[smartmontools-support] smartctl with VMware server raw disks

[Eric S. <ej...@sh...>]

I have a few VMware Server v2 hosts running on CentOS 5.5. I've been 
using smartctl to monitor the drives of a raid-1 array, which contains 
the OS, VMware Server, and all VMs. So far so good.

These servers also have 2 additional drives, accessed by a VM guest in 
raw mode. FWIW, the guest machines are unable to run smartctl against 
the raw drives:
# smartctl -a /dev/sdb
smartctl version 5.38 [i686-redhat-linux-gnu] Copyright (C) 2002-8 Bruce 
Allen
Home page is http://smartmontools.sourceforge.net/

Device: VMware,  VMware Virtual S Version: 1.0
Device type: disk
Local Time is: Wed Jan 19 19:08:41 2011 MST
Device does not support SMART

Error Counter logging not supported
Device does not support Self Test logging
</end>
It would be nice if this worked, but I figure I can live w/out smart 
data from those drives, at least until VMware Server more fully supports 
raw drives.

My problem is this. I'm running smartctl on the host machine, on drives 
/dev/sda and /dev/sdb. I wasn't paying very close attention and failed 
to realize that /dev/sdb is one of the raw drives used by a guest VM. 
/dev/sda and /dev/sdc are the raid-1 mirrors, and /dev/sdb and /dev/sdd 
are the raw disks for the VM guest. I'm a little surprised, but there 
doesn't appear to be a problem with the VM host running smartctl on a 
drive that's being accessed as a raw disk by one of the guest VMs. I was 
a little leery to try this, it's been running daily for 6 weeks now, 
with no problem other than this from smartctl:
  /dev/sdb :
     Usage: Seek_Error_Rate (7) changed to
       100, 200,
     Usage: Temperature_Celsius (194) changed to
       108, 107, 108, 107, 106, 107, 108, 107,
  Currently unreadable (pending) sectors detected:
  	/dev/sdb - 48 Time(s)
  	24 unreadable sectors detected

Q#1: Are the seek errors and unreadable sectors possibly the result of 
running smartctl on the host while a VM guest is accessing the drive in 
raw mode? Or are these errors just a drive that's a little faulty? (I 
have a warranty replacement drive ready to install, but haven't done so yet)

Q#2: Does anyone have an opinion as to whether running smartd against 
drives that are used in raw mode by virtual guests would be a problem 
for some reason? FWIW, these guest drives are part of a raid-1 mirror, 
which is controlled by the guest VM.

Any insight here is appreciated. I'm not very knowledgeable regarding 
smart, so I'm relying on the expertise here.

TIA.
-- 
-Eric 'shubes'