HBASE-20024 Fixed flakyness of TestMergeTableRegionsProcedure

We assumed that we can run for loop from 0 to lastStep sequentially. MergeTableRegionProcedure skips step 2. So, when i is 0 the procedure is already at step 3.
Added a method StateMachineProcedure#getCurrentStateId that can be used from test code only.
This commit is contained in:
Umesh Agashe 2018-02-27 08:25:37 -08:00 committed by Michael Stack
parent ee3a469641
commit 974200fca1
2 changed files with 34 additions and 2 deletions

View File

@ -28,6 +28,9 @@ import org.apache.yetus.audience.InterfaceAudience;
import org.apache.yetus.audience.InterfaceStability;
import org.slf4j.Logger;
import org.slf4j.LoggerFactory;
import org.apache.hbase.thirdparty.com.google.common.annotations.VisibleForTesting;
import org.apache.hadoop.hbase.shaded.protobuf.generated.ProcedureProtos.StateMachineProcedureData;
/**
@ -253,6 +256,16 @@ public abstract class StateMachineProcedure<TEnvironment, TState>
return stateCount > 0 ? getState(states[stateCount-1]) : getInitialState();
}
/**
* This method is used from test code as it cannot be assumed that state transition will happen
* sequentially. Some procedures may skip steps/ states, some may add intermediate steps in
* future.
*/
@VisibleForTesting
public int getCurrentStateId() {
return getStateId(getCurrentState());
}
/**
* Set the next state for the procedure.
* @param stateId the ordinal() of the state enum (or state id)

View File

@ -54,8 +54,10 @@ import org.apache.hadoop.hbase.master.RegionState;
import org.apache.hadoop.hbase.master.TableStateManager;
import org.apache.hadoop.hbase.master.assignment.AssignmentManager;
import org.apache.hadoop.hbase.monitoring.TaskMonitor;
import org.apache.hadoop.hbase.procedure2.Procedure;
import org.apache.hadoop.hbase.procedure2.ProcedureExecutor;
import org.apache.hadoop.hbase.procedure2.ProcedureTestingUtility;
import org.apache.hadoop.hbase.procedure2.StateMachineProcedure;
import org.apache.hadoop.hbase.util.Bytes;
import org.apache.hadoop.hbase.util.FSUtils;
import org.apache.hadoop.hbase.util.MD5Hash;
@ -377,11 +379,28 @@ public class MasterProcedureTestingUtility {
// execute step N - kill before store update
// restart executor/store
// execute step N - save on store
for (int i = 0; i < numSteps; ++i) {
LOG.info("Restart " + i + " exec state=" + procExec.getProcedure(procId));
// NOTE: currently we make assumption that states/ steps are sequential. There are already
// instances of a procedures which skip (don't use) intermediate states/ steps. In future,
// intermediate states/ steps can be added with ordinal greater than lastStep. If and when
// that happens the states can not be treated as sequential steps and the condition in
// following while loop needs to be changed. We can use euqals/ not equals operator to check
// if the procedure has reached the user specified state. But there is a possibility that
// while loop may not get the control back exaclty when the procedure is in lastStep. Proper
// fix would be get all visited states by the procedure and then check if user speccified
// state is in that list. Current assumption of sequential proregression of steps/ states is
// made at multiple places so we can keep while condition below for simplicity.
Procedure proc = procExec.getProcedure(procId);
int stepNum = proc instanceof StateMachineProcedure ?
((StateMachineProcedure) proc).getCurrentStateId() : 0;
while (stepNum < numSteps) {
LOG.info("Restart " + stepNum + " exec state=" + proc);
ProcedureTestingUtility.assertProcNotYetCompleted(procExec, procId);
restartMasterProcedureExecutor(procExec);
ProcedureTestingUtility.waitProcedure(procExec, procId);
// Old proc object is stale, need to get the new one after ProcedureExecutor restart
proc = procExec.getProcedure(procId);
stepNum = proc instanceof StateMachineProcedure ?
((StateMachineProcedure) proc).getCurrentStateId() : stepNum + 1;
}
assertEquals(expectExecRunning, procExec.isRunning());