feat: Use both RssAnon and VmRSS in CI benchmarks (qdrant#133)

* feat: Use RssAnon instead of VmRSS

* feat: Show both RSSAnon and VMRss

Author: KShivendu

tools/qdrant_collect_stats.sh

@@ -16,11 +16,13 @@ BENCH_SERVER_NAME=${SERVER_NAME:-"benchmark-server-1"}
 
 IP_OF_THE_SERVER=$(bash "${SCRIPT_PATH}/${CLOUD_NAME}/get_public_ip.sh" "$BENCH_SERVER_NAME")
 
-MEMORY_USAGE=$(ssh -t "${SERVER_USERNAME}@${IP_OF_THE_SERVER}" "grep VmRSS /proc/\$(pidof qdrant)/status | awk '{print \$2}'")
+VM_RSS_MEMORY_USAGE=$(ssh -t "${SERVER_USERNAME}@${IP_OF_THE_SERVER}" "grep VmRSS /proc/\$(pidof qdrant)/status | awk '{print \$2}'")
+RSS_ANON_MEMORY_USAGE=$(ssh -t "${SERVER_USERNAME}@${IP_OF_THE_SERVER}" "grep RssAnon /proc/\$(pidof qdrant)/status | awk '{print \$2}'")
 
 CURRENT_DATE=$(date +%Y-%m-%d-%H-%M-%S)
 
-echo "$MEMORY_USAGE" > results/memory-usage-"${CURRENT_DATE}".txt
+echo "$VM_RSS_MEMORY_USAGE" > results/vm-rss-memory-usage-"${CURRENT_DATE}".txt
+echo "$RSS_ANON_MEMORY_USAGE" > results/rss-anon-memory-usage-"${CURRENT_DATE}".txt
 
 ROOT_API_RESPONSE=$(ssh -t "${SERVER_USERNAME}@${IP_OF_THE_SERVER}" "curl -s http://localhost:6333/")
 

tools/run_ci.sh

@@ -13,7 +13,8 @@ bash -x "${SCRIPT_PATH}/run_remote_benchmark.sh"
 # -t sorts by modification time
 export SEARCH_RESULTS_FILE=$(ls -t results/*-search-*.json | head -n 1)
 export UPLOAD_RESULTS_FILE=$(ls -t results/*-upload-*.json | head -n 1)
-export MEMORY_USAGE_FILE=$(ls -t results/memory-usage-*.txt | head -n 1)
+export VM_RSS_MEMORY_USAGE_FILE=$(ls -t results/vm-rss-memory-usage-*.txt | head -n 1)
+export RSS_ANON_MEMORY_USAGE_FILE=$(ls -t results/rss-anon-memory-usage-*.txt | head -n 1)
 export ROOT_API_RESPONSE_FILE=$(ls -t results/root-api-*.json | head -n 1)
 
 bash -x "${SCRIPT_PATH}/upload_results_postgres.sh"

tools/run_client_script.sh

@@ -27,7 +27,8 @@ ssh -tt "${SERVER_USERNAME}@${IP_OF_THE_CLIENT}" "${RUN_EXPERIMENT}"
 
 SEARCH_RESULT_FILE=$(ssh "${SERVER_USERNAME}@${IP_OF_THE_CLIENT}" "ls -t results/*-search-*.json | head -n 1")
 UPLOAD_RESULT_FILE=$(ssh "${SERVER_USERNAME}@${IP_OF_THE_CLIENT}" "ls -t results/*-upload-*.json | head -n 1")
-MEMORY_USAGE_FILE=$(ssh "${SERVER_USERNAME}@${IP_OF_THE_CLIENT}" "ls -t results/memory-usage-*.txt | head -n 1")
+VM_RSS_MEMORY_USAGE_FILE=$(ssh "${SERVER_USERNAME}@${IP_OF_THE_CLIENT}" "ls -t results/vm-rss-memory-usage-*.txt | head -n 1")
+RSS_ANON_MEMORY_USAGE_FILE=$(ssh "${SERVER_USERNAME}@${IP_OF_THE_CLIENT}" "ls -t results/rss-anon-memory-usage-*.txt | head -n 1")
 
 mkdir -p results
 

tools/upload_results_postgres.sh

@@ -8,7 +8,7 @@
 # 	id SERIAL PRIMARY key,
 # 	engine VARCHAR(255),
 # 	branch VARCHAR(255),
-#   commit CHAR(40),
+# 	commit CHAR(40),
 # 	dataset VARCHAR(255),
 # 	measure_timestamp TIMESTAMP,
 # 	upload_time real,
@@ -17,12 +17,14 @@
 # 	mean_precisions real,
 # 	p95_time real,
 # 	p99_time real,
-# 	memory_usage real
+# 	vm_rss_mem real,
+# 	rss_anon_mem real
 # );
 
 SEARCH_RESULTS_FILE=${SEARCH_RESULTS_FILE:-""}
 UPLOAD_RESULTS_FILE=${UPLOAD_RESULTS_FILE:-""}
-MEMORY_USAGE_FILE=${MEMORY_USAGE_FILE:-""}
+VM_RSS_MEMORY_USAGE_FILE=${VM_RSS_MEMORY_USAGE_FILE:-""}
+RSS_ANON_MEMORY_USAGE_FILE=${RSS_ANON_MEMORY_USAGE_FILE:-""}
 ROOT_API_RESPONSE_FILE=${ROOT_API_RESPONSE_FILE:-""}
 POSTGRES_TABLE=${POSTGRES_TABLE:-"benchmark"}
 
@@ -41,8 +43,13 @@ if [[ -z "$UPLOAD_RESULTS_FILE" ]]; then
 fi
 
 
-if [[ -z "$MEMORY_USAGE_FILE" ]]; then
-  echo "MEMORY_USAGE_FILE is not set"
+if [[ -z "$VM_RSS_MEMORY_USAGE_FILE" ]]; then
+  echo "VM_RSS_MEMORY_USAGE_FILE is not set"
+  exit 1
+fi
+
+if [[ -z "$RSS_ANON_MEMORY_USAGE_FILE" ]]; then
+  echo "RSS_ANON_MEMORY_USAGE_FILEis not set"
   exit 1
 fi
 
@@ -59,15 +66,16 @@ P99_TIME=$(jq -r '.results.p99_time' "$SEARCH_RESULTS_FILE")
 UPLOAD_TIME=$(jq -r '.results.upload_time' "$UPLOAD_RESULTS_FILE")
 INDEXING_TIME=$(jq -r '.results.total_time' "$UPLOAD_RESULTS_FILE")
 
-MEMORY_USAGE=$(cat "$MEMORY_USAGE_FILE")
+VM_RSS_MEMORY_USAGE=$(cat "$VM_RSS_MEMORY_USAGE_FILE")
+RSS_ANON_MEMORY_USAGE=$(cat "$RSS_ANON_MEMORY_USAGE_FILE")
 
 QDRANT_COMMIT=$(jq -r '.commit' "$ROOT_API_RESPONSE_FILE")
 
 MEASURE_TIMESTAMP=${MEASURE_TIMESTAMP:-$(date -u +"%Y-%m-%dT%H:%M:%SZ")}
 
 
 docker run --rm jbergknoff/postgresql-client "postgresql://qdrant:${POSTGRES_PASSWORD}@${POSTGRES_HOST}:5432/postgres" -c "
-INSERT INTO ${POSTGRES_TABLE} (engine, branch, commit, dataset, measure_timestamp, upload_time, indexing_time, rps, mean_precisions, p95_time, p99_time, memory_usage)
-VALUES ('qdrant-ci', '${QDRANT_VERSION}', '${QDRANT_COMMIT}', '${DATASETS}', '${MEASURE_TIMESTAMP}', ${UPLOAD_TIME}, ${INDEXING_TIME}, ${RPS}, ${MEAN_PRECISIONS}, ${P95_TIME}, ${P99_TIME}, ${MEMORY_USAGE});
+INSERT INTO ${POSTGRES_TABLE} (engine, branch, commit, dataset, measure_timestamp, upload_time, indexing_time, rps, mean_precisions, p95_time, p99_time, vm_rss_mem, rss_anon_mem)
+VALUES ('qdrant-ci', '${QDRANT_VERSION}', '${QDRANT_COMMIT}', '${DATASETS}', '${MEASURE_TIMESTAMP}', ${UPLOAD_TIME}, ${INDEXING_TIME}, ${RPS}, ${MEAN_PRECISIONS}, ${P95_TIME}, ${P99_TIME}, ${VM_RSS_MEMORY_USAGE}, ${RSS_ANON_MEMORY_USAGE});
 "