Red Hat Bugzilla – Bug 812155
filesystem check failure in custom partition setup
Last modified: 2012-04-15 13:43:51 EDT
Description of problem:
Anaconda reports a Unrecoverable Error at the end of the Fedora 17 installation: "ext4 filesystem check failure on /dev/sda9: Filesystem error left uncorrected [...]"
Steps to Reproduce:
1. in the anaconda installer choose "Create Custom Layout"
2. choose a root partition with type ext4 and format it, (I created also a 2 GB swap, and a encrypted home partition with filesystem type xfs)
3. finish the installation
at the end of the procedure "Copying live image to hard drive", it will very likely report this unrecoverable error. After clicking on OK it will reboot the computer automatically. I noticed also that it always reboots when the installer is closed (probably this bug is already reported or known).
- used Fedora 17 pre-release: Beta RC4
Please post anaconda logs. Suspect dupe of 804779.
Created attachment 577313 [details]
ok I attempted it again, without xfs partition and the same error happened again.
I post the logs: anaconda.log, messages, program.log, storage.log
Created attachment 577314 [details]
Created attachment 577315 [details]
Created attachment 577316 [details]
How was the install media (USB stick) created? dd, livecd-tools (version?), or the livecd-iso-to-disk that's on the RC4 LiveCD?
Maybe someone else can sort this out better, but I could use more detail on exactly what the custom partition layout looked like. I'm seeing ntfsresize being called. A 14G /boot on sda4? No sda6 but an sda7.
When sda4 is formatted ext4, the format fails with an error. A subsequent e2fsck on sda4 then also fails. This is not an encrypted partition. When reattempting to install, dumpe2fs also finds errors.
I used livecd-iso-to-disk to create the install media.
the partition table output via parted:
Model: ATA SAMSUNG HM250JI (scsi)
Disk /dev/sda: 250GB
Sector size (logical/physical): 512B/512B
Partition Table: gpt
Disk Flags: pmbr_boot
Number Start End Size File system Name Flags
1 1049kB 32.1GB 32.1GB ntfs hidden, msftres, legacy_boot
2 32.1GB 59.6GB 27.5GB ntfs
3 59.6GB 59.6GB 1049kB bios_grub
4 59.6GB 75.0GB 15.4GB ext4
5 75.0GB 123GB 48.5GB
7 123GB 126GB 2147MB linux-swap(v1)
I created the ntfs partitions via gnome-disk-utilty under the F17 live system.
I'm virtually certain this is not an anaconda bug. I can't reproduce this, even with a 15G /boot (which really makes no sense, but it works for me).
It's unclear why your anaconda log says you are starting out with 6 partitions when based on what you've written, you should have had two NTFS partitions and the rest free space. And it's not clear why you're missing sda6, so you must've deleted a partition at some point in anaconda.
An exact step by step recipe of how you are reproducing this might be helpful because so far I can't reproduce it.
ok I will post a more accurate description later.
this 15 GB partition is mounted as / not /boot
You know what, before you go to the trouble, could you boot the LiveCD, and run
mke2fs -t ext4 /dev/sda4
program.log indicates this fails, and it's the subsequent 'e2fsck -f -p -C 0 /dev/sda4' failure that causes the error reported by anaconda.
ok after I went through the installation the command "e2fsck -f -p -C 0
/dev/sda4" fails indeed. But the command works if I format the partitions fresh in anaconda, after the partitions are formatted and before the copying to hard disk begun this command still worked. But at the and of the copy process it shows this unrecoverable error. And then also the command above show this output:
"_Fedora-17-Beta-: Superblock last mount time (Tue Apr 10 01:06:12 2012,
now = Mon Feb 20 14:04:50 2012) is in the future.
_Fedora-17-Beta-: UNEXPECTED INCONSISTENCY; RUN fsck MANUALLY.
(i.e., without -a or -p options)"
oh damn. I didn't know that the time of the notebook is so wrong, I don't use it often. Probably that's the reason, I will try if I can reproduce the error even if the time is correct.
It makes sense that e2fsck is going to be annoyed and fail, because the mke2fs command failed. The create file system cause is the one I'm curious about, why it's failing.
Sounds something like bug 522969. cc'ing Eric Sandeen. Still not sure this is anaconda.
yes with the corrected notebook time it installs successfully.
Thanks for your help.
Wow, not a graceful failure from an incorrectly set date! Not graceful at all. I'm able to reproduce this as well. Wonky. And certainly not an even remotely helpful error message at the end of the installation, to die the problem to date. This problem prevents the resizing of the file system, and the resulting installation is useless/not bootable.
*** This bug has been marked as a duplicate of bug 811706 ***