wbtiger
/
foundationdb
mirror of https://github.com/apple/foundationdb.git
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

fix: the cluster controller did not consider the master sharing the same process as the cluster controller as bad in all needed locations 

waited too long for good recruitment locations, which would add too much time to recoveries of clusters that do not use machine classes
This commit is contained in:
Evan Tschannen 2018-02-06 11:30:05 -08:00
parent 080a454051
commit b7dde88029
2 changed files with 5 additions and 4 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

5
fdbserver/ClusterController.actor.cpp
View File

@ -734,7 +734,7 @@ public:
if ( oldMasterFit < newMasterFit )
return false;
if ( oldMasterFit > newMasterFit )
if ( oldMasterFit > newMasterFit || ( dbi.master.locality.processId() == clusterControllerProcessId && mworker.first.locality.processId() != clusterControllerProcessId ) )
return true;
// Check tLog fitness
@ -840,7 +840,8 @@ ACTOR Future<Void> clusterWatchDatabase( ClusterControllerData* cluster, Cluster
state double recoveryStart = now();
TraceEvent("CCWDB", cluster->id).detail("Recruiting", "Master");
state std::pair<WorkerInterface, ProcessClass> masterWorker = cluster->getMasterWorker(db->config);
if( masterWorker.second.machineClassFitness( ProcessClass::Master ) > SERVER_KNOBS->EXPECTED_MASTER_FITNESS && now() - cluster->startTime < SERVER_KNOBS->WAIT_FOR_GOOD_RECRUITMENT_DELAY ) {
if( ( masterWorker.second.machineClassFitness( ProcessClass::Master ) > SERVER_KNOBS->EXPECTED_MASTER_FITNESS || masterWorker.first.locality.processId() == cluster->clusterControllerProcessId )
&& now() - cluster->startTime < SERVER_KNOBS->WAIT_FOR_GOOD_RECRUITMENT_DELAY ) {
TraceEvent("CCWDB", cluster->id).detail("Fitness", masterWorker.second.machineClassFitness( ProcessClass::Master ));
Void _ = wait( delay(SERVER_KNOBS->ATTEMPT_RECRUITMENT_DELAY) );
continue;

4
fdbserver/Knobs.cpp
View File

@ -259,8 +259,8 @@ ServerKnobs::ServerKnobs(bool randomize, ClientKnobs* clientKnobs) {
init( SIM_SHUTDOWN_TIMEOUT, 10 );
init( SHUTDOWN_TIMEOUT, 600 ); if( randomize && BUGGIFY ) SHUTDOWN_TIMEOUT = 60.0;
init( MASTER_SPIN_DELAY, 1.0 ); if( randomize && BUGGIFY ) MASTER_SPIN_DELAY = 10.0;
init( WAIT_FOR_GOOD_RECRUITMENT_DELAY, 1.0 );
init( ATTEMPT_RECRUITMENT_DELAY, 0.05 );
init( WAIT_FOR_GOOD_RECRUITMENT_DELAY, 0.1 );
init( ATTEMPT_RECRUITMENT_DELAY, 0.35 );
init( WORKER_FAILURE_TIME, 1.0 ); if( randomize && BUGGIFY ) WORKER_FAILURE_TIME = 10.0;
init( CHECK_BETTER_MASTER_INTERVAL, 1.0 ); if( randomize && BUGGIFY ) CHECK_BETTER_MASTER_INTERVAL = 0.001;
init( INCOMPATIBLE_PEERS_LOGGING_INTERVAL, 600 ); if( randomize && BUGGIFY ) INCOMPATIBLE_PEERS_LOGGING_INTERVAL = 60.0;