I'm looking for a way to get a bash script to determine the sdX device name of the cache drive.  I've got usable results for the other drives using mdcmd status, but the output from that doesn't include the cache drive.  And since some of my device names change between reboots, I don't think I can just hard code the values.

My goal is to setup a cron job to run scheduled SMART tests on each drive once every month, with one drive being tested per day.  This was recently discussed in a couple threads, but no final script was ever posted.  I'm comfortable enough with bash to write this on my own, I just need some way to programmatically return the cache drive's sdX name.

 

Thanks in advance!

As a crutch, you could try grepping through the syslog to see which devices are mounted for cache.

Clunky yes, but how about

 

df /mnt/cache | tail -1 | cut --delimiter=' ' --fields=1

 

I don't have a cache drive on any of my servers, but when I run this for disk1 I get

 

/dev/md1

 

 

Clunky yes, but how about

 

df /mnt/cache | tail -1 | cut --delimiter=' ' --fields=1

 

I don't have a cache drive on any of my servers, but when I run this for disk1 I get

 

/dev/md1

I can confirm this works fine.  For instance on my system I get /dev/sdb1.  It was not clear if the OP wanted the partition number to be provided or not - it would be easy to chop it off if it is not wanted.

Thanks, BRiT.  That actually works out pretty well.

bobkart's answer works also, but since I want to feed the output into smartctl, I'd need to strip the partition number.  Syslog also provides a disk number in line with the rest of the array.

 

Here's what I'm testing now, in case anybody wants to accomplish something similar.  This bit goes at the end of /boot/config/go:

#Get device names for array drives
mdcmd status | gawk 'match($0,/rdevName.(.+)=(...)/,arr) {num=sprintf("%02d",arr[1]); print num, arr[2]}' > /tmp/SMART_Device_List
#Get cache drive device name
gawk 'match($0,/import.(..).cache.*(...)$/,arr) {print arr[1], arr[2]}' /var/log/syslog | head -1 >> /tmp/SMART_Device_List
#Add custom cron entries
crontab -l > /tmp/file
cat /boot/scripts/Custom_Crontab >> /tmp/file
crontab /tmp/file
rm -f /tmp/file

I don't know if gawk is included in unRaid versions prior to 6 -- I didn't really do anything like this on version 5.

Edit:  Just noticed an issue with the timing on this.  The device list was empty after rebooting.  I'm guessing the array needs to be started before any of the drive info is available.    I'll update later with a better solution.

 

SMART_Device_List then contains one line for each drive with the disk number (parity is 0, cache is 24) and device name:

00 sdc
01 sdd
...
21 sdv
24 sdb

 

crontab:

0 5 3-27 * * /boot/scripts/Scheduled_SMART_Test 1> /dev/null

 

Scheduled_SMART_Test:

#!/bin/bash

drivenum=$(printf "%02d" $((`date +%d` - 3)))
devname=`grep $drivenum /tmp/SMART_Device_List | awk '{print $2}'`

if [ $devname ]; then
usr/local/emhttp/plugins/dynamix/scripts/spindowndelay $devname 0
smartctl -t long /dev/$devname
while true; do
	sleep 30
	if ! smartctl -c /dev/$devname | grep -Pom1 '\d+%'; then
		/usr/local/emhttp/plugins/dynamix/scripts/spindowndelay $devname
		break
	fi
done
fi

Spindowndelay included in Dynamix per this post.  The grep statement hangs if $drivenum is negative, which is why the crontab entry starts on the 3rd of each month.  Another if statement to check if $drivenum is greater than or equal to 0 would accomplish the same thing.