1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977 978 979 980 981 982 983 984 985 986 987 988 989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085
|
// dbclient.cpp - connect to a Mongo database as a database, from C++
/* Copyright 2009 10gen Inc.
*
* Licensed under the Apache License, Version 2.0 (the "License");
* you may not use this file except in compliance with the License.
* You may obtain a copy of the License at
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS,
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
* See the License for the specific language governing permissions and
* limitations under the License.
*/
#include "pch.h"
#include "dbclient.h"
#include "../bson/util/builder.h"
#include "../db/jsobj.h"
#include "../db/json.h"
#include "../db/dbmessage.h"
#include "connpool.h"
#include "dbclient_rs.h"
#include "../util/background.h"
namespace mongo {
// --------------------------------
// ----- ReplicaSetMonitor ---------
// --------------------------------
// global background job responsible for checking every X amount of time
class ReplicaSetMonitorWatcher : public BackgroundJob {
public:
ReplicaSetMonitorWatcher() : _safego("ReplicaSetMonitorWatcher::_safego") , _started(false) {}
virtual string name() const { return "ReplicaSetMonitorWatcher"; }
void safeGo() {
// check outside of lock for speed
if ( _started )
return;
scoped_lock lk( _safego );
if ( _started )
return;
_started = true;
go();
}
protected:
void run() {
log() << "starting" << endl;
while ( ! inShutdown() ) {
sleepsecs( 10 );
try {
ReplicaSetMonitor::checkAll( true );
}
catch ( std::exception& e ) {
error() << "check failed: " << e.what() << endl;
}
catch ( ... ) {
error() << "unkown error" << endl;
}
}
}
mongo::mutex _safego;
bool _started;
} replicaSetMonitorWatcher;
string seedString( const vector<HostAndPort>& servers ){
string seedStr;
for ( unsigned i = 0; i < servers.size(); i++ ){
seedStr += servers[i].toString();
if( i < servers.size() - 1 ) seedStr += ",";
}
return seedStr;
}
ReplicaSetMonitor::ReplicaSetMonitor( const string& name , const vector<HostAndPort>& servers )
: _lock( "ReplicaSetMonitor instance" ) , _checkConnectionLock( "ReplicaSetMonitor check connection lock" ), _name( name ) , _master(-1), _nextSlave(0) {
uassert( 13642 , "need at least 1 node for a replica set" , servers.size() > 0 );
if ( _name.size() == 0 ) {
warning() << "replica set name empty, first node: " << servers[0] << endl;
}
log() << "starting new replica set monitor for replica set " << _name << " with seed of " << seedString( servers ) << endl;
string errmsg;
for ( unsigned i = 0; i < servers.size(); i++ ) {
// Don't check servers we have already
if( _find_inlock( servers[i] ) >= 0 ) continue;
auto_ptr<DBClientConnection> conn( new DBClientConnection( true , 0, 5.0 ) );
try{
if( ! conn->connect( servers[i] , errmsg ) ){
throw DBException( errmsg, 15928 );
}
log() << "successfully connected to seed " << servers[i] << " for replica set " << this->_name << endl;
}
catch( DBException& e ){
log() << "error connecting to seed " << servers[i] << causedBy( e ) << endl;
// skip seeds that don't work
continue;
}
string maybePrimary;
_checkConnection( conn.get(), maybePrimary, false, -1 );
}
// Check everything to get the first data
_check( true );
log() << "replica set monitor for replica set " << _name << " started, address is " << getServerAddress() << endl;
}
ReplicaSetMonitor::~ReplicaSetMonitor() {
_nodes.clear();
_master = -1;
}
ReplicaSetMonitorPtr ReplicaSetMonitor::get( const string& name , const vector<HostAndPort>& servers ) {
scoped_lock lk( _setsLock );
ReplicaSetMonitorPtr& m = _sets[name];
if ( ! m )
m.reset( new ReplicaSetMonitor( name , servers ) );
replicaSetMonitorWatcher.safeGo();
return m;
}
ReplicaSetMonitorPtr ReplicaSetMonitor::get( const string& name ) {
scoped_lock lk( _setsLock );
map<string,ReplicaSetMonitorPtr>::const_iterator i = _sets.find( name );
if ( i == _sets.end() )
return ReplicaSetMonitorPtr();
return i->second;
}
void ReplicaSetMonitor::checkAll( bool checkAllSecondaries ) {
set<string> seen;
while ( true ) {
ReplicaSetMonitorPtr m;
{
scoped_lock lk( _setsLock );
for ( map<string,ReplicaSetMonitorPtr>::iterator i=_sets.begin(); i!=_sets.end(); ++i ) {
string name = i->first;
if ( seen.count( name ) )
continue;
LOG(1) << "checking replica set: " << name << endl;
seen.insert( name );
m = i->second;
break;
}
}
if ( ! m )
break;
m->check( checkAllSecondaries );
}
}
void ReplicaSetMonitor::setConfigChangeHook( ConfigChangeHook hook ) {
massert( 13610 , "ConfigChangeHook already specified" , _hook == 0 );
_hook = hook;
}
string ReplicaSetMonitor::getServerAddress() const {
scoped_lock lk( _lock );
return _getServerAddress_inlock();
}
string ReplicaSetMonitor::_getServerAddress_inlock() const {
StringBuilder ss;
if ( _name.size() )
ss << _name << "/";
for ( unsigned i=0; i<_nodes.size(); i++ ) {
if ( i > 0 )
ss << ",";
ss << _nodes[i].addr.toString();
}
return ss.str();
}
bool ReplicaSetMonitor::contains( const string& server ) const {
scoped_lock lk( _lock );
for ( unsigned i=0; i<_nodes.size(); i++ ) {
if ( _nodes[i].addr == server )
return true;
}
return false;
}
void ReplicaSetMonitor::notifyFailure( const HostAndPort& server ) {
scoped_lock lk( _lock );
if ( _master >= 0 && _master < (int)_nodes.size() ) {
if ( server == _nodes[_master].addr ) {
_nodes[_master].ok = false;
_master = -1;
}
}
}
HostAndPort ReplicaSetMonitor::getMaster() {
{
scoped_lock lk( _lock );
if ( _master >= 0 && _nodes[_master].ok )
return _nodes[_master].addr;
}
_check( false );
scoped_lock lk( _lock );
uassert( 10009 , str::stream() << "ReplicaSetMonitor no master found for set: " << _name , _master >= 0 );
return _nodes[_master].addr;
}
HostAndPort ReplicaSetMonitor::getSlave( const HostAndPort& prev ) {
// make sure its valid
bool wasFound = false;
// This is always true, since checked in port()
assert( prev.port() >= 0 );
if( prev.host().size() ){
scoped_lock lk( _lock );
for ( unsigned i=0; i<_nodes.size(); i++ ) {
if ( prev != _nodes[i].addr )
continue;
wasFound = true;
if ( _nodes[i].okForSecondaryQueries() )
return prev;
break;
}
}
if( prev.host().size() ){
if( wasFound ){ LOG(1) << "slave '" << prev << "' is no longer ok to use" << endl; }
else{ LOG(1) << "slave '" << prev << "' was not found in the replica set" << endl; }
}
else LOG(1) << "slave '" << prev << "' is not initialized or invalid" << endl;
return getSlave();
}
HostAndPort ReplicaSetMonitor::getSlave() {
LOG(2) << "dbclient_rs getSlave " << getServerAddress() << endl;
scoped_lock lk( _lock );
for ( unsigned ii = 0; ii < _nodes.size(); ii++ ) {
_nextSlave = ( _nextSlave + 1 ) % _nodes.size();
if ( _nextSlave != _master ) {
if ( _nodes[ _nextSlave ].okForSecondaryQueries() )
return _nodes[ _nextSlave ].addr;
LOG(2) << "dbclient_rs getSlave not selecting " << _nodes[_nextSlave] << ", not currently okForSecondaryQueries" << endl;
}
}
if( _master >= 0 ) {
assert( static_cast<unsigned>(_master) < _nodes.size() );
LOG(2) << "dbclient_rs getSlave no member in secondary state found, returning primary " << _nodes[ _master ] << endl;
return _nodes[_master].addr;
}
LOG(2) << "dbclient_rs getSlave no suitable member found, returning first node " << _nodes[ 0 ] << endl;
assert( _nodes.size() > 0 );
return _nodes[0].addr;
}
/**
* notify the monitor that server has failed
*/
void ReplicaSetMonitor::notifySlaveFailure( const HostAndPort& server ) {
int x = _find( server );
if ( x >= 0 ) {
scoped_lock lk( _lock );
_nodes[x].ok = false;
}
}
void ReplicaSetMonitor::_checkStatus( const string& hostAddr ) {
BSONObj status;
/* replSetGetStatus requires admin auth so use a connection from the pool,
* which are authenticated with the keyFile credentials.
*/
ScopedDbConnection authenticatedConn( hostAddr );
if ( !authenticatedConn->runCommand( "admin", BSON( "replSetGetStatus" << 1 ), status )) {
LOG(1) << "dbclient_rs replSetGetStatus failed" << endl;
authenticatedConn.done(); // connection worked properly, but we got an error from server
return;
}
// Make sure we return when finished
authenticatedConn.done();
if( !status.hasField("members") ) {
log() << "dbclient_rs error expected members field in replSetGetStatus result" << endl;
return;
}
if( status["members"].type() != Array) {
log() << "dbclient_rs error expected members field in replSetGetStatus result to be an array" << endl;
return;
}
BSONObjIterator hi(status["members"].Obj());
while (hi.more()) {
BSONObj member = hi.next().Obj();
string host = member["name"].String();
int m = -1;
if ((m = _find(host)) < 0) {
continue;
}
double state = member["state"].Number();
if (member["health"].Number() == 1 && (state == 1 || state == 2)) {
scoped_lock lk( _lock );
_nodes[m].ok = true;
}
else {
scoped_lock lk( _lock );
_nodes[m].ok = false;
}
}
}
NodeDiff ReplicaSetMonitor::_getHostDiff_inlock( const BSONObj& hostList ){
NodeDiff diff;
set<int> nodesFound;
int index = 0;
BSONObjIterator hi( hostList );
while( hi.more() ){
string toCheck = hi.next().String();
int nodeIndex = _find_inlock( toCheck );
// Node-to-add
if( nodeIndex < 0 ) diff.first.insert( toCheck );
else nodesFound.insert( nodeIndex );
index++;
}
for( size_t i = 0; i < _nodes.size(); i++ ){
if( nodesFound.find( static_cast<int>(i) ) == nodesFound.end() ) diff.second.insert( static_cast<int>(i) );
}
return diff;
}
bool ReplicaSetMonitor::_shouldChangeHosts( const BSONObj& hostList, bool inlock ){
int origHosts = 0;
if( ! inlock ){
scoped_lock lk( _lock );
origHosts = _nodes.size();
}
else origHosts = _nodes.size();
int numHosts = 0;
bool changed = false;
BSONObjIterator hi(hostList);
while ( hi.more() ) {
string toCheck = hi.next().String();
numHosts++;
int index = 0;
if( ! inlock ) index = _find( toCheck );
else index = _find_inlock( toCheck );
if ( index >= 0 ) continue;
changed = true;
break;
}
return changed || origHosts != numHosts;
}
void ReplicaSetMonitor::_checkHosts( const BSONObj& hostList, bool& changed ) {
// Fast path, still requires intermittent locking
if( ! _shouldChangeHosts( hostList, false ) ){
changed = false;
return;
}
// Slow path, double-checked though
scoped_lock lk( _lock );
// Our host list may have changed while waiting for another thread in the meantime,
// so double-check here
// TODO: Do we really need this much protection, this should be pretty rare and not
// triggered from lots of threads, duping old behavior for safety
if( ! _shouldChangeHosts( hostList, true ) ){
changed = false;
return;
}
// LogLevel can be pretty low, since replica set reconfiguration should be pretty rare and
// we want to record our changes
log() << "changing hosts to " << hostList << " from " << _getServerAddress_inlock() << endl;
NodeDiff diff = _getHostDiff_inlock( hostList );
set<string> added = diff.first;
set<int> removed = diff.second;
assert( added.size() > 0 || removed.size() > 0 );
changed = true;
// Delete from the end so we don't invalidate as we delete, delete indices are ascending
for( set<int>::reverse_iterator i = removed.rbegin(), end = removed.rend(); i != end; ++i ){
log() << "erasing host " << _nodes[ *i ] << " from replica set " << this->_name << endl;
_nodes.erase( _nodes.begin() + *i );
}
// Add new nodes
for( set<string>::iterator i = added.begin(), end = added.end(); i != end; ++i ){
log() << "trying to add new host " << *i << " to replica set " << this->_name << endl;
// Connect to new node
HostAndPort h( *i );
DBClientConnection * newConn = new DBClientConnection( true, 0, 5.0 );
string errmsg;
try{
if( ! newConn->connect( h , errmsg ) ){
throw DBException( errmsg, 15927 );
}
log() << "successfully connected to new host " << *i << " in replica set " << this->_name << endl;
}
catch( DBException& e ){
warning() << "cannot connect to new host " << *i << " to replica set " << this->_name << causedBy( e ) << endl;
}
_nodes.push_back( Node( h , newConn ) );
}
}
bool ReplicaSetMonitor::_checkConnection( DBClientConnection* conn,
string& maybePrimary, bool verbose, int nodesOffset ) {
assert( conn );
scoped_lock lk( _checkConnectionLock );
bool isMaster = false;
bool changed = false;
bool errorOccured = false;
if ( nodesOffset >= 0 ){
scoped_lock lk( _lock );
if ( !_checkConnMatch_inlock( conn, nodesOffset )) {
/* Another thread modified _nodes -> invariant broken.
* This also implies that another thread just passed
* through here and refreshed _nodes. So no need to do
* duplicate work.
*/
return false;
}
}
try {
Timer t;
BSONObj o;
conn->isMaster( isMaster, &o );
if ( o["setName"].type() != String || o["setName"].String() != _name ) {
warning() << "node: " << conn->getServerAddress()
<< " isn't a part of set: " << _name
<< " ismaster: " << o << endl;
if ( nodesOffset >= 0 ) {
scoped_lock lk( _lock );
_nodes[nodesOffset].ok = false;
}
return false;
}
if ( nodesOffset >= 0 ) {
scoped_lock lk( _lock );
_nodes[nodesOffset].pingTimeMillis = t.millis();
_nodes[nodesOffset].hidden = o["hidden"].trueValue();
_nodes[nodesOffset].secondary = o["secondary"].trueValue();
_nodes[nodesOffset].ismaster = o["ismaster"].trueValue();
_nodes[nodesOffset].lastIsMaster = o.copy();
}
log( ! verbose ) << "ReplicaSetMonitor::_checkConnection: " << conn->toString()
<< ' ' << o << endl;
// add other nodes
BSONArrayBuilder b;
if ( o["hosts"].type() == Array ) {
if ( o["primary"].type() == String )
maybePrimary = o["primary"].String();
BSONObjIterator it( o["hosts"].Obj() );
while( it.more() ) b.append( it.next() );
}
if (o.hasField("passives") && o["passives"].type() == Array) {
BSONObjIterator it( o["passives"].Obj() );
while( it.more() ) b.append( it.next() );
}
_checkHosts( b.arr(), changed);
_checkStatus( conn->getServerAddress() );
}
catch ( std::exception& e ) {
log( ! verbose ) << "ReplicaSetMonitor::_checkConnection: caught exception "
<< conn->toString() << ' ' << e.what() << endl;
errorOccured = true;
}
if ( errorOccured ) {
scoped_lock lk( _lock );
_nodes[nodesOffset].ok = false;
}
if ( changed && _hook )
_hook( this );
return isMaster;
}
void ReplicaSetMonitor::_check( bool checkAllSecondaries ) {
LOG(1) << "_check : " << getServerAddress() << endl;
int newMaster = -1;
shared_ptr<DBClientConnection> nodeConn;
for ( int retry = 0; retry < 2; retry++ ) {
bool triedQuickCheck = false;
if ( !checkAllSecondaries ) {
scoped_lock lk( _lock );
if ( _master >= 0 ) {
/* Nothing else to do since another thread already
* found the _master
*/
return;
}
}
for ( unsigned i = 0; /* should not check while outside of lock! */ ; i++ ) {
{
scoped_lock lk( _lock );
if ( i >= _nodes.size() ) break;
nodeConn = _nodes[i].conn;
}
string maybePrimary;
if ( _checkConnection( nodeConn.get(), maybePrimary, retry, i ) ) {
scoped_lock lk( _lock );
if ( _checkConnMatch_inlock( nodeConn.get(), i )) {
_master = i;
newMaster = i;
if ( !checkAllSecondaries )
return;
}
else {
/*
* Somebody modified _nodes and most likely set the new
* _master, so try again.
*/
break;
}
}
if ( ! triedQuickCheck && ! maybePrimary.empty() ) {
int probablePrimaryIdx = -1;
shared_ptr<DBClientConnection> probablePrimaryConn;
{
scoped_lock lk( _lock );
probablePrimaryIdx = _find_inlock( maybePrimary );
probablePrimaryConn = _nodes[probablePrimaryIdx].conn;
}
if ( probablePrimaryIdx >= 0 ) {
triedQuickCheck = true;
string dummy;
if ( _checkConnection( probablePrimaryConn.get(), dummy,
false, probablePrimaryIdx ) ) {
scoped_lock lk( _lock );
if ( _checkConnMatch_inlock( probablePrimaryConn.get(),
probablePrimaryIdx )) {
_master = probablePrimaryIdx;
newMaster = probablePrimaryIdx;
if ( ! checkAllSecondaries )
return;
}
else {
/*
* Somebody modified _nodes and most likely set the
* new _master, so try again.
*/
break;
}
}
}
}
}
if ( newMaster >= 0 )
return;
sleepsecs( 1 );
}
}
void ReplicaSetMonitor::check( bool checkAllSecondaries ) {
shared_ptr<DBClientConnection> masterConn;
{
scoped_lock lk( _lock );
// first see if the current master is fine
if ( _master >= 0 ) {
masterConn = _nodes[_master].conn;
}
}
if ( masterConn.get() != NULL ) {
string temp;
if ( _checkConnection( masterConn.get(), temp, false, _master )) {
if ( ! checkAllSecondaries ) {
// current master is fine, so we're done
return;
}
}
}
// we either have no master, or the current is dead
_check( checkAllSecondaries );
}
int ReplicaSetMonitor::_find( const string& server ) const {
scoped_lock lk( _lock );
return _find_inlock( server );
}
int ReplicaSetMonitor::_find_inlock( const string& server ) const {
const size_t size = _nodes.size();
for ( unsigned i = 0; i < size; i++ ) {
if ( _nodes[i].addr == server ) {
return i;
}
}
return -1;
}
void ReplicaSetMonitor::appendInfo( BSONObjBuilder& b ) const {
scoped_lock lk( _lock );
BSONArrayBuilder hosts( b.subarrayStart( "hosts" ) );
for ( unsigned i=0; i<_nodes.size(); i++ ) {
hosts.append( BSON( "addr" << _nodes[i].addr <<
// "lastIsMaster" << _nodes[i].lastIsMaster << // this is a potential race, so only used when debugging
"ok" << _nodes[i].ok <<
"ismaster" << _nodes[i].ismaster <<
"hidden" << _nodes[i].hidden <<
"secondary" << _nodes[i].secondary <<
"pingTimeMillis" << _nodes[i].pingTimeMillis ) );
}
hosts.done();
b.append( "master" , _master );
b.append( "nextSlave" , _nextSlave );
}
bool ReplicaSetMonitor::_checkConnMatch_inlock( DBClientConnection* conn,
size_t nodeOffset ) const {
return ( nodeOffset < _nodes.size() &&
conn->getServerAddress() == _nodes[nodeOffset].conn->getServerAddress() );
}
mongo::mutex ReplicaSetMonitor::_setsLock( "ReplicaSetMonitor" );
map<string,ReplicaSetMonitorPtr> ReplicaSetMonitor::_sets;
ReplicaSetMonitor::ConfigChangeHook ReplicaSetMonitor::_hook;
// --------------------------------
// ----- DBClientReplicaSet ---------
// --------------------------------
DBClientReplicaSet::DBClientReplicaSet( const string& name , const vector<HostAndPort>& servers, double so_timeout )
: _monitor( ReplicaSetMonitor::get( name , servers ) ),
_so_timeout( so_timeout ) {
}
DBClientReplicaSet::~DBClientReplicaSet() {
}
DBClientConnection * DBClientReplicaSet::checkMaster() {
HostAndPort h = _monitor->getMaster();
if ( h == _masterHost && _master ) {
// a master is selected. let's just make sure connection didn't die
if ( ! _master->isFailed() )
return _master.get();
_monitor->notifyFailure( _masterHost );
}
_masterHost = _monitor->getMaster();
_master.reset( new DBClientConnection( true , this , _so_timeout ) );
string errmsg;
if ( ! _master->connect( _masterHost , errmsg ) ) {
_monitor->notifyFailure( _masterHost );
uasserted( 13639 , str::stream() << "can't connect to new replica set master [" << _masterHost.toString() << "] err: " << errmsg );
}
_auth( _master.get() );
return _master.get();
}
DBClientConnection * DBClientReplicaSet::checkSlave() {
HostAndPort h = _monitor->getSlave( _slaveHost );
if ( h == _slaveHost && _slave ) {
if ( ! _slave->isFailed() )
return _slave.get();
_monitor->notifySlaveFailure( _slaveHost );
_slaveHost = _monitor->getSlave();
}
else {
_slaveHost = h;
}
_slave.reset( new DBClientConnection( true , this , _so_timeout ) );
_slave->connect( _slaveHost );
_auth( _slave.get() );
return _slave.get();
}
void DBClientReplicaSet::_auth( DBClientConnection * conn ) {
for ( list<AuthInfo>::iterator i=_auths.begin(); i!=_auths.end(); ++i ) {
const AuthInfo& a = *i;
string errmsg;
if ( ! conn->auth( a.dbname , a.username , a.pwd , errmsg, a.digestPassword ) )
warning() << "cached auth failed for set: " << _monitor->getName() << " db: " << a.dbname << " user: " << a.username << endl;
}
}
DBClientConnection& DBClientReplicaSet::masterConn() {
return *checkMaster();
}
DBClientConnection& DBClientReplicaSet::slaveConn() {
return *checkSlave();
}
bool DBClientReplicaSet::connect() {
try {
checkMaster();
}
catch (AssertionException&) {
if (_master && _monitor) {
_monitor->notifyFailure(_masterHost);
}
return false;
}
return true;
}
bool DBClientReplicaSet::auth(const string &dbname, const string &username, const string &pwd, string& errmsg, bool digestPassword ) {
DBClientConnection * m = checkMaster();
// first make sure it actually works
if( ! m->auth(dbname, username, pwd, errmsg, digestPassword ) )
return false;
// now that it does, we should save so that for a new node we can auth
_auths.push_back( AuthInfo( dbname , username , pwd , digestPassword ) );
return true;
}
// ------------- simple functions -----------------
void DBClientReplicaSet::insert( const string &ns , BSONObj obj , int flags) {
checkMaster()->insert(ns, obj, flags);
}
void DBClientReplicaSet::insert( const string &ns, const vector< BSONObj >& v , int flags) {
checkMaster()->insert(ns, v, flags);
}
void DBClientReplicaSet::remove( const string &ns , Query obj , bool justOne ) {
checkMaster()->remove(ns, obj, justOne);
}
void DBClientReplicaSet::update( const string &ns , Query query , BSONObj obj , bool upsert , bool multi ) {
return checkMaster()->update(ns, query, obj, upsert,multi);
}
auto_ptr<DBClientCursor> DBClientReplicaSet::query(const string &ns, Query query, int nToReturn, int nToSkip,
const BSONObj *fieldsToReturn, int queryOptions, int batchSize) {
if ( queryOptions & QueryOption_SlaveOk ) {
// we're ok sending to a slave
// we'll try 2 slaves before just using master
// checkSlave will try a different slave automatically after a failure
for ( int i=0; i<3; i++ ) {
try {
return checkSlaveQueryResult( checkSlave()->query(ns,query,nToReturn,nToSkip,fieldsToReturn,queryOptions,batchSize) );
}
catch ( DBException &e ) {
LOG(1) << "can't query replica set slave " << i << " : " << _slaveHost << causedBy( e ) << endl;
}
}
}
return checkMaster()->query(ns,query,nToReturn,nToSkip,fieldsToReturn,queryOptions,batchSize);
}
BSONObj DBClientReplicaSet::findOne(const string &ns, const Query& query, const BSONObj *fieldsToReturn, int queryOptions) {
if ( queryOptions & QueryOption_SlaveOk ) {
// we're ok sending to a slave
// we'll try 2 slaves before just using master
// checkSlave will try a different slave automatically after a failure
for ( int i=0; i<3; i++ ) {
try {
return checkSlave()->findOne(ns,query,fieldsToReturn,queryOptions);
}
catch ( DBException &e ) {
LOG(1) << "can't findone replica set slave " << i << " : " << _slaveHost << causedBy( e ) << endl;
}
}
}
return checkMaster()->findOne(ns,query,fieldsToReturn,queryOptions);
}
void DBClientReplicaSet::killCursor( long long cursorID ) {
// we should neve call killCursor on a replica set conncetion
// since we don't know which server it belongs to
// can't assume master because of slave ok
// and can have a cursor survive a master change
assert(0);
}
void DBClientReplicaSet::isntMaster() {
log() << "got not master for: " << _masterHost << endl;
_monitor->notifyFailure( _masterHost );
_master.reset();
}
auto_ptr<DBClientCursor> DBClientReplicaSet::checkSlaveQueryResult( auto_ptr<DBClientCursor> result ){
if ( result.get() == NULL ) return result;
BSONObj error;
bool isError = result->peekError( &error );
if( ! isError ) return result;
// We only check for "not master or secondary" errors here
// If the error code here ever changes, we need to change this code also
BSONElement code = error["code"];
if( code.isNumber() && code.Int() == 13436 /* not master or secondary */ ){
isntSecondary();
throw DBException( str::stream() << "slave " << _slaveHost.toString() << " is no longer secondary", 14812 );
}
return result;
}
void DBClientReplicaSet::isntSecondary() {
log() << "slave no longer has secondary status: " << _slaveHost << endl;
// Failover to next slave
_monitor->notifySlaveFailure( _slaveHost );
_slave.reset();
}
void DBClientReplicaSet::say( Message& toSend, bool isRetry ) {
if( ! isRetry )
_lazyState = LazyState();
int lastOp = -1;
bool slaveOk = false;
if ( ( lastOp = toSend.operation() ) == dbQuery ) {
// TODO: might be possible to do this faster by changing api
DbMessage dm( toSend );
QueryMessage qm( dm );
if ( ( slaveOk = ( qm.queryOptions & QueryOption_SlaveOk ) ) ) {
for ( int i = _lazyState._retries; i < 3; i++ ) {
try {
DBClientConnection* slave = checkSlave();
slave->say( toSend );
_lazyState._lastOp = lastOp;
_lazyState._slaveOk = slaveOk;
_lazyState._retries = i;
_lazyState._lastClient = slave;
return;
}
catch ( DBException &e ) {
LOG(1) << "can't callLazy replica set slave " << i << " : " << _slaveHost << causedBy( e ) << endl;
}
}
}
}
DBClientConnection* master = checkMaster();
master->say( toSend );
_lazyState._lastOp = lastOp;
_lazyState._slaveOk = slaveOk;
_lazyState._retries = 3;
_lazyState._lastClient = master;
return;
}
bool DBClientReplicaSet::recv( Message& m ) {
assert( _lazyState._lastClient );
// TODO: It would be nice if we could easily wrap a conn error as a result error
try {
return _lazyState._lastClient->recv( m );
}
catch( DBException& e ){
log() << "could not receive data from " << _lazyState._lastClient << causedBy( e ) << endl;
return false;
}
}
void DBClientReplicaSet::checkResponse( const char* data, int nReturned, bool* retry, string* targetHost ){
// For now, do exactly as we did before, so as not to break things. In general though, we
// should fix this so checkResponse has a more consistent contract.
if( ! retry ){
if( _lazyState._lastClient )
return _lazyState._lastClient->checkResponse( data, nReturned );
else
return checkMaster()->checkResponse( data, nReturned );
}
*retry = false;
if( targetHost && _lazyState._lastClient ) *targetHost = _lazyState._lastClient->getServerAddress();
else if (targetHost) *targetHost = "";
if( ! _lazyState._lastClient ) return;
if( nReturned != 1 && nReturned != -1 ) return;
BSONObj dataObj;
if( nReturned == 1 ) dataObj = BSONObj( data );
// Check if we should retry here
if( _lazyState._lastOp == dbQuery && _lazyState._slaveOk ){
// Check the error code for a slave not secondary error
if( nReturned == -1 ||
( hasErrField( dataObj ) && ! dataObj["code"].eoo() && dataObj["code"].Int() == 13436 ) ){
bool wasMaster = false;
if( _lazyState._lastClient == _slave.get() ){
isntSecondary();
}
else if( _lazyState._lastClient == _master.get() ){
wasMaster = true;
isntMaster();
}
else
warning() << "passed " << dataObj << " but last rs client " << _lazyState._lastClient->toString() << " is not master or secondary" << endl;
if( _lazyState._retries < 3 ){
_lazyState._retries++;
*retry = true;
}
else{
(void)wasMaster; // silence set-but-not-used warning
// assert( wasMaster );
// printStackTrace();
log() << "too many retries (" << _lazyState._retries << "), could not get data from replica set" << endl;
}
}
}
}
bool DBClientReplicaSet::call( Message &toSend, Message &response, bool assertOk , string * actualServer ) {
const char * ns = 0;
if ( toSend.operation() == dbQuery ) {
// TODO: might be possible to do this faster by changing api
DbMessage dm( toSend );
QueryMessage qm( dm );
ns = qm.ns;
if ( qm.queryOptions & QueryOption_SlaveOk ) {
for ( int i=0; i<3; i++ ) {
try {
DBClientConnection* s = checkSlave();
if ( actualServer )
*actualServer = s->getServerAddress();
return s->call( toSend , response , assertOk );
}
catch ( DBException &e ) {
LOG(1) << "can't call replica set slave " << i << " : " << _slaveHost << causedBy( e ) << endl;
if ( actualServer )
*actualServer = "";
}
}
}
}
DBClientConnection* m = checkMaster();
if ( actualServer )
*actualServer = m->getServerAddress();
if ( ! m->call( toSend , response , assertOk ) )
return false;
if ( ns ) {
QueryResult * res = (QueryResult*)response.singleData();
if ( res->nReturned == 1 ) {
BSONObj x(res->data() );
if ( str::contains( ns , "$cmd" ) ) {
if ( isNotMasterErrorString( x["errmsg"] ) )
isntMaster();
}
else {
if ( isNotMasterErrorString( getErrField( x ) ) )
isntMaster();
}
}
}
return true;
}
}
|