Nagios check for SSD Wear Level

This is a small script to check SSD Wear Level Indicators, specifically SMART ID 177 and 233, or the wear-leveling attribute in nvme-cli.

This script tries to find SSDs itself by checking for RAID controller cards we use within our organization. As of this moment that is 3Ware and LSI/AVAGO cards.

It also supports IDE/SATA/NVMe drives.

Example

IDE/SATA:

# check_ssd
SSD OK: (auto) Drive /dev/sda WLC/MWI 100. Drive /dev/sdc WLC/MWI 100.

NVMe:

# check_ssd
SSD OK: (auto) Drive /dev/sda on 0 WLC/MWI 99. Drive /dev/sdc on 0 WLC/MWI 99. Drive /dev/nvme0n1 on 0 WLC/MWI 100. Drive /dev/nvme1n1 on 0 WLC/MWI 100.

3Ware 9750 Controller:

# check_ssd
SSD OK: (3ware) Drive 2/0 WLC/MWI 99. Drive 3/0 WLC/MWI 99.

LSI 9341 Controller:

# check_ssd
SSD OK: (megaraid) Drive 4/1 WLC/MWI 83. Drive 5/1 WLC/MWI 83.

Usage

The script supports the following parameters:

-W=,  --warning=           Set the warning threshold of the check. Default is 40.
-C=,  --critical=          Set the critical threshold of the check. Default is 20.

-c=,  --card=              Instead of autodetecting using lspci, set the card type. We accept "lsi", "3ware" and "auto" for now. Auto is autodetect
-d=,  --device=            The blockdevice we should use for calling smartmontools. Can be any blockdevice in /dev, but is driver specific
-b=,  --brand=             The brand of SSD to search for. We accept "samsung" and "intel"

-d,   --debug              Enable debug output
-t,   --test               Only test if there are SSDs present in the system. exits with 0 when found, 1 if not
-s=,  --storcli=           Add a custom path for storcli binary.
-n,   --nossd              Not having SSD is OK.
-h,   --help               Show what you are reading now!
-v,   --version            Show version

Requirements

The script uses the following tools:

• smartctl for retrieving SMART data
• storcli for managing LSI controllers. Vendor specific tools might be required, for example for Dell PERC.
• tw_cli for managing 3Ware controllers
• nvme-cli for managing nvme disks
• bc for calculations
• awk for string manipulation
• head/tail for output filtering
• sed for string manipulation
• tr for removing characters
• lsblk to identify SSDs on HBAs (non-RAID)
• lsscsi to lookup blockdevice by controller

Example

In our case we call the script after the RAID status check completes with exit code zero, and explicitly set the card type within the script. e.g.:

in check_lsi_raid:

sub getSSDstatus {
    system("/usr/bin/check_ssd -t -c=lsi");
    if ($? == 0) {
      # We have SSDs!
      my $command = "/usr/bin/check_ssd -c=lsi";
      my @output = `$command`;
      printf (@output);
      return $? >> 8;
    } else {
      return 0;
    }
}

...

if($exitstatus == 0) {
            print "LSIRAID OK (Ctrl #$controller) | STATUS=$exitstatus\n";
            my $exitcode = getSSDstatus();
            exit($exitcode);

ChangeLog:

• Version 1.6: Add performance data, fix issues with non standar lsi cards, several fixes and extra options. Check changelog
• Version 1.5: Tiny bugfix
• Version 1.4: Support for NVMe disks.
• Version 1.3: Replace lsblk -S with lsscsi. -S flag on lsblk is not available on Wheezy
• Version 1.2: Correctly identify which blockdevice (/dev/sdX) belongs to which controller
• Version 1.1: Support for multiple LSI cards. Support for multiple 3Ware cards are not yet implemented.