been updated frequently with negligible issues. But recently within a month or two, after a few updates, including a genkernel kernel update the PXE bootable thinclient isn't working. Genkernel version is 3.4.10-r1 Here is my pxelinux.cfg/default menu item for this thin client: Code: label dev menu label Gentoo (dev-only for testing) kernel /pookydev/boot/kernel-genkernel-x86-2.6.26-gentoo append initrd=/pookydev/boot/initramfs-genkernel-x86-2.6.26-gentoo looptype=squashfs root=/dev/loop0 nfsroot=172.16.4.23:/diskless/pookydev real_root=/dev/nfs ip=dhcp nodevfs cdroot panic=10 Reproducible: Always Steps to Reproduce: 1.See description 2. 3. Actual Results: Here is the error from booting into the thinclient. Code: >>Attempting to mount NFS CD image on 172.16.4.23:/diskless/pookydev !!The filesystem mounted at /dev/nfs does not appear to be a valid /, try again !!Could not find the root block device in . Please specify another value or: press Enter for the same, type "shell" for a shell, or "q" to skip... root block device() :: Expected Results: Booting normally. This was solved by downgrading to 3.4.9.
Is there a file called "livecd" at 172.16.4.23:/diskless/pookydev ? If not, you shouldn't be using "cdroot" on your command line.
Here are the files in /diskless/pookydev: boot/ livecd livecd.squashfs There is in fact a file called livecd. This format has not changed, and has been like this when it was working. It was the change from genkernel 3.4.9 to 3.4.10-r1 that has caused my issue. The problem has been solved by reinstalling 3.4.9 and using genkernel to rebuild the kernel and initrd. If you look at /usr/share/genkernel/generic/linuxrc ~line 410 you can see: if [ -d ${NEW_ROOT}/dev -a -x ${NEW_ROOT}/sbin/init ] then break else bad_msg "The filesystem mounted at ${REAL_ROOT} does not appear to be a valid /, try again" got_good_root=0 REAL_ROOT='' fi (taken from Gentoo forums, http://forums.gentoo.org/viewtopic-p-5202867.html#5202867) This seems to indicate the root of the issue as PXE/Squash genkernel thinclients do not have a /dev or /etc/init in the root directory. I hope this helps.
I've just faced this. This is a bug in the generic/linuxrc script. After correctly mounting the nfsroot, it performs checks on the mounted nfs filesystem expecting to find a valid baselayout and ignoring the possibility of the nfs partition containing a cdroot image. The script checks for the presence of /newroot/mnt/cdrom/dev as a directory and the existence of an executable /newroot/mnt/cdrom/sbin/init. To get this working quickly I have just patched it to break out of the loop if the nfs mount succeeds with no error, leaving the other checks to the follow on script.
We'll be looking into this for 3.4.11's release.
*** This bug has been marked as a duplicate of bug 246759 ***
(In reply to comment #5) > > *** This bug has been marked as a duplicate of bug 246759 *** > Actually, that bug is a duplicate of this one. When will this be released in portage?
I know this one came first, but I don't really care. The resolution information is on that one, so this one gets marked as a DUPE. We'll do a new release when we do one. I don't have a date for you. *** This bug has been marked as a duplicate of bug 246759 ***
I remove this bug from tracker bug #245389 "Features for inclusion by 3.4.11" as it does not seem related.
