What is smartd?

[Matt Morgan]

On 03/24/2004 10:27 PM, Wolfgang Gill wrote:

>On Wed, 24 Mar 2004 23:49:41 +0530, Kaustubh Ghosh wrote
>  
>
>>I have a peculiar situation:
>>When my FC1 boots up,it gives a message:
>>Starting smartd:                        [FAILED]
>>When it shuts down ,it gives message:
>>Stopping smartd:                 [FAILED]
>>Everything ,the desktop etc. works fine -there is no problem.But 
>>what is 'smartd'  and why does it fail? Thanks in advance.
>>
>You need to enable S.M.A.R.T in you systems BIOS (It's disabled by default).
>Else you will get this error. ALL new drives these days have S.M.A.R.T
>technology built-in. The HDD's hardware simply monitors the HDD and when there
>are problems, it will inform either the BIOS or OS's S.M.A.R.T monitoring
>software of a possible HDD failure.
>  
>
This is getting interesting. I wondered, when there's a problem, how 
does smartd let you (user or root) know that there's a problem? It's all 
there in the man page. It checks every 30 minutes (configurable) and 
uses SYSLOG, so the errors probably end up in /var/log/messages on the 
local computer, unless you've set up a syslog server. Apparently at 
startup it can get a lot of "missing block-major-xx" errors, which are 
"mostly harmless," if it's been configured to look for non-existent 
drives especially.

I was pleased to see that even though I've put zero effort into smartd, 
it appears to be working perfectly on my computer (these are the smartd 
startup messages in my /var/log/messages) :

Mar 24 09:20:58 morgan4-f smartd[3246]: smartd version 5.21 Copyright 
(C) 2002-3 Bruce Allen
Mar 24 09:20:58 morgan4-f smartd[3246]: Home page is 
http://smartmontools.sourceforge.net/
Mar 24 09:20:58 morgan4-f smartd[3246]: Opened configuration file 
/etc/smartd.conf
Mar 24 09:20:58 morgan4-f smartd[3246]: Configuration file 
/etc/smartd.conf parsed.
Mar 24 09:20:58 morgan4-f smartd[3246]: Device: /dev/hda, opened
Mar 24 09:20:58 morgan4-f smartd[3246]: Device: /dev/hda, not found in 
smartd database.
Mar 24 09:20:59 morgan4-f smartd[3246]: Device: /dev/hda, is SMART 
capable. Adding to "monitor" list.
Mar 24 09:20:59 morgan4-f smartd[3246]: Monitoring 1 ATA and 0 SCSI devices
Mar 24 09:20:59 morgan4-f smartd[3248]: smartd has fork()ed into 
background mode. New PID=3248.
Mar 24 09:20:59 morgan4-f smartd: smartd startup succeeded

(I rebooted yesterday morning). There are no other smartd messages since 
then, which is good news!

Here's what I still wonder: how well does SMART work? If I aggressively 
monitor my smartd errors, will I be guaranteed to catch a failing drive 
before it dies completely? Or when they die, do they mostly die all at 
once, not giving smartd a chance to warn you ahead of time? And are 
there any UI elements to smartd, so that you'll be warned when a problem 
is logged, without having to go check /var/log/messages, or are you 
expected to use whatever other log monitoring tools you've already set up?

I also think it's kind of funny that this is out there now, now that 
drives hardly fail anymore. I wish I had this 10 years ago. Although I 
still see situations where certain groups of drives are bad (for 
example, a bunch we bought for some Macs in our Design department last 
year all failed). Pretty neat.

--Matt