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

steps:
  - command: "system-test/testnet-automation.sh"
    label: "GCE - GPU Enabled 50 Nodes"
    env:
      UPLOAD_RESULTS_TO_SLACK: "true"
      CLOUD_PROVIDER: "gce"
      TESTNET_TAG: "gce-perf-gpu-enabled"
      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: 1
      CLIENT_OPTIONS: "bench-tps=1=--tx_count 10000 --thread-batch-sleep-ms 250"
      TESTNET_ZONES: "us-west1-a,us-west1-b,us-central1-a,europe-west4-a"
      ALLOW_BOOT_FAILURES: "true"
      USE_PUBLIC_IP_ADDRESSES: "false"
      ADDITIONAL_FLAGS: "--dedicated"
      TEST_TYPE: "fixed_duration"
    agents:
      - "queue=gce-deploy"
Implement nightly performance tests (#6140) * Implement nightly performance tests on colo 2019-10-10 08:12:13 -07:00			`steps:`
Refactor system-test dir structure and add stability test (#7530) 2019-12-17 11:58:10 -08:00			`- command: "system-test/testnet-automation.sh"`
Define 10, 25, 50 node testcases (#6557) 2019-10-25 10:43:53 -07:00			`label: "GCE - GPU Enabled 50 Nodes"`
Implement nightly performance tests (#6140) * Implement nightly performance tests on colo 2019-10-10 08:12:13 -07:00			`env:`
Push perf test results to slack app (#6371) * Add script to publish testnet results to slack * Obscure webhook URL * fixup * Replace read with cat redirection * Turn back on net restart * Pick nits * Make symlink before trying to delete its contents * Display test config in slack and pick Trents nit not to maybe rm -rf /* * Clean up results print * Minor nits * Turn the test settings back up to 11 * typo * Shellcheck * Just a few more fields * fix payload formatting * Del clear-config.sh * Mount secondary * Add commit SHA link and Grafana time range URL * Add fancy buttons instead of text URLs * Tighten up test config display * Fixup display nits * chellsheck * Rebase and fix typo 2019-10-21 17:00:17 -07:00			`UPLOAD_RESULTS_TO_SLACK: "true"`
Implement nightly performance tests (#6140) * Implement nightly performance tests on colo 2019-10-10 08:12:13 -07:00			`CLOUD_PROVIDER: "gce"`
Clean up DB names in automation (#6949) 2019-11-14 07:20:10 -08:00			`TESTNET_TAG: "gce-perf-gpu-enabled"`
Rename TEST_DURATION to TEST_DURATION_SECONDS (#6493) 2019-10-21 20:24:46 -07:00			`TEST_DURATION_SECONDS: 600`
Push perf test results to slack app (#6371) * Add script to publish testnet results to slack * Obscure webhook URL * fixup * Replace read with cat redirection * Turn back on net restart * Pick nits * Make symlink before trying to delete its contents * Display test config in slack and pick Trents nit not to maybe rm -rf /* * Clean up results print * Minor nits * Turn the test settings back up to 11 * typo * Shellcheck * Just a few more fields * fix payload formatting * Del clear-config.sh * Mount secondary * Add commit SHA link and Grafana time range URL * Add fancy buttons instead of text URLs * Tighten up test config display * Fixup display nits * chellsheck * Rebase and fix typo 2019-10-21 17:00:17 -07:00			`NUMBER_OF_VALIDATOR_NODES: 50`
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 2019-10-24 14:12:25 -07:00			`ENABLE_GPU: "true"`
Implement nightly performance tests (#6140) * Implement nightly performance tests on colo 2019-10-10 08:12:13 -07:00			`VALIDATOR_NODE_MACHINE_TYPE: "--machine-type n1-standard-16 --accelerator count=2,type=nvidia-tesla-v100"`
Limit performance testcases to 40k TPS client (#9906) 2020-05-06 20:36:13 -07:00			`NUMBER_OF_CLIENT_NODES: 1`
			`CLIENT_OPTIONS: "bench-tps=1=--tx_count 10000 --thread-batch-sleep-ms 250"`
Implement nightly performance tests (#6140) * Implement nightly performance tests on colo 2019-10-10 08:12:13 -07:00			`TESTNET_ZONES: "us-west1-a,us-west1-b,us-central1-a,europe-west4-a"`
Implement allowing validator boot failure into automation (#6589) * Pass allow boot failures through create AND start * Extend sleep timeout to all nodes * Add 100 node testcase * Reduce consistent sleep 2019-10-28 15:43:40 -07:00			`ALLOW_BOOT_FAILURES: "true"`
removes use of public ip addresses from system tests Using global IPs causes outbound traffic which costs money: https://github.com/solana-labs/solana/pull/18728#issuecomment-884290209 2021-08-11 07:50:19 -07:00			`USE_PUBLIC_IP_ADDRESSES: "false"`
Implement allowing validator boot failure into automation (#6589) * Pass allow boot failures through create AND start * Extend sleep timeout to all nodes * Add 100 node testcase * Reduce consistent sleep 2019-10-28 15:43:40 -07:00			`ADDITIONAL_FLAGS: "--dedicated"`
Automated test framework can run scripts on launched clusters. Add offline stake operations test case and script. (#8510) automerge 2020-03-18 14:57:19 -07:00			`TEST_TYPE: "fixed_duration"`
Implement nightly performance tests (#6140) * Implement nightly performance tests on colo 2019-10-10 08:12:13 -07:00			`agents:`
Update testnet testcases to point at new buildkite agent queues (#7544) automerge 2019-12-17 20:32:31 -08:00			`- "queue=gce-deploy"`