Specify machine type without necessarily enabling GPU (#6529)

* Specifiy machine type without necessarily enabling GPU

* Make long arg, extend --enable-gpu to automation

* Set machine types only in one place

* Fixup

* Fixup flag in automation

* Typo

* shellcheck

system-test/testnet-performance/colo-gpu-perf.yml

@@ -5,6 +5,7 @@ steps:
       UPLOAD_RESULTS_TO_SLACK: "true"
       CLOUD_PROVIDER: "colo"
       TESTNET_TAG: "colo-edge-perf-gpu-enabled"
+      ENABLE_GPU: "true"
       RAMP_UP_TIME: 0
       TEST_DURATION_SECONDS: 600
       NUMBER_OF_VALIDATOR_NODES: 4

system-test/testnet-performance/gce-cpu-only-perf-10-node.yml

@@ -8,6 +8,7 @@ steps:
       RAMP_UP_TIME: 60
       TEST_DURATION_SECONDS: 300
       NUMBER_OF_VALIDATOR_NODES: 10
+      ENABLE_GPU: "false"
       VALIDATOR_NODE_MACHINE_TYPE: "--machine-type n1-standard-16"
       NUMBER_OF_CLIENT_NODES: 1
       CLIENT_OPTIONS: "bench-tps=1=--tx_count 80000 --thread-batch-sleep-ms 1000"

system-test/testnet-performance/gce-cpu-only-perf-5-node.yml

@@ -8,6 +8,8 @@ steps:
       RAMP_UP_TIME: 0
       TEST_DURATION_SECONDS: 600
       NUMBER_OF_VALIDATOR_NODES: 5
+      ENABLE_GPU: "false"
+      VALIDATOR_NODE_MACHINE_TYPE: "--machine-type n1-standard-16"
       NUMBER_OF_CLIENT_NODES: 2
       CLIENT_OPTIONS: "bench-tps=2=--tx_count 15000 --thread-batch-sleep-ms 250"
       TESTNET_ZONES: "us-west1-a,us-west1-b,us-central1-a,europe-west4-a"

system-test/testnet-performance/gce-gpu-perf-5-node.yml

@@ -8,6 +8,7 @@ steps:
       RAMP_UP_TIME: 0
       TEST_DURATION_SECONDS: 600
       NUMBER_OF_VALIDATOR_NODES: 5
+      ENABLE_GPU: "true"
       VALIDATOR_NODE_MACHINE_TYPE: "--machine-type n1-standard-16 --accelerator count=2,type=nvidia-tesla-v100"
       NUMBER_OF_CLIENT_NODES: 2
       CLIENT_OPTIONS: "bench-tps=2=--tx_count 15000 --thread-batch-sleep-ms 250"

system-test/testnet-performance/gce-gpu-perf-50-node.yml

@@ -8,10 +8,11 @@ steps:
       RAMP_UP_TIME: 0
       TEST_DURATION_SECONDS: 600
       NUMBER_OF_VALIDATOR_NODES: 50
+      ENABLE_GPU: "true"
       VALIDATOR_NODE_MACHINE_TYPE: "--machine-type n1-standard-16 --accelerator count=2,type=nvidia-tesla-v100"
       NUMBER_OF_CLIENT_NODES: 2
       CLIENT_OPTIONS: "bench-tps=2=--tx_count 15000 --thread-batch-sleep-ms 250"
       TESTNET_ZONES: "us-west1-a,us-west1-b,us-central1-a,europe-west4-a"
-      ADDITIONAL_FLAGS: ""
+      ADDITIONAL_FLAGS: "--dedicated --allow-boot-failures"
     agents:
       - "queue=testnet-deploy"

system-test/testnet-performance/testnet-automation.sh

@@ -74,16 +74,18 @@ function launchTestnet() {
   case $CLOUD_PROVIDER in
     gce)
     # shellcheck disable=SC2068
+    # shellcheck disable=SC2086
       net/gce.sh create \
         -d pd-ssd \
         -n "$NUMBER_OF_VALIDATOR_NODES" -c "$NUMBER_OF_CLIENT_NODES" \
-        "$maybeMachineType" "$VALIDATOR_NODE_MACHINE_TYPE" \
+        $maybeCustomMachineType $VALIDATOR_NODE_MACHINE_TYPE "$maybeEnableGpu" \
         -p "$TESTNET_TAG" ${TESTNET_CLOUD_ZONES[@]/#/"-z "} ${ADDITIONAL_FLAGS[@]/#/" "}
       ;;
     colo)
     # shellcheck disable=SC2068
+    # shellcheck disable=SC2086
       net/colo.sh create \
-        -n "$NUMBER_OF_VALIDATOR_NODES" -c "$NUMBER_OF_CLIENT_NODES" -g \
+        -n "$NUMBER_OF_VALIDATOR_NODES" -c "$NUMBER_OF_CLIENT_NODES" "$maybeEnableGpu" \
         -p "$TESTNET_TAG" ${ADDITIONAL_FLAGS[@]/#/" "}
       ;;
     *)
@@ -169,6 +171,13 @@ if [[ -z $NUMBER_OF_VALIDATOR_NODES ]] ; then
   exit 1
 fi
 
+if [[ -z $ENABLE_GPU ]] ; then
+  ENABLE_GPU=false
+fi
+if [[ "$ENABLE_GPU" = "true" ]] ; then
+  maybeEnableGpu="--enable-gpu"
+fi
+
 if [[ -z $NUMBER_OF_CLIENT_NODES ]] ; then
   echo NUMBER_OF_CLIENT_NODES not defined
   exit 1
@@ -193,7 +202,7 @@ source ci/upload-ci-artifact.sh
 source system-test/testnet-performance/upload_results_to_slack.sh
 
 maybeClientOptions=${CLIENT_OPTIONS:+"-c"}
-maybeMachineType=${VALIDATOR_NODE_MACHINE_TYPE:+"-G"}
+maybeCustomMachineType=${VALIDATOR_NODE_MACHINE_TYPE:+"--custom-machine-type"}
 
 IFS=, read -r -a TESTNET_CLOUD_ZONES <<<"${TESTNET_ZONES}"
 
@@ -203,6 +212,7 @@ RESULT_DETAILS="Test failed to finish"
 
 TEST_PARAMS_TO_DISPLAY=(CLOUD_PROVIDER \
                         NUMBER_OF_VALIDATOR_NODES \
+                        ENABLE_GPU \
                         VALIDATOR_NODE_MACHINE_TYPE \
                         NUMBER_OF_CLIENT_NODES \
                         CLIENT_OPTIONS \