With my new box I decided to enable the smart daemon that is packaged with smartmontools. I quickly noticed error messages regarding my primary IDE drive. Allegedly I had at least one bad sector.
Running fsck at boot cured nothing and manually testing with smartctl always passed.
I next read the
Bad block HOWTO for smartmontools and learned about the debugfs tool.
When I calculated the location of the bad block and queried with debugfs, I received a "file not found" message. I scratched
my head for several moments and then realized that no files were stored on that block. So what to do to eliminate the smartd error messages?
Well, a little further into the HOWTO was the hat trick. I could have copied the /dev/zero device to the bad block, but the solution was much simpler. I copied a DVD ISO image, which was about the same size as the remaining space in the partition (7.4 GB). The hard drive firmware remedied the problem nicely. I don't know whether the block remains bad or whether the firmware marked the block as unusable, but I think the latter as the error messages changed thereafter.
After I restarted the smartd I no longer received any error messages. No errors from the smartctl long test either. Very nice trick.
Through this journey this afternoon, I also noticed that the version of smartmontools packaged with Slackware 12.0 (and 12.1) did not support my new WD3200AAKS SE16 Caviar SATA hard drive. I could use the
-d ata option to work with the drive, but I preferred otherwise.
I downloaded the 5.38 version, compiled with the stock Slackware build script (updated for the new version), and then installed version 5.38. The new database recognized my new drive. I emailed PV and asked him to consider updating smartmontools in 12.1.
I also modified /etc/syslog.conf to create a separate log for smartd:
# Route smartd messages
local3.* /var/log/smartd
I then modified the /etc/rc.d/rc.smartd:
SMARTD_OPTIONS="-l local3"
I like separate logs as much as possible and this trick was nice.
One last observation, not directly related to smartmontools. When I was trying to troubleshoot the smartmontools error messages, I naturally ran a full fsck at boot (through
touch /etc/forcefsck). The console screen was mildly discomforting to watch. With two hard drives, the primary an IDE and the secondary the new SATA, fsck seemed confused about displaying the progress bar results, or perhaps the two drives were fighting for time, or both. In /etc/rc.d/rc.S, in the section for checking all non-root partitions with fsck, I added the
-s option. That solved the problem. Perhaps the -s option is slower, but I could not really tell. Perhaps later I'll time the results with a stop watch, but the on screen console display behaved thereafter and provided me some peace of mind.
Just some tid bits I thought somebody might find useful. With my schedule these days I don't get much chance to participate in the forum as I would like
and hope this info helps.
