What is smart-check?

After several undetected silent hard drive failures i looked (once again) for a working Nagios/Icinga plugin. There are many, many plugins which claim to be able to detect early signs of failing hard drives. Most of them just don't.

The few which actually do more than just checking for the temperature or some bogus attributes fall short on some other aspects, so i decided to add just another plugin to world which tries to solve this problem.

Basic usage

usage: smart-check [-h] [--disks-file DISKS_FILE] [-s]
                   [--smartctl-path SMARTCTL_PATH] [-a SMARTCTL_ARGS]
                   [-i INTERFACE] [-f FILE] [-x]
                   [--ignore-attributes [IGNORE_ATTRIBUTES [IGNORE_ATTRIBUTES ...]]]
                   [-v] [--debug]
                   [drive]

positional arguments:
  drive                 The device as passed to smartctl's positional argument

optional arguments:
  -h, --help            show this help message and exit
  --disks-file DISKS_FILE
  -s, --sudo            Use sudo to execute smartctl
  --smartctl-path SMARTCTL_PATH
                        Path to smartctl (default: /usr/sbin/smartctl)
  -a SMARTCTL_ARGS, --smartctl-args SMARTCTL_ARGS
                        Other arguments passed to smartctl (default: -n
                        standby)
  -i INTERFACE, --interface INTERFACE
                        The smartctl interface specification (passed to
                        smartctl's -d parameter
  -f FILE, --file FILE  Use S.M.A.R.T. report from file instead of calling
                        smartctl (Use - to read from stdin)
  -x, --exclude-notices
                        Do not report NOTICE warnings (default: False)
  --ignore-attributes [IGNORE_ATTRIBUTES [IGNORE_ATTRIBUTES ...]]
                        Ignore this S.M.A.R.T. attributes (id or name)
  -v, --verbose         Verbose messages
  --debug               Print debug messages

The basic usage is

smart-check <device>

e.g.

root@host# smart-check /dev/sda

If your devices are behind a raid controller, pass smartctl's --device argument via the -i/--interface parameter:

root@host# smart-check -i areca,1 /dev/sg13

Attributes and thresholds

smart-check checks for several well-known S.M.A.R.T. attributes to predict signs a hardware failure. Note: This code is taken from gsmartcontrol. It checks that the following attributes have a non-zero value for VALUE/RAW_VALUE and emits a NOTICE.

• reallocated_sector_count / reallocated_sector_ct
• spin_up_retry_count
• soft_read_error_rate
• temperature_celsius / temperature_celsius_x10
• reallocation_event_count
• current_pending_sector / current_pending_sector_count / total_pending_sectors
• offline_uncorrectable / total_offline_uncorrectable
• ssd_life_left

smart-check also contains a disk database (taken from check_smart_attributes) which can be used to specify more fine-grained thresholds for specific hard drive device models.

If you want to ignore certain attributes (e.g. because your hard drive reports bogus values) pass the --ignore-attributes parameter to smart-check, e.g.:

smart-check --ignore-attributes Temperature_Celsius /dev/sda