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

fix: give time to detect failed servers before building teams

This commit is contained in:
Evan Tschannen 2018-06-10 20:21:39 -07:00
parent 0bc7274d0e
commit 4903df5ce9
3 changed files with 6 additions and 2 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

6
fdbserver/DataDistribution.actor.cpp
View File

@ -533,6 +533,7 @@ struct DDTeamCollection {
bool primary;
Reference<AsyncVar<bool>> processingUnhealthy;
Future<Void> readyToStart;
Future<Void> checkTeamDelay;
DDTeamCollection(
Database const& cx,
@ -547,8 +548,8 @@ struct DDTeamCollection {
Future<Void> readyToStart, Reference<AsyncVar<bool>> zeroHealthyTeams, bool primary,
Reference<AsyncVar<bool>> processingUnhealthy)
:cx(cx), masterId(masterId), lock(lock), output(output), shardsAffectedByTeamFailure(shardsAffectedByTeamFailure), doBuildTeams( true ), teamBuilder( Void() ),
configuration(configuration), serverChanges(serverChanges), readyToStart(readyToStart),
initialFailureReactionDelay( delayed( readyToStart, BUGGIFY ? 0 : SERVER_KNOBS->INITIAL_FAILURE_REACTION_DELAY, TaskDataDistribution ) ), healthyTeamCount( 0 ),
configuration(configuration), serverChanges(serverChanges), readyToStart(readyToStart), checkTeamDelay( delay( SERVER_KNOBS->CHECK_TEAM_DELAY, TaskDataDistribution) ),
initialFailureReactionDelay( delayed( readyToStart, SERVER_KNOBS->INITIAL_FAILURE_REACTION_DELAY, TaskDataDistribution ) ), healthyTeamCount( 0 ),
initializationDoneActor(logOnCompletion(readyToStart && initialFailureReactionDelay, this)), optimalTeamCount( 0 ), recruitingStream(0), restartRecruiting( SERVER_KNOBS->DEBOUNCE_RECRUITING_DELAY ),
unhealthyServers(0), includedDCs(includedDCs), otherTrackedDCs(otherTrackedDCs), zeroHealthyTeams(zeroHealthyTeams), zeroOptimalTeams(true), primary(primary), processingUnhealthy(processingUnhealthy)
{
@ -586,6 +587,7 @@ struct DDTeamCollection {
ACTOR Future<Void> checkBuildTeams( DDTeamCollection* self ) {
state Promise<Void> restart;
Void _ = wait( self->checkTeamDelay );
while( !self->teamBuilder.isReady() )
Void _ = wait( self->teamBuilder );

1
fdbserver/Knobs.cpp
View File

@ -131,6 +131,7 @@ ServerKnobs::ServerKnobs(bool randomize, ClientKnobs* clientKnobs) {
init( METRIC_DELAY, 0.1 ); if( randomize && BUGGIFY ) METRIC_DELAY = 1.0;
init( ALL_DATA_REMOVED_DELAY, 1.0 );
init( INITIAL_FAILURE_REACTION_DELAY, 30.0 ); if( randomize && BUGGIFY ) INITIAL_FAILURE_REACTION_DELAY = 0.0;
init( CHECK_TEAM_DELAY, 30.0 );
init( LOG_ON_COMPLETION_DELAY, DD_QUEUE_LOGGING_INTERVAL );
init( BEST_TEAM_MAX_TEAM_TRIES, 10 );
init( BEST_TEAM_OPTION_COUNT, 4 );

1
fdbserver/Knobs.h
View File

@ -96,6 +96,7 @@ public:
double METRIC_DELAY;
double ALL_DATA_REMOVED_DELAY;
double INITIAL_FAILURE_REACTION_DELAY;
double CHECK_TEAM_DELAY;
double LOG_ON_COMPLETION_DELAY;
int BEST_TEAM_MAX_TEAM_TRIES;
int BEST_TEAM_OPTION_COUNT;