1#!/bin/sh 2# 3# Show SMART stats 4# 5 6helpstr=" 7smart: Show SMART temperature and error stats (specific to drive type) 8smartx: Show SMART extended drive stats (specific to drive type). 9temp: Show SMART drive temperature in celsius (all drives). 10health: Show reported SMART status (all drives). 11r_proc: Show SMART read GBytes processed over drive lifetime (SAS). 12w_proc: Show SMART write GBytes processed over drive lifetime (SAS). 13r_ucor: Show SMART read uncorrectable errors (SAS). 14w_ucor: Show SMART write uncorrectable errors (SAS). 15nonmed: Show SMART non-medium errors (SAS). 16defect: Show SMART grown defect list (SAS). 17hours_on: Show number of hours drive powered on (all drives). 18realloc: Show SMART reallocated sectors count (ATA). 19rep_ucor: Show SMART reported uncorrectable count (ATA). 20cmd_to: Show SMART command timeout count (ATA). 21pend_sec: Show SMART current pending sector count (ATA). 22off_ucor: Show SMART offline uncorrectable errors (ATA). 23ata_err: Show SMART ATA errors (ATA). 24pwr_cyc: Show SMART power cycle count (ATA). 25serial: Show disk serial number. 26nvme_err: Show SMART NVMe errors (NVMe). 27smart_test: Show SMART self-test results summary. 28test_type: Show SMART self-test type (short, long... ). 29test_status: Show SMART self-test status. 30test_progress: Show SMART self-test percentage done. 31test_ended: Show when the last SMART self-test ended (if supported). 32" 33 34# Hack for developer testing 35# 36# If you set $samples to a directory containing smartctl output text files, 37# we will use them instead of running smartctl on the vdevs. This can be 38# useful if you want to test a bunch of different smartctl outputs. Also, if 39# $samples is set, and additional 'file' column is added to the zpool output 40# showing the filename. 41samples= 42 43# get_filename_from_dir DIR 44# 45# Look in directory DIR and return a filename from it. The filename returned 46# is chosen quasi-sequentially (based off our PID). This allows us to return 47# a different filename every time this script is invoked (which we do for each 48# vdev), without having to maintain state. 49get_filename_from_dir() 50{ 51 dir=$1 52 pid="$$" 53 num_files=$(find "$dir" -maxdepth 1 -type f | wc -l) 54 mod=$((pid % num_files)) 55 i=0 56 find "$dir" -type f -printf "%f\n" | while read -r file ; do 57 if [ "$mod" = "$i" ] ; then 58 echo "$file" 59 break 60 fi 61 i=$((i+1)) 62 done 63} 64 65script=$(basename "$0") 66 67if [ "$1" = "-h" ] ; then 68 echo "$helpstr" | grep "$script:" | tr -s '\t' | cut -f 2- 69 exit 70fi 71 72smartctl_path=$(command -v smartctl) 73 74# shellcheck disable=SC2015 75if [ -b "$VDEV_UPATH" ] && [ -x "$smartctl_path" ] || [ -n "$samples" ] ; then 76 if [ -n "$samples" ] ; then 77 # cat a smartctl output text file instead of running smartctl 78 # on a vdev (only used for developer testing). 79 file=$(get_filename_from_dir "$samples") 80 echo "file=$file" 81 raw_out=$(cat "$samples/$file") 82 else 83 raw_out=$(eval "sudo $smartctl_path -a $VDEV_UPATH") 84 fi 85 86 # What kind of drive are we? Look for the right line in smartctl: 87 # 88 # SAS: 89 # Transport protocol: SAS 90 # 91 # SATA: 92 # ATA Version is: 8 93 # 94 # NVMe: 95 # SMART/Health Information (NVMe Log 0xnn, NSID 0xnn) 96 # 97 out=$(echo "$raw_out" | awk ' 98# SAS specific 99/read:/{print "rrd="$4"\nr_cor="$5"\nr_proc="$7"\nr_ucor="$8} 100/write:/{print "rwr="$4"\nw_cor="$5"\nw_proc="$7"\nw_ucor="$8} 101/Non-medium error count/{print "nonmed="$4} 102/Elements in grown defect list/{print "defect="$6} 103 104# SAS common 105/SAS/{type="sas"} 106/Drive Temperature:/{print "temp="$4} 107# Status can be a long string, substitute spaces for '_' 108/SMART Health Status:/{printf "health="; for(i=4;i<=NF-1;i++){printf "%s_", $i}; printf "%s\n", $i} 109/number of hours powered up/{print "hours_on="$7; hours_on=int($7)} 110/Serial number:/{print "serial="$3} 111 112# SATA specific 113/Reallocated_Sector_Ct/{print "realloc="$10} 114/Reported_Uncorrect/{print "rep_ucor="$10} 115/Command_Timeout/{print "cmd_to="$10} 116/Current_Pending_Sector/{print "pend_sec="$10} 117/Offline_Uncorrectable/{print "off_ucor="$10} 118/ATA Error Count:/{print "ata_err="$4} 119/Power_Cycle_Count/{print "pwr_cyc="$10} 120 121# SATA common 122/SATA/{type="sata"} 123/Temperature_Celsius/{print "temp="$10} 124/Airflow_Temperature_Cel/{print "temp="$10} 125/Current Temperature:/{print "temp="$3} 126/SMART overall-health self-assessment test result:/{print "health="$6} 127/Power_On_Hours/{print "hours_on="$10; hours_on=int($10)} 128/Serial Number:/{print "serial="$3} 129 130# NVMe common 131/NVMe/{type="nvme"} 132/Temperature:/{print "temp="$2} 133/SMART overall-health self-assessment test result:/{print "health="$6} 134/Power On Hours:/{gsub("[^0-9]","",$4); print "hours_on="$4} 135/Serial Number:/{print "serial="$3} 136/Power Cycles:/{print "pwr_cyc="$3} 137 138# NVMe specific 139/Media and Data Integrity Errors:/{print "nvme_err="$6} 140 141# SMART self-test info 142/Self-test execution status:/{progress=tolower($4)} # SAS 143/SMART Self-test log/{test_seen=1} # SAS 144/SMART Extended Self-test Log/{test_seen=1} # SATA 145/# 1/{ 146 test_type=tolower($3"_"$4); 147 # Status could be one word ("Completed") or multiple ("Completed: read 148 # failure"). Look for the ":" to see if we need to grab more words. 149 150 if ($5 ~ ":") 151 status=tolower($5""$6"_"$7) 152 else 153 status=tolower($5) 154 if (status=="self") 155 status="running"; 156 157 if (type == "sas") { 158 hours=int($(NF-4)) 159 } else { 160 hours=int($(NF-1)) 161 # SATA reports percent remaining, rather than percent done 162 # Convert it to percent done. 163 progress=(100-int($(NF-2)))"%" 164 } 165 # When we int()-ify "hours", it converts stuff like "NOW" and "-" into 166 # 0. In those cases, set it to hours_on, so they will cancel out in 167 # the "hours_ago" calculation later on. 168 if (hours == 0) 169 hours=hours_on 170 171 if (test_seen) { 172 print "test="hours_on 173 print "test_type="test_type 174 print "test_status="status 175 print "test_progress="progress 176 } 177 # Not all drives report hours_on 178 if (hours_on && hours) { 179 total_hours_ago=(hours_on-hours) 180 days_ago=int(total_hours_ago/24) 181 hours_ago=(total_hours_ago % 24) 182 if (days_ago != 0) 183 ago_str=days_ago"d" 184 if (hours_ago !=0) 185 ago_str=ago_str""hours_ago"h" 186 print "test_ended="ago_str 187 } 188} 189 190END {print "type="type; ORS="\n"; print ""} 191'); 192fi 193type=$(echo "$out" | grep '^type=' | cut -d '=' -f 2) 194 195# If type is not set by now, either we don't have a block device 196# or smartctl failed. Either way, default to ATA and set $out to 197# nothing. 198if [ -z "$type" ]; then 199 type="sata" 200 out= 201fi 202 203case $script in 204smart) 205 # Print temperature plus common predictors of drive failure 206 if [ "$type" = "sas" ] ; then 207 scripts="temp|health|r_ucor|w_ucor" 208 elif [ "$type" = "sata" ] ; then 209 scripts="temp|health|ata_err|realloc|rep_ucor|cmd_to|pend_sec|off_ucor" 210 elif [ "$type" = "nvme" ] ; then 211 scripts="temp|health|nvme_err" 212 fi 213 ;; 214smartx) 215 # Print some other interesting stats 216 if [ "$type" = "sas" ] ; then 217 scripts="hours_on|defect|nonmed|r_proc|w_proc" 218 elif [ "$type" = "sata" ] ; then 219 scripts="hours_on|pwr_cyc" 220 elif [ "$type" = "nvme" ] ; then 221 scripts="hours_on|pwr_cyc" 222 fi 223 ;; 224smart_test) 225 scripts="test_type|test_status|test_progress|test_ended" 226 ;; 227*) 228 scripts="$script" 229esac 230 231with_vals=$(echo "$out" | grep -E "$scripts") 232if [ -n "$with_vals" ]; then 233 echo "$with_vals" 234 without_vals=$(echo "$scripts" | tr "|" "\n" | 235 grep -v -E "$(echo "$with_vals" | 236 awk -F "=" '{print $1}')" | awk '{print $0"="}') 237else 238 without_vals=$(echo "$scripts" | tr "|" "\n" | awk '{print $0"="}') 239fi 240 241if [ -n "$without_vals" ]; then 242 echo "$without_vals" 243fi 244