heatmap-per-vm.sh

#!/bin/bash

function dumpVSDL1()
{
    local vsd="$1"
    echo $vsd;
    detail=$(sc vsd show display detail uuid $vsd)
    cap=`sc vsd show display detail uuid $vsd | grep CapacityBytes | awk '{print $3}'`
    headers=`sc vsd show display detail uuid $vsd | grep  Headers | awk '{ print $3 }' | sed -e 's/,//g'`
    echo "Headers are $headers"
    l1=$(sc rsd interpret extent $headers-1024 | grep L1table | awk '{ print $4 }');
    echo "Capacity:"
    echo "$cap"
    echo "L1 address: $l1"
    ext=$(sc rsd interpret extent ${l1%,*}-$((16 * 1024)) | grep chunks) # read the L0
    if [ -z "$ext" ]; then
        # spoof a chunk for pre-L0 tables
        ext="[0]${l1%,*}-$(( ($cap / 1048576) / 29 * 512 + 512))"
    fi
    for w in $(echo $ext); do # break into words
        chunk=$(echo "$w" | egrep '\[[0-9]+\][0-9]:[0-9]+:[0-9]:0x'| gawk ' BEGIN { mb = 1048576 / 512} /VSD/ { print $0 } /:/{ match ($1, /(0:.*)(0x[0-9a-f]+)-([0-9]+)/, addr); c = addr[3] / 1048576; if (c == 1) { print $0 ; } else  { for (i = 0; i < c; ++i) { printf "%s0x%x-1048576 %s\n", addr[1], strtonum(addr[2]) + i * mb, $2; }}} ') # chunks
        if ! [ -z "$chunk" ]; then
            addr=${chunk%%-*};
            addr=${addr##*]};
            length=${chunk##*-};
            echo "Found L1 extent $addr-$length"
            sc rsd interpret extent $addr-$length
        fi
    done
}

function get_percentile
{
    awk "BEGIN { print int (.$2 * $1); }"
}

function analyzeVSDHeatmap()
{
    local uuid="$1"
    vmguid="$(sc vsd show display attachments uuid $uuid | awk '{ print $2 }' | tail -n +3)"
    vmname="$(sc vm show display detail guid $vmguid | grep Name | cut -d ":" -f 2)"
    if ! [ -z "$vmguid" ]; then
        echo -e "From VM: $vmname\n"
    fi
    echo "--- Analyzing Heatmap for VSD $uuid ---"
    echo ""
    if [ -z "$(sc vsd show | grep ONLINE | grep "$uuid")" ]; then
        echo "No ONLINE vsd with uuid $1";
        return 1
    fi



    echo "Counting entry allocations..."
    entryCount="$(dumpVSDL1 "$uuid" | gawk ' /^ address/ { match($1, "([0-9]+)", ary); if (ary[1] % 2 == 0) print $0; }' | grep -v '0:0:0:0x0' | wc -l)"

    echo "Total allocated entries: $entryCount";

    if [ $entryCount == "0" ]; then
        echo "No allocated entries; skipping VSD $uuid";
        return 0
    fi

    echo "Getting complete heatmap..."
    distribution="$(sc vsdheatmap show display data uuid $uuid edge 0 | sed -s 's/\] \[/\]X\[/g' | tr 'X' '\n' | gawk ' BEGIN { FS=","; top=0; } /^\[/ { match($1, "([0-9]+)", h); match($2, "([0-9]+)", e); dist[e[1]]=h[1]; if (e[1] > top) { top = e[1]; } } END { for (i = 0; i <= top; ++i) { count=dist[i]; if (count == "") { count = 0; }  printf "%5d: %5d ", count, i; for (j = 0; j < count; ++j) { printf "#"; } printf "\n"; } }' | sort -n)"

    totalHeat="$(echo "$distribution" | gawk '{ total += $1 } END { print total; } ')";
    maxHeat="$(echo "$distribution" | tail -1 | gawk '{ total += $1 } END { print total; } ')";
    echo "Total heat: $totalHeat";

    echo "Getting heat distributions:"
    for i in 60 70 80 90 95 96 97 98 99; do
        hp="$(get_percentile $maxHeat $i)"
        percCount="$(echo "$distribution" | awk "BEGIN { total = 0; } { if (int(\$1) >= $hp) total += 1; } END { print  total; } ")"
        if [ -z "$percCount" ]; then
            percCount=0;
        fi
        entryPercent="$(awk "BEGIN { print ($percCount / $entryCount) * 100; }")"
        echo "   Percentile $i: entries with heat >= $hp: $percCount ($entryPercent%)"
    done;

    echo "Getting entry distributions:"
    for i in 1 2 3 5 8 10 20 30 40 50; do
        numInPerc="$(awk " BEGIN { print int($entryCount * ($i / 100)); } ")"
        heatInPerc="$(echo "$distribution" | tail -n $numInPerc | awk " { total += int(\$1); } END { print total }")"
        if [ -z "$heatInPerc" ]; then
            heatInPerc=0;
        fi
        heatSharePerc="$(awk " BEGIN { print ($heatInPerc / $totalHeat) * 100; } ")"
        gigs="$(awk " BEGIN { print ($numInPerc * 2) / 1024 } ")"
        echo "   There are $numInPerc entries, or ${gigs}GB, (at replicationFactor=2), in the hottest $i%, which account for $heatSharePerc% of all heat ($heatInPerc/$totalHeat)"
    done;
}

if [ -z "$1" ]; then
    # run on all online VSDs
    uuids="$(sc vsd show | grep ONLINE | awk '{ print $1 }')"
else
    uuids=$@;
fi

for i in $uuids; do
    analyzeVSDHeatmap "$i";
done