git.sesse.net Git - remoteglot/blob - server/serve-analysis.js

   1 // node.js version of analysis.pl; hopefully scales a bit better
   2 // for this specific kind of task.
   3
   4 // Modules.
   5 var http = require('http');
   6 var fs = require('fs');
   7 var url = require('url');
   8 var querystring = require('querystring');
   9 var path = require('path');
  10 var zlib = require('zlib');
  11 var readline = require('readline');
  12 var child_process = require('child_process');
  13 var delta = require('../www/js/json_delta.js');
  14 var hash_lookup = require('./hash-lookup.js');
  15
  16 // Constants.
  17 var HISTORY_TO_KEEP = 5;
  18 var MINIMUM_VERSION = null;
  19 var COUNT_FROM_VARNISH_LOG = true;
  20
  21 // Filename to serve.
  22 var json_filename = '/srv/analysis.sesse.net/www/analysis.json';
  23 if (process.argv.length >= 3) {
  24         json_filename = process.argv[2];
  25 }
  26
  27 // Expected destination filenames.
  28 var serve_url = '/analysis.pl';
  29 var hash_serve_url = '/hash';
  30 if (process.argv.length >= 4) {
  31         serve_url = process.argv[3];
  32 }
  33 if (process.argv.length >= 5) {
  34         hash_serve_url = process.argv[4];
  35 }
  36
  37 // TCP port to listen on.
  38 var port = 5000;
  39 if (process.argv.length >= 6) {
  40         port = parseInt(process.argv[5]);
  41 }
  42
  43 // gRPC backends.
  44 var grpc_backends = ["localhost:50051", "localhost:50052"];
  45 if (process.argv.length >= 7) {
  46         grpc_backends = process.argv[6].split(",");
  47 }
  48 hash_lookup.init(grpc_backends);
  49
  50 // If set to 1, we are already processing a JSON update and should not
  51 // start a new one. If set to 2, we are _also_ having one in the queue.
  52 var json_lock = 0;
  53
  54 // The current contents of the file to hand out, and its last modified time.
  55 var json = undefined;
  56
  57 // The last five timestamps, and diffs from them to the latest version.
  58 var historic_json = [];
  59 var diff_json = {};
  60
  61 // The list of clients that are waiting for new data to show up.
  62 // Uniquely keyed by request_id so that we can take them out of
  63 // the queue if they close the socket.
  64 var sleeping_clients = {};
  65 var request_id = 0;
  66
  67 // List of when clients were last seen, keyed by their unique ID.
  68 // Used to show a viewer count to the user.
  69 var last_seen_clients = {};
  70
  71 // The timer used to touch the file every 30 seconds if nobody
  72 // else does it for us. This makes sure we don't have clients
  73 // hanging indefinitely (which might have them return errors).
  74 var touch_timer = undefined;
  75
  76 // If we are behind Varnish, we can't count the number of clients
  77 // ourselves, so we need to get it from parsing varnishncsa.
  78 var viewer_count_override = undefined;
  79
  80 var replace_json = function(new_json_contents, mtime) {
  81         // Generate the list of diffs from the last five versions.
  82         if (json !== undefined) {
  83                 // If two versions have the same mtime, clients could have either.
  84                 // Note the fact, so that we never insert it.
  85                 if (json.last_modified == mtime) {
  86                         json.invalid_base = true;
  87                 }
  88                 if (!json.invalid_base) {
  89                         historic_json.push(json);
  90                         if (historic_json.length > HISTORY_TO_KEEP) {
  91                                 historic_json.shift();
  92                         }
  93                 }
  94         }
  95
  96         var parsed = JSON.parse(new_json_contents);
  97
  98         if (parsed['internal']) {
  99                 if (parsed['internal']['grpc_backends'] &&
 100                     hash_lookup.need_reinit(parsed['internal']['grpc_backends'])) {
 101                         hash_lookup.init(parsed['internal']['grpc_backends']);
 102                 }
 103                 delete parsed['internal'];
 104                 new_json_contents = JSON.stringify(parsed);
 105         }
 106
 107         var new_json = {
 108                 parsed: parsed,
 109                 plain: new_json_contents,
 110                 last_modified: mtime
 111         };
 112         create_json_historic_diff(new_json, historic_json.slice(0), {}, function(new_diff_json) {
 113                 // gzip the new version (non-delta), and put it into place.
 114                 zlib.gzip(new_json_contents, function(err, buffer) {
 115                         if (err) throw err;
 116
 117                         new_json.gzip = buffer;
 118                         json = new_json;
 119                         diff_json = new_diff_json;
 120                         json_lock = 0;
 121
 122                         // Finally, wake up any sleeping clients.
 123                         possibly_wakeup_clients();
 124                 });
 125         });
 126 }
 127
 128 var create_json_historic_diff = function(new_json, history_left, new_diff_json, cb) {
 129         if (history_left.length == 0) {
 130                 cb(new_diff_json);
 131                 return;
 132         }
 133
 134         var histobj = history_left.shift();
 135         var diff = delta.JSON_delta.diff(histobj.parsed, new_json.parsed);
 136         var diff_text = JSON.stringify(diff);
 137
 138         // Verify that the delta is correct
 139         var base = JSON.parse(histobj.plain);
 140         delta.JSON_delta.patch(base, diff);
 141         var correct_pv = JSON.stringify(base['pv']);
 142         var wrong_pv = JSON.stringify(new_json.parsed['pv']);
 143         if (correct_pv !== wrong_pv) {
 144                 console.log("Patch went wrong:", histobj.plain, new_json.plain);
 145                 exit();
 146         }
 147
 148         zlib.gzip(diff_text, function(err, buffer) {
 149                 if (err) throw err;
 150                 new_diff_json[histobj.last_modified] = {
 151                         parsed: diff,
 152                         plain: diff_text,
 153                         gzip: buffer,
 154                         last_modified: new_json.last_modified,
 155                 };
 156                 create_json_historic_diff(new_json, history_left, new_diff_json, cb);
 157         });
 158 }
 159
 160 function read_entire_file(filename, callback) {
 161         fs.open(filename, 'r', function(err, fd) {
 162                 if (err) throw err;
 163                 fs.fstat(fd, function(err, st) {
 164                         if (err) throw err;
 165                         var buffer = new Buffer(1048576);
 166                         fs.read(fd, buffer, 0, 1048576, 0, function(err, bytesRead, buffer) {
 167                                 if (err) throw err;
 168                                 fs.close(fd, function() {
 169                                         var contents = buffer.toString('utf8', 0, bytesRead);
 170                                         callback(contents, st.mtime.getTime());
 171                                 });
 172                         });
 173                 });
 174         });
 175 }
 176
 177 var reread_file = function(event, filename) {
 178         if (filename != path.basename(json_filename)) {
 179                 return;
 180         }
 181         if (json_lock >= 2) {
 182                 return;
 183         }
 184         if (json_lock == 1) {
 185                 // Already processing; wait a bit.
 186                 json_lock = 2;
 187                 setTimeout(function() { if (json_lock == 2) json_lock = 1; reread_file(event, filename); }, 100);
 188                 return;
 189         }
 190         json_lock = 1;
 191
 192         console.log("Rereading " + json_filename);
 193         read_entire_file(json_filename, function(new_json_contents, mtime) {
 194                 replace_json(new_json_contents, mtime);
 195         });
 196
 197         if (touch_timer !== undefined) {
 198                 clearTimeout(touch_timer);
 199         }
 200         touch_timer = setTimeout(function() {
 201                 console.log("Touching analysis.json due to no other activity");
 202                 var now = Date.now() / 1000;
 203                 fs.utimes(json_filename, now, now, function() {});
 204         }, 30000);
 205 }
 206 var possibly_wakeup_clients = function() {
 207         var num_viewers = count_viewers();
 208         for (var i in sleeping_clients) {
 209                 mark_recently_seen(sleeping_clients[i].unique);
 210                 send_json(sleeping_clients[i].response,
 211                           sleeping_clients[i].ims,
 212                           sleeping_clients[i].accept_gzip,
 213                           num_viewers);
 214         }
 215         sleeping_clients = {};
 216 }
 217 var send_404 = function(response) {
 218         response.writeHead(404, {
 219                 'Content-Type': 'text/plain',
 220         });
 221         response.write('Something went wrong. Sorry.');
 222         response.end();
 223 }
 224 var send_json = function(response, ims, accept_gzip, num_viewers) {
 225         var this_json = diff_json[ims] || json;
 226
 227         var headers = {
 228                 'Content-Type': 'text/json',
 229                 'X-RGLM': this_json.last_modified,
 230                 'X-RGNV': num_viewers,
 231                 'Access-Control-Expose-Headers': 'X-RGLM, X-RGNV, X-RGMV',
 232                 'Vary': 'Accept-Encoding',
 233         };
 234
 235         if (MINIMUM_VERSION) {
 236                 headers['X-RGMV'] = MINIMUM_VERSION;
 237         }
 238
 239         if (accept_gzip) {
 240                 headers['Content-Length'] = this_json.gzip.length;
 241                 headers['Content-Encoding'] = 'gzip';
 242                 response.writeHead(200, headers);
 243                 response.write(this_json.gzip);
 244         } else {
 245                 headers['Content-Length'] = this_json.plain.length;
 246                 response.writeHead(200, headers);
 247                 response.write(this_json.plain);
 248         }
 249         response.end();
 250 }
 251 var mark_recently_seen = function(unique) {
 252         if (unique) {
 253                 last_seen_clients[unique] = (new Date).getTime();
 254         }
 255 }
 256 var count_viewers = function() {
 257         if (viewer_count_override !== undefined) {
 258                 return viewer_count_override;
 259         }
 260
 261         var now = (new Date).getTime();
 262
 263         // Go through and remove old viewers, and count them at the same time.
 264         var new_last_seen_clients = {};
 265         var num_viewers = 0;
 266         for (var unique in last_seen_clients) {
 267                 if (now - last_seen_clients[unique] < 5000) {
 268                         ++num_viewers;
 269                         new_last_seen_clients[unique] = last_seen_clients[unique];
 270                 }
 271         }
 272
 273         // Also add sleeping clients that we would otherwise assume timed out.
 274         for (var request_id in sleeping_clients) {
 275                 var unique = sleeping_clients[request_id].unique;
 276                 if (unique && !(unique in new_last_seen_clients)) {
 277                         ++num_viewers;
 278                 }
 279         }
 280
 281         last_seen_clients = new_last_seen_clients;
 282         return num_viewers;
 283 }
 284 var log = function(str) {
 285         console.log("[" + ((new Date).getTime()*1e-3).toFixed(3) + "] " + str);
 286 }
 287
 288 // Set up a watcher to catch changes to the file, then do an initial read
 289 // to make sure we have a copy.
 290 fs.watch(path.dirname(json_filename), reread_file);
 291 reread_file(null, path.basename(json_filename));
 292
 293 if (COUNT_FROM_VARNISH_LOG) {
 294         // Note: We abuse serve_url as a regex.
 295         var varnishncsa = child_process.spawn(
 296                 'varnishncsa', ['-F', '%{%s}t %U %q tffb=%{Varnish:time_firstbyte}x',
 297                 '-q', 'ReqURL ~ "^' + serve_url + '"']);
 298         var rl = readline.createInterface({
 299                 input: varnishncsa.stdout,
 300                 output: varnishncsa.stdin,
 301                 terminal: false
 302         });
 303
 304         var uniques = [];
 305         rl.on('line', function(line) {
 306                 var v = line.match(/(\d+) .*\?ims=\d+&unique=(.*) tffb=(.*)/);
 307                 if (v) {
 308                         uniques[v[2]] = {
 309                                 last_seen: (parseInt(v[1]) + parseFloat(v[3])) * 1e3,
 310                                 grace: null,
 311                         };
 312                         log(v[1] + " " + v[2] + " " + v[3]);
 313                 } else {
 314                         log("VARNISHNCSA UNPARSEABLE LINE: " + line);
 315                 }
 316         });
 317         setInterval(function() {
 318                 var mtime = json.last_modified - 1000;  // Compensate for subsecond issues.
 319                 var now = (new Date).getTime();
 320                 var num_viewers = 0;
 321
 322                 for (var unique in uniques) {
 323                         ++num_viewers;
 324                         var last_seen = uniques[unique].last_seen;
 325                         if (now - last_seen <= 5000) {
 326                                 // We've seen this user in the last five seconds;
 327                                 // it's okay.
 328                                 continue;
 329                         }
 330                         if (last_seen >= mtime) {
 331                                 // This user has the latest version;
 332                                 // they are probably just hanging.
 333                                 continue;
 334                         }
 335                         if (uniques[unique].grace === null) {
 336                                 // They have five seconds after a new JSON has been
 337                                 // provided to get get it, or they're out.
 338                                 // We don't simply use mtime, since we don't want to
 339                                 // reset the grace timer just because a new JSON is
 340                                 // published.
 341                                 uniques[unique].grace = mtime;
 342                         }
 343                         if (now - uniques[unique].grace > 5000) {
 344                                 log("Timing out " + unique + " (last_seen=" + last_seen + ", now=" + now +
 345                                         ", mtime=" + mtime, ", grace=" + uniques[unique].grace + ")");
 346                                 delete uniques[unique];
 347                                 --num_viewers;
 348                         }
 349                 }
 350
 351                 log(num_viewers + " entries in hash, mtime=" + mtime);
 352                 viewer_count_override = num_viewers;
 353         }, 1000);
 354 }
 355
 356 var server = http.createServer();
 357 server.on('request', function(request, response) {
 358         var u = url.parse(request.url, true);
 359         var ims = (u.query)['ims'];
 360         var unique = (u.query)['unique'];
 361
 362         log(request.url);
 363         if (u.pathname === hash_serve_url) {
 364                 var fen = (u.query)['fen'];
 365                 hash_lookup.handle_request(fen, response);
 366                 return;
 367         }
 368         if (u.pathname !== serve_url) {
 369                 // This is not the request you are looking for.
 370                 send_404(response);
 371                 return;
 372         }
 373
 374         mark_recently_seen(unique);
 375
 376         var accept_encoding = request.headers['accept-encoding'];
 377         var accept_gzip;
 378         if (accept_encoding !== undefined && accept_encoding.match(/\bgzip\b/)) {
 379                 accept_gzip = true;
 380         } else {
 381                 accept_gzip = false;
 382         }
 383
 384         // If we already have something newer than what the user has,
 385         // just send it out and be done with it.
 386         if (json !== undefined && (!ims || json.last_modified > ims)) {
 387                 send_json(response, ims, accept_gzip, count_viewers());
 388                 return;
 389         }
 390
 391         // OK, so we need to hang until we have something newer.
 392         // Put the user on the wait list.
 393         var client = {};
 394         client.response = response;
 395         client.request_id = request_id;
 396         client.accept_gzip = accept_gzip;
 397         client.unique = unique;
 398         client.ims = ims;
 399         sleeping_clients[request_id++] = client;
 400
 401         request.socket.client = client;
 402 });
 403 server.on('connection', function(socket) {
 404         socket.on('close', function() {
 405                 var client = socket.client;
 406                 if (client) {
 407                         mark_recently_seen(client.unique);
 408                         delete sleeping_clients[client.request_id];
 409                 }
 410         });
 411 });
 412
 413 server.listen(port);