High Block I/O metrics in monitoring

[emimontesdeoca]

Hey everyone,

I'm a bit obssesed with the monitoring on my Dokploy instance and I see that I have high Block I/O metrics that do not match with individual monitoring on each service that I have deployed.

I'd like to find where I'm making this such big writes.

Is there any place where I can find more information about this and possible check where this metrics are coming from?

As I've said on top, I've already checked monitoring on each service but it's way lower when you sum them all.

Thanks!

[djsisson]

rb=0; wb=0
for dev in $(lsblk -dn -o NAME); do
  case $dev in
    loop*|ram*|sr*|fd*) continue ;;
  esac
  ss=$(cat /sys/block/$dev/queue/hw_sector_size 2>/dev/null)
  vals=$(awk -v ss=$ss -v dev=$dev '$3==dev {print $6*ss, $10*ss}' /proc/diskstats)
  readb=$(echo $vals | awk '{print $1}')
  writeb=$(echo $vals | awk '{print $2}')
  rb=$((rb+readb))
  wb=$((wb+writeb))
done
echo "readBytes=$rb writeBytes=$wb"

or

iostat -d -k | awk 'NR>3 && $1 !~ /^(loop|ram|sr|fd)/ {readKB+=$6; writeKB+=$7} END {print "readKB="readKB, "writeKB="writeKB, "readBytes="readKB*1024, "writeBytes="writeKB*1024}'

its probably correct