forked from Mirrors/privacore-open-source-search-engine
catdb is now generated successfully.
This commit is contained in:
@ -29,7 +29,11 @@ bool Catdb::init ( ) {
|
||||
// . what's max # of tree nodes?
|
||||
// . assume avg tagdb rec size (siteUrl) is about 82 bytes we get:
|
||||
// . NOTE: 32 bytes of the 82 are overhead
|
||||
long treeMem = g_conf.m_catdbMaxTreeMem;
|
||||
//long treeMem = g_conf.m_catdbMaxTreeMem;
|
||||
// speed up gen catdb, use 15MB. later maybe once gen is complete
|
||||
// we can free this tree or something...
|
||||
// TODO!
|
||||
long treeMem = 15000000;
|
||||
//long treeMem = 100000000;
|
||||
//long maxTreeNodes = g_conf.m_catdbMaxTreeMem / 82;
|
||||
long maxTreeNodes = treeMem / 82;
|
||||
|
10
Msg1.cpp
10
Msg1.cpp
@ -355,6 +355,12 @@ bool Msg1::sendSomeOfList ( ) {
|
||||
if ( m_list->m_fixedDataSize != getDataSizeFromRdbId(m_rdbId) ) {
|
||||
char *xx=NULL;*xx=0; }
|
||||
|
||||
// little debug thing for genCatdb from msg9b's huge list add
|
||||
if ( m_list->m_listSize > 10000000 )
|
||||
log("msg1: adding chunk @ %li of %li bytes",
|
||||
dataStart - m_list->m_list ,
|
||||
m_list->m_listSize );
|
||||
|
||||
// . now send this list to the host
|
||||
// . this returns false if blocked, true otherwise
|
||||
// . it also sets g_errno on error
|
||||
@ -447,7 +453,9 @@ bool Msg1::sendData ( unsigned long groupId, char *listData , long listSize ) {
|
||||
if ( ! g_errno ) sendToSelf = false;
|
||||
else {
|
||||
log("rdb: msg1 had error: %s",mstrerror(g_errno));
|
||||
return true;
|
||||
// this is messing up generate catdb's huge rdblist add
|
||||
// why did we put it in there??? from msg9b.cpp
|
||||
//return true;
|
||||
}
|
||||
|
||||
QUICKPOLL(m_niceness);
|
||||
|
@ -328,6 +328,12 @@ bool Msg2a::makeCatdb( char *coll,
|
||||
// null terminate
|
||||
m_urls[urlp] = '\0';
|
||||
currUrl++;
|
||||
// debug
|
||||
//SafeBuf sb;
|
||||
//sb.safeMemcpy(&m_urls[urlp-urlLen],urlLen);
|
||||
//sb.nullTerm();
|
||||
//log("gencat: url=%s",sb.getBufStart());
|
||||
|
||||
}
|
||||
log(LOG_INFO, "db: Wrote %li urls to update (%li)\n",
|
||||
currUrl - m_numRemoveUrls, m_numUpdateIndexes);
|
||||
|
@ -45,7 +45,7 @@ bool Msg8b::getCatRec ( Url *url ,
|
||||
// clear g_errno
|
||||
g_errno = 0;
|
||||
// warning
|
||||
if ( ! coll ) log(LOG_LOGIC,"net: NULL collection. msg8b.");
|
||||
//if ( ! coll ) log(LOG_LOGIC,"net: NULL collection. msg8b.");
|
||||
// store the calling parameters in this class for retrieval by callback
|
||||
m_state = state;
|
||||
m_callback = callback;
|
||||
|
@ -123,7 +123,10 @@ bool Msg9b::addCatRecs ( char *urls ,
|
||||
}
|
||||
else if ( ! m_list.addRecord ( key, dataSize, data ) )
|
||||
return true;
|
||||
|
||||
|
||||
// debug
|
||||
//log("gencat: adding url=%s",sr.m_url);
|
||||
|
||||
//skip:
|
||||
// now advance p to e
|
||||
p = e;
|
||||
@ -133,7 +136,8 @@ bool Msg9b::addCatRecs ( char *urls ,
|
||||
|
||||
QUICKPOLL((niceness));
|
||||
}
|
||||
log ( LOG_INFO, "Msg9b: %li sites and %li links added", k , c );
|
||||
log ( LOG_INFO, "Msg9b: %li sites and %li links added. "
|
||||
"listSize=%li", k , c , m_list.m_listSize );
|
||||
// . now add the m_list to tagdb using msg1
|
||||
// . use high priority (niceness of 0)
|
||||
// . i raised niceness from 0 to 1 so multicast does not use the
|
||||
|
File diff suppressed because one or more lines are too long
Reference in New Issue
Block a user