tools/datalab/ratelimiting.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Analyzing Rate Limit Exceeded events\n",
    "\n",
    "Use this notebook to dig into Rate Limit Exceeded events on Monorail."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "import gcp\n",
    "import gcp.bigquery as bq\n",
    "\n",
    "context = gcp.Context.default()\n",
    "print 'The current project is %s' % context.project_id\n",
    "\n",
    "# Set the date to analyze here:\n",
    "date = 20160514"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "%%sql --module by_ip\n",
    "SELECT\n",
    "  protoPayload.ip as ip,\n",
    "  COUNT(protoPayload.requestId) AS num\n",
    "FROM\n",
    "  [logs.appengine_googleapis_com_request_log_$date]\n",
    "WHERE\n",
    "  protoPayload.moduleId is null # == \"default\", otherwise you get backend queries too.\n",
    "  AND\n",
    "  protoPayload.line.logMessage LIKE \"Rate Limit Exceeded%\"\n",
    "GROUP BY\n",
    "  ip\n",
    "ORDER BY\n",
    "  num DESC\n",
    "LIMIT\n",
    "  100;"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "%%sql --module by_ip_class\n",
    "SELECT\n",
    "  REGEXP_EXTRACT(protoPayload.ip,r'^(?:[^\\.]*\\.){0}([^\\.]*)\\.?') AS a,\n",
    "  REGEXP_EXTRACT(protoPayload.ip,r'^(?:[^\\.]*\\.){1}([^\\.]*)\\.?') AS b,\n",
    "  REGEXP_EXTRACT(protoPayload.ip,r'^(?:[^\\.]*\\.){2}([^\\.]*)\\.?') AS c,\n",
    "  REGEXP_EXTRACT(protoPayload.ip,r'^(?:[^\\.]*\\.){3}([^\\.]*)\\.?') AS d,\n",
    "  COUNT(protoPayload.requestId) AS num\n",
    "FROM\n",
    "  [logs.appengine_googleapis_com_request_log_$date]\n",
    "WHERE\n",
    "  protoPayload.moduleId is null # == \"default\", otherwise you get backend queries too.\n",
    "  AND\n",
    "  protoPayload.line.logMessage LIKE \"Rate Limit Exceeded%\"\n",
    "GROUP BY\n",
    "  a,\n",
    "  b,\n",
    "  c,\n",
    "  d\n",
    "ORDER BY\n",
    "  num DESC\n",
    "LIMIT\n",
    "  100;"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "%%sql --module by_country\n",
    "SELECT\n",
    "  protoPayload.line.logMessage as line,\n",
    "  COUNT(DISTINCT protoPayload.ip) as ip_count,\n",
    "  COUNT(protoPayload.requestId) AS req_count\n",
    "FROM\n",
    "  FLATTEN ([logs.appengine_googleapis_com_request_log_$date], protoPayload.line)\n",
    "WHERE\n",
    "  protoPayload.moduleId is null # == \"default\", otherwise you get backend queries too.\n",
    "  AND\n",
    "  protoPayload.line.logMessage LIKE \"Rate Limit Exceeded%\"\n",
    "  AND\n",
    "  REGEXP_MATCH(protoPayload.line.logMessage, 'X-AppEngine-Country')\n",
    "GROUP BY\n",
    "  line\n",
    "ORDER BY\n",
    "  req_count DESC\n",
    "LIMIT\n",
    "  100;"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "%%sql --module by_resource\n",
    "SELECT\n",
    "  protoPayload.resource as resource,\n",
    "  COUNT(protoPayload.requestId) AS req_count\n",
    "FROM\n",
    "  [logs.appengine_googleapis_com_request_log_$date]\n",
    "WHERE\n",
    "  protoPayload.moduleId is null # == \"default\", otherwise you get backend queries too.\n",
    "  AND\n",
    "  protoPayload.line.logMessage LIKE \"Rate Limit Exceeded%\"\n",
    "GROUP BY\n",
    "  resource\n",
    "ORDER BY\n",
    "  req_count DESC\n",
    "LIMIT\n",
    "  100;"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Requests by IP"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "df = bq.Query(by_ip, date=date).to_dataframe()\n",
    "df.head(20)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "if len(df):\n",
    "  df.plot()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Requests by IP Class"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "df = bq.Query(by_ip_class,  date=date).to_dataframe()\n",
    "df.head(20)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "if len(df):\n",
    "  df.plot()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Requests by Country Code"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "df = bq.Query(by_country, date=date).to_dataframe()\n",
    "df.head(20)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "if len(df):\n",
    "  df.plot()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Requests by Requested Resource"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "df = bq.Query(by_resource, date=date).to_dataframe()\n",
    "df.head(20)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "if len(df):\n",
    "  df.plot()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 2",
   "language": "python",
   "name": "python2"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 2
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython2",
   "version": "2.7.9"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 0
}