Article ID: 113107, created on Jan 10, 2012, last review on Jun 17, 2016

  • Applies to:
  • Virtuozzo containers for Linux 4.7


It is impossible to boot into the PVCfL 4.7 kernel on CentOS 5.x after upgrading to PVCfL if the root is placed on an SCSI device.

A similar output may be observed on the server console:

switchroot mount failed: No such file or directory.
3.668771] Kernel panic e not syncing. Attempted to kill init!
3.668887] Pid: 1, comm: init veid2 0 Not tainted 2.6.32—042stab044.11 #1
3.669000] Call Trace2
3.669111] [<ffffffff814ca02c>] ? panic+0x78/0x143
3.669228] [<ffffffff8106f229>] ? do_exit+0x909/0x910
3.669401] [<ffffffff8106f288>] ? do_group_exit+0x58/0xd0
3.669515] [<ffffffff8106f317>] ? sys_exit_group+0x1?/0x20
3.669632] [<ffffffff8100b242>] ? system_call_fastpath+0x16/0x1b


The issue is with the CentOS/RedHat initrd generation mechanism, which omits the necessary delay for initialization of SCSI devices if the "scsi_mod" module is not used explicitly.

It was reported to CentOS a long ago ( ), but unfortunately marked as 'wontfix'. However, in 2.6.32 kernel branch the functionality of this module is included in the kernel.


  1. Edit /sbin/mkinitrd as follows:

    [root@pvclin46 ~]# cp /sbin/mkinitrd{,.orig}
    [root@pvclin46 ~]# vi /sbin/mkinitrd
    [root@pvclin46 ~]# diff -pruN /sbin/mkinitrd{.orig,}
    --- /sbin/mkinitrd.orig 2012-01-09 18:14:55.000000000 +0700
    +++ /sbin/mkinitrd      2012-01-09 18:17:19.000000000 +0700
    @@ -923,7 +923,7 @@ EOF
        unset usb_mounted
    -    if [ -n "$scsi" ]; then
    +    if [ -n "$scsi" -o -n "$scsimodules" ]; then
            emit "echo Waiting for driver initialization."
            emit "stabilized --hash --interval 1000 /proc/scsi/scsi"

    Or, in short, on a line 923 locate

    if [ -n "$scsi" ]; then

    and replace it with

    if [ -n "$scsi" -o -n "$scsimodules" ]; then.
  2. Re-generate the initrd image for the PVCfL 4.7 kernel.

    The easiest way to do this is to re-install the appropriate vzkernel RPM package.

NOTE there is a similar problem observed on nodes with RAID, it is described in the following article:

122340 - Kernel panic after upgrading to PVC 4.7 on CentOS 5 host with software RAID

Search Words


0c05f0c76fec3dd785e9feafce1099a9 d02f9caf3e11b191a38179103495106f e8e50b42231236b82df27684e7ec0beb 2897d76d56d2010f4e3a28f864d69223

Email subscription for changes to this article
Save as PDF