From Bugzilla Helper: User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.4.3) Gecko/20041004 Description of problem: Installer always crashes at the exact same point. It does not matter if I use CDs, NFS or hard drive image. I am trying to install on an empty /dev/hda drive. I have a redhat 9 setup already on a software RAID on /dev/hde and /dev/hdg that I do not want to disturb. Here is the error (I will attach the dump file also): Traceback (most recent call last): File "/usr/bin/anaconda", line 1165, in ? intf.run(id, dispatch, configFileData) File "/usr/src/build/475969-i386/install//usr/lib/anaconda/text.py", line 507, in run dispatch.gotoNext() File "/usr/src/build/475969-i386/install//usr/lib/anaconda/dispatch.py", line 171, in gotoNext self.moveStep() File "/usr/src/build/475969-i386/install//usr/lib/anaconda/dispatch.py", line 239, in moveStep rc = apply(func, self.bindArgs(args)) File "/usr/src/build/475969-i386/install//usr/lib/anaconda/packages.py", line 557, in turnOnFilesystems thefsset.makeFilesystems (instPath) File "/usr/src/build/475969-i386/install//usr/lib/anaconda/fsset.py", line 1479, in makeFilesystems self.labelEntry(entry, chroot) File "/usr/src/build/475969-i386/install//usr/lib/anaconda/fsset.py", line 1369, in labelEntry entry.fsystem.labelDevice(entry, chroot) File "/usr/src/build/475969-i386/install//usr/lib/anaconda/fsset.py", line 517, in labelDevice label = labelFactory.createLabel(entry.mountpoint, self.maxLabelChars) File "/usr/src/build/475969-i386/install//usr/lib/anaconda/fsset.py", line 112, in createLabel labels = diskset.getLabels() File "/usr/src/build/475969-i386/install//usr/lib/anaconda/partedUtils.py", line 568, in getLabels label = isys.readFSLabel(dev) File "/usr/lib/anaconda/isys.py", line 639, in readFSLabel label = readJFSLabel(device, makeDevNode) File "/usr/lib/anaconda/isys.py", line 594, in readJFSLabel label = readJFSLabel_int("/tmp/disk") File "/usr/lib/anaconda/isys.py", line 582, in readJFSLabel_int os.lseek(fd,32768,0) OSError: [Errno 22] Invalid argument Local variables in innermost frame: device: /tmp/disk jfslabel: None fd: 43 Version-Release number of selected component (if applicable): How reproducible: Always Steps to Reproduce: 1.Install from CD images. 2.Format partitions as ext3/LVM... does not matter 3.Choose any or all packages ... does not matter 4.Accept defaults until it crashes. Actual Results: Crashes every time at the exact same place with the same error dump. Expected Results: Successful install Additional info: I have downloaded the install image 3 times. Each time the MD5SUMs have matched. I ran the media check on the CDs and they were good. The NFS and harddrive installs gave the exact same results as the CDs. Package selection and partition type do not make a difference.
Created attachment 106057 [details] Anaconda Dump file This is as default as I could make the install with 'everything' selected for packages.
If I add 'hde=none hdg=none' to the kernel options it installs. Something in Anaconda must be trying to mount and use the /dev/md* arrays even though I have not selected them for installation.
Even though you're not using them for installation, we have to try to look at the filesystem label on there to avoid duplicate labels. I've added some code to do better error handling on the reads, but I'm curious as to why it's failing. Is there anything odd about the md arrays?
Maybe its because there are no labels? I use fixed devices instead of labels so I do not run into problems on dual booting linux installs. Well not until recently. Here is what I have: # cat /proc/mdstat Personalities : [linear] [raid1] read_ahead 1024 sectors md6 : active raid1 hde9[0] hdg9[1] 16659136 blocks [2/2] [UU] md5 : active raid1 hde7[0] hdg7[1] 2000256 blocks [2/2] [UU] md4 : active raid1 hde6[0] hdg6[1] 30000512 blocks [2/2] [UU] md3 : active raid1 hde5[0] hdg5[1] 25000320 blocks [2/2] [UU] md2 : active raid1 hde3[0] hdg3[1] 1000320 blocks [2/2] [UU] md1 : active linear hde2[0] hdg2[1] 2000640 blocks 32k rounding md0 : active raid1 hde1[0] hdg1[1] 488768 blocks [2/2] [UU] unused devices: <none> # for i in 0 1 2 3 4 5 6 ; do echo /dev/md${i}: \"`e2label dev/md$i`\" ; done /dev/md0: "" /dev/md1: "" /dev/md2: "" /dev/md3: "" /dev/md4: "" /dev/md5: "" /dev/md6: ""
Hmm, maybe. Although it still shouldn't traceback on the reads. Oh well, should be fixed in CVS.
*** Bug 139274 has been marked as a duplicate of this bug. ***
*** Bug 141957 has been marked as a duplicate of this bug. ***
*** Bug 142610 has been marked as a duplicate of this bug. ***
Created attachment 109155 [details] Anaconda exception dump (text mode) Trying to install Fedora Core 3 onto i686 system with SCSI hard disks. Software RAID-0 used as root "/" across seven SCSI hard disks (34GB total capacity). Note: "/boot" partition is on 170MB IDE hard disk; Boot Loader installed to MBR of same IDE hard disk.) No Operating Systems exist on any hard disk in the system. Looks like anaconda is trying to create the Software RAID device /dev/md0 when exception is thrown. Using Adaptec AHA-2940U/2940UW PCI SCSI controller.