Red Hat Bugzilla – Bug 27218
mkinitrd error, All of your loopback devices are in use!
Last modified: 2005-10-31 17:00:50 EST
From Bugzilla Helper: User-Agent: Mozilla/4.76 [en] (Win98; U) I used the Redhat Network to download the new kernel rpms (2.2.17-14smp) and installed them. I then rebooted and discovered that it was still using the old kernel 2.2.16-22smp. The ethernet and sound modules did not load properly either. I was surprised to find out that the lilo.conf wasn't appended with the new kernel information and that the initrd.img files were not created either. After realizing this I attempted to create the initrd files and encountered the following error from mkinitrd: "All of your loopback devices are in use!" I can't seem to get mkinitrd to work at all now. System configuration: Celeron 500 x2, on Abit BP6 motherboard, 128MB RAM, Adaptec 2940U2W SCSI controller, IBM SCSI 9GB hard drive. Reproducible: Always Steps to Reproduce: 1. Use Redhat Update Agent to download and install kernel-2.2.17-14 ( and associated files) 2. Reboot machine after installation completes. 3. attempt to run "mkinitrd -v 2.2.17-14smp.img 2.2.17-14smp" to create the appropriate .img files Actual Results: mkinitrd fails with the following error: All of your loopback devices are in use! Expected Results: 2.2.17-14smp.img is created
It upgraded your kernel, so 2.2.16-22 isn't there anymore. Since it isn't there is no loop module to insert, so you can't make an initrd.
I understand that. How do I go about correcting this? That was my main concern.
If you have an install CD, you can boot off of that in 'rescue' mode, and then make an initrd.
I attempted to create initrd after booting into rescue mode from CD and mounting /boot. However, mkinitrd does not appear to exist in /sbin folder used by rescue mode. I attempted to mount the system's root drive under /system. I then attempted to use /system/sbin/mkinitrd but received the error: "No such file or directory" How am I supposed to create the initrd in this situation?
It may be complaining that it doesnt know where your /etc/fstab is. It uses /etc/fstab as the default if one is not given. Try adding the extra argument --fstab=/system/etc/fstab and see if that helps.
I am wondering, could the kernel RPM be told to 'modprobe loop' just before it gets upgraded? As in the unfortunate case where the sys-admin forgets the special case of -i for kernel upgrades, this would at least allow the generation of initrds. Perhaps this could also be automated. I understand the reluctance to take the next logical step and actualy install the kernel in lilo, but at least getting the sysadmin that far would make things *very* convenient.
I've just had the same problem, i had it because i wasn't root on the file system, and only root is able to handle that. (of course, you need to have support for loop devices) That errormsg is a little confusing, that's all
error message added for case where user isnt' logged in as root
Another similar case is where you (not RH default, but a scheme I like) mount /tmp as tmpfs. Becouse you can't do loopback on tmpfs (I assume becouse loop requires block backing) it all fails rather miserably. A better error message could be useful - but this one might be a little harder to detect.