llvm-premerge-checks/scripts/buildkite/analyze_jobs.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## License\n",
    "\n",
    "Copyright 2020 Google LLC\n",
    "\n",
    "Licensed under the the Apache License v2.0 with LLVM Exceptions (the \"License\");\n",
    "you may not use this file except in compliance with the License.\n",
    "You may obtain a copy of the License at\n",
    "\n",
    "https://llvm.org/LICENSE.txt\n",
    "\n",
    "Unless required by applicable law or agreed to in writing, software\n",
    "distributed under the License is distributed on an \"AS IS\" BASIS,\n",
    "WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.\n",
    "See the License for the specific language governing permissions and\n",
    "limitations under the License."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "import requests\n",
    "import os\n",
    "import logging\n",
    "import argparse\n",
    "import json\n",
    "import pandas as pd\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "cache_file = \"cache.json\"\n",
    "token = f'Bearer {os.getenv(\"BUILDKITE_API_TOKEN\")}'\n",
    "builds = []"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "if os.path.exists(cache_file):\n",
    "    with open(cache_file) as f:\n",
    "        builds = json.load(f)\n",
    "        print(f'loaded {len(builds)} entries') "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# load new jobs from Buildkite API\n",
    "if True:\n",
    "    existing = set()\n",
    "    for j in builds:\n",
    "        existing.add(j['id'])\n",
    "\n",
    "    # uncomment to reset\n",
    "#     builds = []\n",
    "#     existing = set()\n",
    "    page = 1\n",
    "    stop = False\n",
    "    while not stop:\n",
    "        print('loading page', page)\n",
    "        re = requests.get('https://api.buildkite.com/v2/organizations/llvm-project/builds',\n",
    "                          params={'page': page},\n",
    "                          headers={'Authorization': token})\n",
    "        if re.status_code != 200:\n",
    "            print('response status', re.status_code, re)\n",
    "            break\n",
    "        x = re.json()\n",
    "        if x == []:\n",
    "            print('empty response')\n",
    "            break\n",
    "        for j in x:\n",
    "            if j['id'] in existing:\n",
    "                print('found existing job', j['id'])\n",
    "                stop = True\n",
    "                break\n",
    "            # TODO: skip running jobs            \n",
    "            if (j['state'] == 'running') or (j['state'] == 'scheduled'):\n",
    "                print(j['web_url'], 'is', j['state'], ', skipping')\n",
    "                continue\n",
    "            builds.append(j)\n",
    "        page += 1\n",
    "    print(len(builds), 'jobs in total')    \n",
    "    with open(cache_file, 'w') as f:\n",
    "        json.dump(builds, f)\n",
    "    print(f'saved {len(builds)} entries')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "d = {\n",
    "    'id': [],\n",
    "    'number': [],\n",
    "    'pipeline': [],\n",
    "}\n",
    "\n",
    "jobs = {\n",
    "    'pipeline': [],\n",
    "    'name': [],\n",
    "    'step_key': [],\n",
    "    'state': [],\n",
    "    'exit_status': [],\n",
    "    'agent_id': [],\n",
    "    \n",
    "    'agent_name': [],\n",
    "    'runnable_at': [],\n",
    "    'started_at': [],\n",
    "    'wait_duration': [],\n",
    "    'finished_at': [],\n",
    "    'run_duration': [],\n",
    "}\n",
    "\n",
    "sec = np.timedelta64(1, 's')\n",
    "for b in builds:\n",
    "    d['id'].append(b['id'])\n",
    "    d['number'].append(b['number'])\n",
    "    d['pipeline'].append(b['pipeline']['slug'])\n",
    "    for x in b['jobs']:\n",
    "        if x['state'] in ['waiting_failed', 'canceled', 'skipped', 'broken']:\n",
    "            continue\n",
    "        try:\n",
    "            jobs['pipeline'].append(b['pipeline']['slug'])\n",
    "            jobs['name'].append(x['name'])\n",
    "            jobs['step_key'].append(x['step_key'] if 'step_key' in x else '')\n",
    "            jobs['state'].append(x['state'] )\n",
    "            jobs['exit_status'].append(x['exit_status'] if 'exit_status' in x else -1)\n",
    "            jobs['agent_id'].append(x['agent']['id'] if 'agent' in x else '')\n",
    "            jobs['agent_name'].append(x['agent']['name'] if 'agent' in x else '')\n",
    "            runnable = np.datetime64(x['runnable_at'].replace('Z', ''))\n",
    "            started = np.datetime64(x['started_at'].replace('Z', ''))\n",
    "            finished = np.datetime64(x['finished_at'].replace('Z', ''))\n",
    "            jobs['runnable_at'].append(runnable)\n",
    "            jobs['started_at'].append(started)\n",
    "            jobs['wait_duration'].append((started - runnable) / sec)\n",
    "            jobs['finished_at'].append(finished)\n",
    "            jobs['run_duration'].append((finished - started) / sec)\n",
    "        except Exception as e:\n",
    "            print(x)\n",
    "            raise e            \n",
    "jobs = pd.DataFrame(jobs)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "jobs.pipeline.unique()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "jobs"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "ds = jobs[jobs['pipeline'] == 'premerge-checks'][jobs['step_key'] == 'windows'][jobs['state']=='passed'][~jobs['agent_name'].str.startswith('buildkite-')]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "ds"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "  fig, ax = plt.subplots(figsize=(10,10)) # size of the plot (width, height)\n",
    "\n",
    "  sns.violinplot(\n",
    "        ax=ax,\n",
    "        x='run_duration',\n",
    "        y='agent_name',\n",
    "        split=True,\n",
    "        data=ds)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "t = pd.pivot_table(ds, values=['run_duration'], index=['agent_name'],\n",
    "                    aggfunc=[np.median, np.mean, np.std, np.count_nonzero])\n",
    "t = t.reindex(t.sort_values(by=('median', 'run_duration'), ascending=False).index)\n",
    "t"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.5rc1"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}