From Bugzilla Helper: User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.6) Gecko/20040207 Firefox/0.8 Description of problem: Running up2date to install the latest kernel fix generates the following error (note: I'm using LILO as boot loader) : <snipped traceback statements> ... Error installing lilo.conf The message was: test install of lilo failed then up2date hung. I clicked "Cancel" on the up2date UI. Restarting up2date seems to be okay, and gave the message that my system is fully updated. The new kernel is however sucessfully installed (2.4.21-9.0.1). $> rpm -q kernel <snip> kernel-2.4.21-9.0.1.EL $> rpm -q kernel-smp <snip> kernel-smp-2.4.21-9.0.1.EL /etc/lilo.conf seems to be corret also: image=/boot/vmlinuz-2.4.21-9.0.1.ELsmp label=2.4.21-9.0.1.E1 initrd=/boot/initrd-2.4.21-9.0.1.ELsmp.img read-only append="root=LABEL=/" Running /sbin/lilo manually seems to be OK too. I also tried with '/sbin/lilo -t -v' to see the generated output and it seems fine (attached). So I don't know why up2date failed Version-Release number of selected component (if applicable): up2date-4.2.5-1 How reproducible: Didn't try Steps to Reproduce: 1. Use LILO as boot loader & Wait for new kernel fixes in RHN 2. Run 'up2date' 3. Choose the new kernel for update Actual Results: Error as described above. Expected Results: up2date installed the new kernel, put it in lilo.conf, re-install LILO, and quit gracefully after giving me the summary of what packages were updated. Additional info:
Created attachment 98403 [details] Output of up2date
Created attachment 98404 [details] Ouput of 'lilo -t -v'
can I get a copy of the current working lilo.conf? Trying to duplicate this and not having much luck
Created attachment 99400 [details] lilo.conf Here my current lilo.conf. I tweaked this by hand to make the latest kernel default when boot up. (sorry for late reply, been out much). Let me know if I can help w/ anything else.
The latest kernel update failed again. Up2date hung after the following message is displayed on the console. The rest of the symptons are the same as the one I described above. The /etc/lilo.conf content is the same as my previous attachment. The new lilo.conf after running this (failed) up2date is attached. Running /sbin/lilo manually works fine, although the new kernel is not put as the default in lilo.conf by up2date. root]# up2date Traceback (most recent call last): File "/usr/share/rhn/up2date_client/gui.py", line 2069, in doInstallation kernelsToInstall = up2date.installPackages(self.selectedPkgList, self.rpmCallback) File "/usr/share/rhn/up2date_client/up2date.py", line 680, in installPackages runPkgSpecialCases(pkgList) File "/usr/share/rhn/up2date_client/up2date.py", line 729, in runPkgSpecialCases apply(packageProvidesMethods[provideName], [providesPackages[provideName]]) File "/usr/share/rhn/up2date_client/up2date.py", line 825, in installBootLoader __install_lilo(kernelList) File "/usr/share/rhn/up2date_client/up2date.py", line 842, in __install_lilo ret = lilocfg.installNewImages(kernelList,test=0) File "/usr/share/rhn/up2date_client/lilocfg.py", line 384, in installNewImages raise LiloConfError("test install of lilo failed") up2date_client.lilocfg.LiloConfError: Error installing lilo.conf The message was: test install of lilo failed
Created attachment 99950 [details] Latest lilo.conf after running the failed up2date
I am having the same problem. When the up2date kernel install is running, should it be renaming current kernel area to 'label=linux.bak', and new one to 'label=linux'? The kernel does install right, but a reboot won't boot it since the default boot is the one with 'label=linux'. Or, is this the expected results?
I am also having the same problem. up2date installs the new kernel fine. [root@morpheus root]# rpm -qa | grep kernel-2 kernel-2.4.21-4.EL kernel-2.4.21-15.0.3.EL But at the end of up2date, I get this message (there is no debug info): Error installing lilo.conf The message was: test install of lilo failed /etc/lilo.conf is also updated with the new kernel: prompt timeout=50 default=linux boot=/dev/hda map=/boot/map install=/boot/boot.b serial=0 linear image=/boot/vmlinuz-2.4.21-15.0.3.EL label=2.4.21-15.0.3.E initrd=/boot/initrd-2.4.21-15.0.3.EL.img read-only append="console=ttyS0,115200 root=LABEL=/" image=/boot/vmlinuz-2.4.21-4.EL label=linux initrd=/boot/initrd-2.4.21-4.EL.img read-only append="console=ttyS0,115200 root=LABEL=/" The only piece missing is that the "default" entry is not changed, and lilo is not run. I have to do these steps manually on a number of servers. My up2date is updating from a satellite server (not from RHN).
I belive this should be fixed with the 2.1-U5 update and the rhel3.0-U3 updates.
I installed latest up2date with RHEL 3 U3 updates. Then up2date restarted and continued downloading the U3 updates. The same problem happens at the end of the updates process when up2date tried to install LILO. After killing the up2date, I had to manually changed lilo.conf to use the latest kernel as default, then run /sbin/lilo, which worked fine. So it looks like there's still a problem with up2date and LILO.
I am having /have had the same problem for quite awhile. I will not talk about my experience with RH support on this, only that I am now looking for answers in other places. The problem is the same as others describe, run up2date, download and install kernel updates and Up2date appears to hang. Pressing cancel kills up2date, but appears to install the latest kernel. I have modified lilo.conf to run the latest kernel then run /sbin/ lilo which has worked fine in the past.
The lilo configuration seems to be done twice : firstly by the rpm itself and {secon,ba}dly by up2date. Try to change the line noBootLoader=0 to noBootLoader=1 in /etc/sysconfig/rhn/up2date The above kludge avoid up2date to configure lilo and will work until the kernel rpms stop to handle lilo's configuration.