From Bugzilla Helper: User-Agent: Mozilla/4.0 (compatible; MSIE 5.01; Windows 98) Description of problem: If you ever try to umount a reiserfs filesystem the command line will never return. Happens when shutting down machine too, it just hangs at: Unmounting Filesystems Have to use the Reset/Power Button to get any further. Which of course causes FSCK to kick off everytime it starts up. How reproducible: Always Steps to Reproduce: 1. Mount a reiserfs filesystem at first or second+ level (/var = 1st level, /var/spool/squid = second+). Have it mounted at bootup or not makes no difference just make sure it is mounted. 2. Umount a reiserfs filesystem or restart machine 3. Wait for a LOOOOOOOONG time as it fails completely Actual Results: 1. Hit reset/power button, cause your boss is breathing down your neck, after waiting 1+ hours for it to return. 2. Cycle box, wait for fsck to fix everything (you hope) 3. Announce cache and webserver is back and running. After first announcing it should have been a 5 minute item. Expected Results: 1. Issue command "shutdown -r now" 2. Wait a few moments for box to restart 3. Change Helpdesk announcement to "no current problems". Additional info: Known issue with the 2.4.3 kernel. Goto http://www.namesys.com/download.html and read the bit about "Warning: VFS...etc"
Please try the 2.4.3-15 kernel at http://people.redhat.com/arjanv/testkernels/ it should have the fix for this.
I installed kernel-2.4.3-15... No luck, still hangs. Currently made a new kernel(bzImage) with reiserfs in it (not a module), from source (2.4.7 from kernel.org), using config "kernel-2.4.3-i686.config" from 2.4.3-12 source. With the one change (a biggy I know) it now, works. It was the only way I can get it to NOT hang on umount, so far. This does not mean 2.4.7 is going to stay there, it is not default 2.4.3-15 is. Actual machine is: http://web.grcc.cc.mi.us
Are you _very_ sure you were running -15 with the hang? The "common" hang on unmount was fixed in that, and several people reported success with it so far.
You are correct, in assuming -12 was running. I archive configs as I go, yep My label said -15 but was running -12 (darn vi with yank and put, and me forgetting to edit the image line along with the label). My bad. When actually running the -15 Kernel, it does fix the problem. Although what intrigues me, is the 2.4.7 did the same as a module, but not as a compiled in component. Guess this is closed then.
*** Bug 50576 has been marked as a duplicate of this bug. ***