Add e2e script for creating branch with benchmark results based on experiment runs.

perf-benchmarking-for-releases/create_benchmark_tables.sh

@@ -0,0 +1,195 @@
+#!/bin/bash
+
+# Copyright 2025 Google LLC
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#       http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+# Fail on anything unexpected.
+set -euo pipefail
+
+# Script Documentation
+if [ "$#" -ne 5 ]; then
+    echo "Usage: $0 <GCSFUSE_VERSION> <REGION> <MACHINE_TYPE> <NETWORKING> <DISK_TYPE>"
+    echo ""
+    echo "Example:"
+    echo "  bash create_benchmark_tables.sh  <tag, commit-id-on-master, branch-name> us-south1 c4-standard-96 'gVNIC+ tier_1 networking (200Gbps)' 'Hyperdisk balanced'"
+    exit 1
+fi
+
+TMP_DIR=$(mktemp -d -t create_benchmark_tables.XXXXXX)
+trap 'rm -rf "$TMP_DIR"' EXIT # Ensure cleanup on exit
+
+TABLES_FILE="${TMP_DIR}/tables.md"
+RESULTS_BUCKET_NAME="gcsfuse-release-benchmarks-results"
+GCSFUSE_VERSION=$1
+REGION=$2
+MACHINE_TYPE=$3
+NETWORKING=$4
+DISK_TYPE=$5
+
+# Files to read from GCS. Basenames
+RANDOM_READ_RES_BASENAME="gcsfuse-random-read-workload-benchmark.json"
+SEQ_READ_RES_BASENAME="gcsfuse-sequential-read-workload-benchmark.json"
+WRITE_RES_BASENAME="gcsfuse-write-workload-benchmark.json"
+
+JSON_FILES_BASENAMES=(
+    "$RANDOM_READ_RES_BASENAME"
+    "$SEQ_READ_RES_BASENAME"
+    "$WRITE_RES_BASENAME"
+)
+
+if ! gcloud storage objects describe "gs://${RESULTS_BUCKET_NAME}/${GCSFUSE_VERSION}/${MACHINE_TYPE}/success.txt" &>/dev/null; then
+    echo "Unable to locate success.txt file in the bucket gs://${RESULTS_BUCKET_NAME}/${GCSFUSE_VERSION}/${MACHINE_TYPE}/. Exiting..." >&2
+    exit 1
+fi
+echo "Found the success.txt file in the bucket."
+
+for basename in "${JSON_FILES_BASENAMES[@]}"; do
+    gcloud storage cp "gs://${RESULTS_BUCKET_NAME}/${GCSFUSE_VERSION}/${MACHINE_TYPE}/${basename}" "${TMP_DIR}/${basename}"
+done
+
+# Update variables to point to local copies
+RANDOM_READ_RES="${TMP_DIR}/${RANDOM_READ_RES_BASENAME}"
+SEQ_READ_RES="${TMP_DIR}/${SEQ_READ_RES_BASENAME}"
+WRITE_RES="${TMP_DIR}/${WRITE_RES_BASENAME}"
+
+# Global arrays for storing column data
+ROW_COUNT=0
+declare -a filesize bw_bytes nrfiles iops lat_mean bs
+
+# Helper methods
+# Get size of jobs
+job_count() {
+    local file=$1
+    local count
+    count=$(jq '.jobs | length' "$file")
+    if [[ "$count" -le 0 ]]; then
+        echo "Exiting: found no jobs in experiment result file: $file" >&2
+        exit 1
+    fi
+    echo "$count"
+}
+
+# Generalized function to populate an array from jq output
+# Args: $1: json_file, $2: jq_path_job_level, $3: jq_path_global_level, $4: field_to_extract, $5: array_name_ref
+populate_array_from_jq() {
+    local file="$1"
+    local jq_job_path_template="$2" # e.g., '.jobs[]?."job options"' or '.jobs[]?.read'
+    local jq_global_path_template="$3" # e.g., '."global options"'
+    local field_name="$4"
+    local -n array_ref="$5" # Nameref to the array to populate
+
+    array_ref=() # Clear the array
+
+    local jq_job_query="${jq_job_path_template}[\"${field_name}\"]?"
+    local jq_global_query="${jq_global_path_template}[\"${field_name}\"]?"
+
+    # Read values using mapfile for efficiency
+    mapfile -t values < <(jq -r "$jq_job_query" "$file")
+
+    for i in "${!values[@]}"; do
+        local value="${values[i]}"
+        if [[ "$value" == "null" ]]; then
+            value=$(jq -r "$jq_global_query" "$file")
+        fi
+        if [[ "$value" == "null" ]]; then
+            echo "Value is null for field [$field_name] from file [$file] (job path: $jq_job_query, global path: $jq_global_query) and not found in global options." >&2
+            exit 1
+        fi
+        array_ref+=("$value")
+    done
+
+    if [[ "${#array_ref[@]}" -ne "$ROW_COUNT" ]]; then
+        echo "Not enough rows (${#array_ref[@]} vs $ROW_COUNT) for field [$field_name] for file [$file]." >&2
+        exit 1
+    fi
+}
+
+# Populate all columns based on operation type (read/write)
+# Args: $1: json_file, $2: operation_type ("read" or "write")
+populate_all_columns() {
+    local file="$1"
+    local op_type="$2" # "read" or "write"
+    local lat_ns_op_type="${op_type}.lat_ns" # "read.lat_ns" or "write.lat_ns"
+
+    populate_array_from_jq "$file" '.jobs[]?."job options"' '."global options"' "bs" "bs"
+    populate_array_from_jq "$file" '.jobs[]?."job options"' '."global options"' "filesize" "filesize"
+    populate_array_from_jq "$file" '.jobs[]?."job options"' '."global options"' "nrfiles" "nrfiles"
+
+    populate_array_from_jq "$file" ".jobs[]?.$op_type" '."global options"' "bw_bytes" "bw_bytes"
+    populate_array_from_jq "$file" ".jobs[]?.$op_type" '."global options"' "iops" "iops"
+    populate_array_from_jq "$file" ".jobs[]?.$lat_ns_op_type" '."global options"' "mean" "lat_mean"
+}
+
+declare -A bytes_in=( [MB]="1000000" [GB]="1000000000" )
+convert_bytes_to() {
+    local division="${bytes_in[$1]}"
+    local precision=3
+    for ((i = 0; i < ROW_COUNT; i++)); do
+        bw_bytes[i]=$(awk -v prec="$precision" -v div="$division" -v bval="${bw_bytes[i]}" 'BEGIN { printf "%.*f\n", prec, bval / div }')
+    done
+}
+
+format_iops_to_kilo() {
+    for ((i = 0; i < ROW_COUNT; i++)); do
+        iops[i]=$(awk -v n="${iops[i]}" 'BEGIN { printf "%.2fK\n", n / 1000 }')
+    done
+}
+
+convert_lat_mean_ns_to_ms() {
+    local -n lat_array_ref="$1"
+    local precision=2
+    for ((i = 0; i < ROW_COUNT; i++)); do
+        lat_array_ref[i]=$(awk -v p="$precision" -v ns="${lat_array_ref[i]}" 'BEGIN { printf "%.*fms\n", p, ns / 1E6 }')
+    done
+}
+
+create_table() {
+    local table_name="$1"
+    local file="$2"
+    local workflow_type="$3" # "read" or "write"
+    local bw_in="$4"
+    echo "### $table_name"
+    ROW_COUNT=$(job_count "$file")
+    populate_all_columns "$file" "$workflow_type"
+    convert_bytes_to "${bw_in}"
+    format_iops_to_kilo
+    convert_lat_mean_ns_to_ms "lat_mean"
+
+    echo "| File Size | BlockSize | nrfiles | Bandwidth in (${bw_in}/sec) | IOPs | IOPs Avg Latency (ms) |"
+    echo "|---|---|---|---|---|---|---|---|"
+    for ((i = 0; i < ROW_COUNT; i++)); do
+        echo "| ${filesize[i]} | ${bs[i]} | ${nrfiles[i]} | ${bw_bytes[i]} | ${iops[i]} | ${lat_mean[i]} |"
+    done
+    echo ""
+}
+
+create_tables_markdown_content() {
+    echo "## GCSFuse Benchmarking on ${MACHINE_TYPE%%-*} machine-type"
+    echo "* VM Type: ${MACHINE_TYPE}"
+    echo "* VM location: ${REGION}"
+    echo "* Networking: ${NETWORKING}"
+    echo "* Disk Type: ${DISK_TYPE}"
+    echo "* GCS Bucket location: ${REGION}"
+    echo ""
+    create_table "Sequential Reads" "$SEQ_READ_RES" "read" "GB"
+    create_table "Random Reads" "$RANDOM_READ_RES" "read" "MB"
+    create_table "Sequential Writes" "$WRITE_RES" "write" "MB"
+}
+
+create_tables_markdown_content > "${TABLES_FILE}"
+
+# copy file to results bucket
+gcloud storage cp "${TABLES_FILE}" "gs://${RESULTS_BUCKET_NAME}/${GCSFUSE_VERSION}/${MACHINE_TYPE}/"
+
+echo "Benchmark table successfully created and uploaded to gs://${RESULTS_BUCKET_NAME}/${GCSFUSE_VERSION}/${MACHINE_TYPE}/tables.md"

perf-benchmarking-for-releases/fio-job-files/random-read-workload.fio

@@ -75,4 +75,4 @@ stonewall
 directory=${DIR}
 bs=1M
 filesize=1GB
-nrfiles=2
+nrfiles=10

perf-benchmarking-for-releases/fio-job-files/sequential-read-workload.fio

@@ -83,4 +83,4 @@ directory=${DIR}
 bs=1M
 filesize=1GB
 rw=read
-nrfiles=2
+nrfiles=10