src/test_cache_digest.cc

   1
   2 /*
   3  * $Id: test_cache_digest.cc,v 1.17 1998/04/06 22:32:22 wessels Exp $
   4  *
   5  * AUTHOR: Alex Rousskov
   6  *
   7  * SQUID Internet Object Cache  http://squid.nlanr.net/Squid/
   8  * --------------------------------------------------------
   9  *
  10  *  Squid is the result of efforts by numerous individuals from the
  11  *  Internet community.  Development is led by Duane Wessels of the
  12  *  National Laboratory for Applied Network Research and funded by
  13  *  the National Science Foundation.
  14  *
  15  *  This program is free software; you can redistribute it and/or modify
  16  *  it under the terms of the GNU General Public License as published by
  17  *  the Free Software Foundation; either version 2 of the License, or
  18  *  (at your option) any later version.
  19  *
  20  *  This program is distributed in the hope that it will be useful,
  21  *  but WITHOUT ANY WARRANTY; without even the implied warranty of
  22  *  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  23  *  GNU General Public License for more details.
  24  *
  25  *  You should have received a copy of the GNU General Public License
  26  *  along with this program; if not, write to the Free Software
  27  *  Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
  28  *
  29  */
  30
  31 /*
  32  * Test-suite for playing with cache digests
  33  */
  34
  35 #include "squid.h"
  36
  37 typedef struct {
  38     int query_count;
  39     int true_hit_count;
  40     int true_miss_count;
  41     int false_hit_count;
  42     int false_miss_count;
  43 } CacheQueryStats;
  44
  45 typedef struct _Cache Cache;
  46 struct _Cache {
  47     const char *name;
  48     hash_table *hash;
  49     CacheDigest *digest;
  50     Cache *peer;
  51     CacheQueryStats qstats;
  52     int count;                  /* #currently cached entries */
  53     int req_count;              /* #requests to this cache */
  54     int bad_add_count;          /* #duplicate adds */
  55     int bad_del_count;          /* #dels with no prior add */
  56 };
  57
  58
  59 typedef struct _CacheEntry {
  60     const cache_key *key;
  61     struct _CacheEntry *next;
  62     unsigned char key_arr[MD5_DIGEST_CHARS];
  63     /* storeSwapLogData s; */
  64 } CacheEntry;
  65
  66 /* parsed access log entry */
  67 typedef struct {
  68     cache_key key[MD5_DIGEST_CHARS];
  69     time_t timestamp;
  70     short int use_icp;          /* true/false */
  71 } RawAccessLogEntry;
  72
  73 typedef enum {
  74     frError = -2, frMore = -1, frEof = 0, frOk = 1
  75 } fr_result;
  76 typedef struct _FileIterator FileIterator;
  77 typedef fr_result(*FI_READER) (FileIterator * fi);
  78
  79 struct _FileIterator {
  80     const char *fname;
  81     FILE *file;
  82     time_t inner_time;          /* timestamp of the current entry */
  83     time_t time_offset;         /* to adjust time set by reader */
  84     int line_count;             /* number of lines scanned */
  85     int bad_line_count;         /* number of parsing errors */
  86     int time_warp_count;        /* number of out-of-order entries in the file */
  87     FI_READER reader;           /* reads next entry and updates inner_time */
  88     void *entry;                /* buffer for the current entry, freed with xfree() */
  89 };
  90
  91 /* globals */
  92 static time_t cur_time = -1;    /* timestamp of the current log entry */
  93
  94 #if 0
  95
  96 static int cacheIndexScanCleanPrefix(CacheIndex * idx, const char *fname, FILE * file);
  97 static int cacheIndexScanAccessLog(CacheIndex * idx, const char *fname, FILE * file);
  98
  99 #endif
 100
 101 /* copied from url.c */
 102 const char *RequestMethodStr[] =
 103 {
 104     "NONE",
 105     "GET",
 106     "POST",
 107     "PUT",
 108     "HEAD",
 109     "CONNECT",
 110     "TRACE",
 111     "PURGE"
 112 };
 113 /* copied from url.c */
 114 static method_t
 115 methodStrToId(const char *s)
 116 {
 117     if (strcasecmp(s, "GET") == 0) {
 118         return METHOD_GET;
 119     } else if (strcasecmp(s, "POST") == 0) {
 120         return METHOD_POST;
 121     } else if (strcasecmp(s, "PUT") == 0) {
 122         return METHOD_PUT;
 123     } else if (strcasecmp(s, "HEAD") == 0) {
 124         return METHOD_HEAD;
 125     } else if (strcasecmp(s, "CONNECT") == 0) {
 126         return METHOD_CONNECT;
 127     } else if (strcasecmp(s, "TRACE") == 0) {
 128         return METHOD_TRACE;
 129     } else if (strcasecmp(s, "PURGE") == 0) {
 130         return METHOD_PURGE;
 131     }
 132     return METHOD_NONE;
 133 }
 134
 135 /* FileIterator */
 136
 137 static void fileIteratorAdvance(FileIterator * fi);
 138
 139 static FileIterator *
 140 fileIteratorCreate(const char *fname, FI_READER reader)
 141 {
 142     FileIterator *fi = xcalloc(1, sizeof(FileIterator));
 143     assert(fname && reader);
 144     fi->fname = fname;
 145     fi->reader = reader;
 146     fi->file = fopen(fname, "r");
 147     if (!fi->file) {
 148         fprintf(stderr, "cannot open %s: %s\n", fname, strerror(errno));
 149         return NULL;
 150     } else
 151         fprintf(stderr, "opened %s\n", fname);
 152     fileIteratorAdvance(fi);
 153     return fi;
 154 }
 155
 156 static void
 157 fileIteratorDestroy(FileIterator * fi)
 158 {
 159     assert(fi);
 160     if (fi->file) {
 161         fclose(fi->file);
 162         fprintf(stderr, "closed %s\n", fi->fname);
 163     }
 164     xfree(fi->entry);
 165     xfree(fi);
 166 }
 167
 168 static void
 169 fileIteratorSetCurTime(FileIterator * fi, time_t ct)
 170 {
 171     assert(fi);
 172     assert(fi->inner_time > 0);
 173     fi->time_offset = ct - fi->inner_time;
 174 }
 175
 176 static void
 177 fileIteratorAdvance(FileIterator * fi)
 178 {
 179     int res;
 180     assert(fi);
 181     do {
 182         const time_t last_time = fi->inner_time;
 183         fi->inner_time = -1;
 184         res = fi->reader(fi);
 185         fi->line_count++;
 186         if (fi->inner_time < 0)
 187             fi->inner_time = last_time;
 188         else
 189             fi->inner_time += fi->time_offset;
 190         if (res == frError)
 191             fi->bad_line_count++;
 192         else if (res == frEof) {
 193             fprintf(stderr, "exhausted %s (%d entries) at %s",
 194                 fi->fname, fi->line_count, ctime(&fi->inner_time));
 195             fi->inner_time = -1;
 196         } else if (fi->inner_time < last_time) {
 197             assert(last_time >= 0);
 198             fi->time_warp_count++;
 199             fi->inner_time = last_time;
 200         }
 201         /* report progress */
 202         if (!(fi->line_count % 50000))
 203             fprintf(stderr, "%s scanned %d K entries (%d bad) at %s",
 204                 fi->fname, fi->line_count / 1000, fi->bad_line_count,
 205                 ctime(&fi->inner_time));
 206     } while (res < 0);
 207 }
 208
 209 /* CacheEntry */
 210
 211 static CacheEntry *
 212 cacheEntryCreate(const storeSwapLogData * s)
 213 {
 214     CacheEntry *e = xcalloc(1, sizeof(CacheEntry));
 215     assert(s);
 216     /* e->s = *s; */
 217     xmemcpy(e->key_arr, s->key, MD5_DIGEST_CHARS);
 218     e->key = &e->key_arr[0];
 219     return e;
 220 }
 221
 222 static void
 223 cacheEntryDestroy(CacheEntry * e)
 224 {
 225     assert(e);
 226     xfree(e);
 227 }
 228
 229
 230 /* Cache */
 231
 232 static Cache *
 233 cacheCreate(const char *name)
 234 {
 235     Cache *c;
 236     assert(name && strlen(name));
 237     c = xcalloc(1, sizeof(Cache));
 238     c->name = name;
 239     c->hash = hash_create(storeKeyHashCmp, 2e6, storeKeyHashHash);
 240     return c;
 241 }
 242
 243 static void
 244 cacheDestroy(Cache * cache)
 245 {
 246     CacheEntry *e = NULL;
 247     hash_table *hash;
 248     assert(cache);
 249     hash = cache->hash;
 250     /* destroy hash table contents */
 251     for (e = hash_first(hash); e; e = hash_next(hash)) {
 252         hash_remove_link(hash, (hash_link *) e);
 253         cacheEntryDestroy(e);
 254     }
 255     /* destroy the hash table itself */
 256     hashFreeMemory(hash);
 257     if (cache->digest)
 258         cacheDigestDestroy(cache->digest);
 259     xfree(cache);
 260 }
 261
 262 /* re-digests currently hashed entries */
 263 static void
 264 cacheResetDigest(Cache * cache)
 265 {
 266     CacheEntry *e = NULL;
 267     hash_table *hash;
 268     struct timeval t_start, t_end;
 269
 270     assert(cache);
 271     fprintf(stderr, "%s: init-ing digest with %d entries\n", cache->name, cache->count);
 272     if (cache->digest)
 273         cacheDigestDestroy(cache->digest);
 274     hash = cache->hash;
 275     cache->digest = cacheDigestCreate(2 * cache->count + 1);    /* 50% utilization */
 276     if (!cache->count)
 277         return;
 278     gettimeofday(&t_start, NULL);
 279     for (e = hash_first(hash); e; e = hash_next(hash)) {
 280         cacheDigestAdd(cache->digest, e->key);
 281     }
 282     gettimeofday(&t_end, NULL);
 283     assert(cache->digest->count == cache->count);
 284     fprintf(stderr, "%s: init-ed  digest with %d entries\n",
 285         cache->name, cache->digest->count);
 286     fprintf(stderr, "%s: init took: %f sec, %f sec/M\n",
 287         cache->name,
 288         tvSubDsec(t_start, t_end),
 289         (double) 1e6 * tvSubDsec(t_start, t_end) / cache->count);
 290     /* check how long it takes to traverse the hash */
 291     gettimeofday(&t_start, NULL);
 292     for (e = hash_first(hash); e; e = hash_next(hash)) {
 293     }
 294     gettimeofday(&t_end, NULL);
 295     fprintf(stderr, "%s: hash scan took: %f sec, %f sec/M\n",
 296         cache->name,
 297         tvSubDsec(t_start, t_end),
 298         (double) 1e6 * tvSubDsec(t_start, t_end) / cache->count);
 299 }
 300
 301 static void
 302 cacheQueryPeer(Cache * cache, const cache_key * key)
 303 {
 304     const int peer_has_it = hash_lookup(cache->peer->hash, key) != NULL;
 305     const int we_think_we_have_it = cacheDigestTest(cache->digest, key);
 306
 307     cache->qstats.query_count++;
 308     if (peer_has_it) {
 309         if (we_think_we_have_it)
 310             cache->qstats.true_hit_count++;
 311         else
 312             cache->qstats.false_miss_count++;
 313     } else {
 314         if (we_think_we_have_it)
 315             cache->qstats.false_hit_count++;
 316         else
 317             cache->qstats.true_miss_count++;
 318     }
 319 }
 320
 321 static void
 322 cacheQueryReport(Cache * cache, CacheQueryStats * stats)
 323 {
 324     fprintf(stdout, "%s: peer queries: %d (%d%%)\n",
 325         cache->name,
 326         stats->query_count, xpercentInt(stats->query_count, cache->req_count)
 327         );
 328     fprintf(stdout, "%s: t-hit: %d (%d%%) t-miss: %d (%d%%) t-*: %d (%d%%)\n",
 329         cache->name,
 330         stats->true_hit_count, xpercentInt(stats->true_hit_count, stats->query_count),
 331         stats->true_miss_count, xpercentInt(stats->true_miss_count, stats->query_count),
 332         stats->true_hit_count + stats->true_miss_count,
 333         xpercentInt(stats->true_hit_count + stats->true_miss_count, stats->query_count)
 334         );
 335     fprintf(stdout, "%s: f-hit: %d (%d%%) f-miss: %d (%d%%) f-*: %d (%d%%)\n",
 336         cache->name,
 337         stats->false_hit_count, xpercentInt(stats->false_hit_count, stats->query_count),
 338         stats->false_miss_count, xpercentInt(stats->false_miss_count, stats->query_count),
 339         stats->false_hit_count + stats->false_miss_count,
 340         xpercentInt(stats->false_hit_count + stats->false_miss_count, stats->query_count)
 341         );
 342 }
 343
 344 static void
 345 cacheReport(Cache * cache)
 346 {
 347     fprintf(stdout, "%s: entries: %d reqs: %d bad-add: %d bad-del: %d\n",
 348         cache->name, cache->count, cache->req_count,
 349         cache->bad_add_count, cache->bad_del_count);
 350
 351 }
 352
 353 static void
 354 cacheFetch(Cache * cache, const RawAccessLogEntry * e)
 355 {
 356     assert(e);
 357     cache->req_count++;
 358     if (e->use_icp)
 359         cacheQueryPeer(cache, e->key);
 360 }
 361
 362 static fr_result
 363 swapStateReader(FileIterator * fi)
 364 {
 365     storeSwapLogData *entry;
 366     if (!fi->entry)
 367         fi->entry = xcalloc(1, sizeof(storeSwapLogData));
 368     entry = fi->entry;
 369     if (fread(entry, sizeof(*entry), 1, fi->file) != 1)
 370         return frEof;
 371     fi->inner_time = entry->lastref;
 372     if (entry->op != SWAP_LOG_ADD && entry->op != SWAP_LOG_DEL) {
 373         fprintf(stderr, "%s:%d: unknown swap log action\n", fi->fname, fi->line_count);
 374         exit(-3);
 375     }
 376     return frOk;
 377 }
 378
 379 static fr_result
 380 accessLogReader(FileIterator * fi)
 381 {
 382     static char buf[4096];
 383     RawAccessLogEntry *entry;
 384     char *url;
 385     char *method;
 386     int method_id = METHOD_NONE;
 387     char *hier = NULL;
 388
 389     assert(fi);
 390     if (!fi->entry)
 391         fi->entry = xcalloc(1, sizeof(RawAccessLogEntry));
 392     else
 393         memset(fi->entry, 0, sizeof(RawAccessLogEntry));
 394     entry = fi->entry;
 395     if (!fgets(buf, sizeof(buf), fi->file))
 396         return frEof;           /* eof */
 397     entry->timestamp = fi->inner_time = (time_t) atoi(buf);
 398     url = strstr(buf, "://");
 399     hier = url ? strstr(url, " - ") : NULL;
 400
 401     if (!url || !hier) {
 402         /*fprintf(stderr, "%s:%d: strange access log entry '%s'\n",
 403          * fname, scanned_count, buf); */
 404         return frError;
 405     }
 406     method = url;
 407     while (!isdigit(*method)) {
 408         if (*method == ' ')
 409             *method = '\0';
 410         --method;
 411     }
 412     method += 2;
 413     method_id = methodStrToId(method);
 414     if (method_id == METHOD_NONE) {
 415         /*fprintf(stderr, "%s:%d: invalid method %s in '%s'\n",
 416          * fname, scanned_count, method, buf); */
 417         return frError;
 418     }
 419     while (*url)
 420         url--;
 421     url++;
 422     *hier = '\0';
 423     hier += 3;
 424     *strchr(hier, '/') = '\0';
 425     /*fprintf(stdout, "%s:%d: %s %s %s\n",
 426      * fname, count, method, url, hier); */
 427     entry->use_icp = strcmp(hier, "NONE");
 428     /* no ICP lookup for these status codes */
 429 /*      strcmp(hier, "NONE") &&
 430  * strcmp(hier, "DIRECT") &&
 431  * strcmp(hier, "FIREWALL_IP_DIRECT") &&
 432  * strcmp(hier, "LOCAL_IP_DIRECT") &&
 433  * strcmp(hier, "NO_DIRECT_FAIL") &&
 434  * strcmp(hier, "NO_PARENT_DIRECT") &&
 435  * strcmp(hier, "SINGLE_PARENT") &&
 436  * strcmp(hier, "PASSTHROUGH_PARENT") &&
 437  * strcmp(hier, "SSL_PARENT_MISS") &&
 438  * strcmp(hier, "DEFAULT_PARENT");
 439  */
 440     memcpy(entry->key, storeKeyPublic(url, method_id), sizeof(entry->key));
 441     /*fprintf(stdout, "%s:%d: %s %s %s %s\n",
 442      * fname, count, method, storeKeyText(entry->key), url, hier); */
 443     return frOk;
 444 }
 445
 446
 447 static void
 448 cachePurge(Cache * cache, storeSwapLogData * s, int update_digest)
 449 {
 450     CacheEntry *olde = (CacheEntry *) hash_lookup(cache->hash, s->key);
 451     if (!olde) {
 452         cache->bad_del_count++;
 453     } else {
 454         assert(cache->count);
 455         hash_remove_link(cache->hash, (hash_link *) olde);
 456         if (update_digest)
 457             cacheDigestDel(cache->digest, s->key);
 458         cacheEntryDestroy(olde);
 459         cache->count--;
 460     }
 461 }
 462
 463 static void
 464 cacheStore(Cache * cache, storeSwapLogData * s, int update_digest)
 465 {
 466     CacheEntry *olde = (CacheEntry *) hash_lookup(cache->hash, s->key);
 467     if (olde) {
 468         cache->bad_add_count++;
 469     } else {
 470         CacheEntry *e = cacheEntryCreate(s);
 471         hash_join(cache->hash, (hash_link *) e);
 472         cache->count++;
 473         if (update_digest)
 474             cacheDigestAdd(cache->digest, e->key);
 475     }
 476 }
 477
 478 static void
 479 cacheUpdateStore(Cache * cache, storeSwapLogData * s, int update_digest)
 480 {
 481     switch (s->op) {
 482     case SWAP_LOG_ADD:
 483         cacheStore(cache, s, update_digest);
 484         break;
 485     case SWAP_LOG_DEL:
 486         cachePurge(cache, s, update_digest);
 487         break;
 488     default:
 489         assert(0);
 490     }
 491 }
 492
 493 static int
 494 usage(const char *prg_name)
 495 {
 496     fprintf(stderr, "usage: %s <access_log> <swap_state> ...\n",
 497         prg_name);
 498     return -1;
 499 }
 500
 501 int
 502 main(int argc, char *argv[])
 503 {
 504     FileIterator **fis = NULL;
 505     const int fi_count = argc - 1;
 506     int active_fi_count = 0;
 507     time_t ready_time;
 508     Cache *them, *us;
 509     int i;
 510
 511     if (argc < 3)
 512         return usage(argv[0]);
 513
 514     them = cacheCreate("them");
 515     us = cacheCreate("us");
 516     them->peer = us;
 517     us->peer = them;
 518
 519     fis = xcalloc(fi_count, sizeof(FileIterator *));
 520     /* init iterators with files */
 521     fis[0] = fileIteratorCreate(argv[1], accessLogReader);
 522     for (i = 2; i < argc; ++i)
 523         fis[i - 1] = fileIteratorCreate(argv[i], swapStateReader);
 524     /* check that all files were found */
 525     for (i = 0; i < fi_count; ++i)
 526         if (!fis[i])
 527             return -2;
 528     /* read prefix to get start-up contents of the peer cache */
 529     ready_time = -1;
 530     for (i = 1; i < fi_count; ++i) {
 531         FileIterator *fi = fis[i];
 532         while (fi->inner_time > 0) {
 533             if (((storeSwapLogData *) fi->entry)->op == SWAP_LOG_DEL) {
 534                 cachePurge(them, fi->entry, 0);
 535                 if (ready_time < 0)
 536                     ready_time = fi->inner_time;
 537             } else {
 538                 if (ready_time > 0 && fi->inner_time > ready_time)
 539                     break;
 540                 cacheStore(them, fi->entry, 0);
 541             }
 542             fileIteratorAdvance(fi);
 543         }
 544     }
 545     /* digest peer cache content */
 546     cacheResetDigest(them);
 547     us->digest = cacheDigestClone(them->digest);        /* @netw@ */
 548
 549     /* shift the time in access log to match ready_time */
 550     fileIteratorSetCurTime(fis[0], ready_time);
 551
 552     /* iterate, use the iterator with the smallest positive inner_time */
 553     cur_time = -1;
 554     do {
 555         int next_i = -1;
 556         time_t next_time = -1;
 557         active_fi_count = 0;
 558         for (i = 0; i < fi_count; ++i) {
 559             if (fis[i]->inner_time >= 0) {
 560                 if (!active_fi_count || fis[i]->inner_time < next_time) {
 561                     next_i = i;
 562                     next_time = fis[i]->inner_time;
 563                 }
 564                 active_fi_count++;
 565             }
 566         }
 567         if (next_i >= 0) {
 568             cur_time = next_time;
 569             /*fprintf(stderr, "%2d time: %d %s", next_i, (int)cur_time, ctime(&cur_time)); */
 570             if (next_i == 0)
 571                 cacheFetch(us, fis[next_i]->entry);
 572             else
 573                 cacheUpdateStore(them, fis[next_i]->entry, 1);
 574             fileIteratorAdvance(fis[next_i]);
 575         }
 576     } while (active_fi_count);
 577
 578     /* report */
 579     cacheReport(them);
 580     cacheReport(us);
 581     cacheQueryReport(us, &us->qstats);
 582
 583     /* clean */
 584     for (i = 0; i < argc - 1; ++i) {
 585         fileIteratorDestroy(fis[i]);
 586     }
 587     xfree(fis);
 588     cacheDestroy(them);
 589     cacheDestroy(us);
 590     return 0;
 591 }