// init our rdb
bool Titledb::init ( ) {
// key sanity tests
int64_t uh48 = 0x1234567887654321LL & 0x0000ffffffffffffLL;
int64_t docId = 123456789;
key96_t k = makeKey(docId,uh48,false);
if ( getDocId(&k) != docId ) { g_process.shutdownAbort(true);}
if ( getUrlHash48(&k) != uh48 ) { g_process.shutdownAbort(true);}
const char *url = "http://.ezinemark.com/int32_t-island-child-custody-attorneys-new-york-visitation-lawyers-melville-legal-custody-law-firm-45f00bbed18.html";
Url uu;
uu.set(url);
const char *d1 = uu.getDomain();
int32_t dlen1 = uu.getDomainLen();
int32_t dlen2 = 0;
const char *d2 = getDomFast ( url , &dlen2 );
if ( !d1 || !d2 ) { g_process.shutdownAbort(true); }
if ( dlen1 != dlen2 ) { g_process.shutdownAbort(true); }
// another one
url = "http://ok/";
uu.set(url);
const char *d1a = uu.getDomain();
dlen1 = uu.getDomainLen();
dlen2 = 0;
const char *d2a = getDomFast ( url , &dlen2 );
if ( d1a || d2a ) { g_process.shutdownAbort(true); }
if ( dlen1 != dlen2 ) { g_process.shutdownAbort(true); }
// . what's max # of tree nodes?
// . assume avg TitleRec size (compressed html doc) is about 1k we get:
// . NOTE: overhead is about 32 bytes per node
int32_t maxTreeNodes = g_conf.m_titledbMaxTreeMem / (1*1024);
// initialize our own internal rdb
return m_rdb.init ( "titledb" ,
-1 , // fixed record size
//g_conf.m_titledbMinFilesToMerge ,
// this should not really be changed...
-1,
g_conf.m_titledbMaxTreeMem ,
maxTreeNodes ,
false, // half keys?
12, // key size
false, //isCollectionLess
false); //useIndexFile
// validate
//return verify ( );
}
// init our rdb
bool Titledb::init ( ) {
// key sanity tests
int64_t uh48 = 0x1234567887654321LL & 0x0000ffffffffffffLL;
int64_t docId = 123456789;
key_t k = makeKey(docId,uh48,false);
if ( getDocId(&k) != docId ) { char *xx=NULL;*xx=0;}
if ( getUrlHash48(&k) != uh48 ) { char *xx=NULL;*xx=0;}
char *url = "http://.ezinemark.com/int32_t-island-child-custody-attorneys-new-york-visitation-lawyers-melville-legal-custody-law-firm-45f00bbed18.html";
Url uu;
uu.set(url);
char *d1 = uu.getDomain();
int32_t dlen1 = uu.getDomainLen();
int32_t dlen2 = 0;
char *d2 = getDomFast ( url , &dlen2 );
if ( dlen1 != dlen2 ) { char *xx=NULL;*xx=0; }
// another one
url = "http://ok/";
uu.set(url);
d1 = uu.getDomain();
dlen1 = uu.getDomainLen();
dlen2 = 0;
d2 = getDomFast ( url , &dlen2 );
if ( dlen1 != dlen2 ) { char *xx=NULL;*xx=0; }
int64_t maxMem = 200000000; // 200MB
// . what's max # of tree nodes?
// . assume avg TitleRec size (compressed html doc) is about 1k we get:
// . NOTE: overhead is about 32 bytes per node
int32_t maxTreeNodes = maxMem / (1*1024);
// . we now use a disk page cache for titledb as opposed to the
// old rec cache. i am trying to do away with the Rdb::m_cache rec
// cache in favor of cleverly used disk page caches, because
// the rec caches are not real-time and get stale.
// . just hard-code 30MB for now
int32_t pcmem = 30000000; // = g_conf.m_titledbMaxDiskPageCacheMem;
// f**k that we need all the mem!
//pcmem = 0;
// do not use any page cache if doing tmp cluster in order to
// prevent swapping
if ( g_hostdb.m_useTmpCluster ) pcmem = 0;
int32_t pageSize = GB_INDEXDB_PAGE_SIZE;
// init the page cache
// . MDW: "minimize disk seeks" not working otherwise i'd enable it!
if ( ! m_pc.init ( "titledb",
RDB_TITLEDB,
pcmem ,
pageSize ) )
return log("db: Titledb init failed.");
// each entry in the cache is usually just a single record, no lists
//int32_t maxCacheNodes = g_conf.m_titledbMaxCacheMem / (10*1024);
// initialize our own internal rdb
if ( ! m_rdb.init ( g_hostdb.m_dir ,
"titledb" ,
true , // dedup same keys?
-1 , // fixed record size
//g_hostdb.m_groupMask ,
//g_hostdb.m_groupId ,
//g_conf.m_titledbMinFilesToMerge ,
// this should not really be changed...
-1,//3,//230 minfilestomerge mintomerge
maxMem, // g_conf.m_titledbMaxTreeMem ,
maxTreeNodes ,
// now we balance so Sync.cpp can ordered huge list
true , // balance tree?
// turn off cache for now because the page cache
// is just as fast and does not get out of date
// so bad??
//0 ,
0,//g_conf.m_titledbMaxCacheMem ,
0,//maxCacheNodes ,
false ,// half keys?
false ,// g_conf.m_titledbSav
&m_pc , // page cache ptr
true ) )// is titledb?
return false;
return true;
// validate
//return verify ( );
}
