Apache
/
hbase
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

HBASE-25774 TestSyncReplicationStandbyKillRS#testStandbyKillRegionServer is flaky (#3189) 

Wait for the restarter thread to finish before checking the state
Add more detailed logs

Signed-off-by: meiyi <myimeiyi@gmail.com>
This commit is contained in:
Duo Zhang 2021-04-22 10:10:15 +08:00 committed by GitHub
parent d5c5e48839
commit 50920ee306
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
1 changed files with 18 additions and 8 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

26
hbase-server/src/test/java/org/apache/hadoop/hbase/replication/TestSyncReplicationStandbyKillRS.java
View File

@ -68,12 +68,14 @@ public class TestSyncReplicationStandbyKillRS extends SyncReplicationTestBase {
UTIL1.shutdownMiniCluster();
JVMClusterUtil.MasterThread activeMaster = UTIL2.getMiniHBaseCluster().getMasterThread();
String threadName = "RegionServer-Restarter";
Thread t = new Thread(() -> {
try {
List<JVMClusterUtil.RegionServerThread> regionServers =
UTIL2.getMiniHBaseCluster().getLiveRegionServerThreads();
for (JVMClusterUtil.RegionServerThread rst : regionServers) {
ServerName serverName = rst.getRegionServer().getServerName();
LOG.debug("Going to stop [{}]", serverName);
rst.getRegionServer().stop("Stop RS for test");
waitForRSShutdownToStartAndFinish(activeMaster, serverName);
JVMClusterUtil.RegionServerThread restarted =
@ -83,9 +85,11 @@ public class TestSyncReplicationStandbyKillRS extends SyncReplicationTestBase {
} catch (Exception e) {
LOG.error("Failed to kill RS", e);
}
});
}, threadName);
t.start();
LOG.debug("Going to transit peer {} to {} state", PEER_ID,
SyncReplicationState.DOWNGRADE_ACTIVE);
// Transit standby to DA to replay logs
try {
UTIL2.getAdmin().transitReplicationPeerSyncReplicationState(PEER_ID,
@ -94,11 +98,18 @@ public class TestSyncReplicationStandbyKillRS extends SyncReplicationTestBase {
LOG.error("Failed to transit standby cluster to " + SyncReplicationState.DOWNGRADE_ACTIVE, e);
}
while (UTIL2.getAdmin().getReplicationPeerSyncReplicationState(PEER_ID)
!= SyncReplicationState.DOWNGRADE_ACTIVE) {
LOG.debug("Waiting for the restarter thread {} to quit", threadName);
t.join();
while (UTIL2.getAdmin()
.getReplicationPeerSyncReplicationState(PEER_ID) != SyncReplicationState.DOWNGRADE_ACTIVE) {
LOG.debug("Waiting for peer {} to be in {} state", PEER_ID,
SyncReplicationState.DOWNGRADE_ACTIVE);
Thread.sleep(SLEEP_TIME);
}
LOG.debug("Going to verify the result, {} records expected", COUNT);
verify(UTIL2, 0, COUNT);
LOG.debug("Verification successfully done");
}
private void waitForRSShutdownToStartAndFinish(JVMClusterUtil.MasterThread activeMaster,
@ -106,15 +117,14 @@ public class TestSyncReplicationStandbyKillRS extends SyncReplicationTestBase {
ServerManager sm = activeMaster.getMaster().getServerManager();
// First wait for it to be in dead list
while (!sm.getDeadServers().isDeadServer(serverName)) {
LOG.debug("Waiting for [" + serverName + "] to be listed as dead in master");
LOG.debug("Waiting for {} to be listed as dead in master", serverName);
Thread.sleep(SLEEP_TIME);
}
LOG.debug("Server [" + serverName + "] marked as dead, waiting for it to " +
"finish dead processing");
LOG.debug("Server {} marked as dead, waiting for it to finish dead processing", serverName);
while (sm.areDeadServersInProgress()) {
LOG.debug("Server [" + serverName + "] still being processed, waiting");
LOG.debug("Server {} still being processed, waiting", serverName);
Thread.sleep(SLEEP_TIME);
}
LOG.debug("Server [" + serverName + "] done with server shutdown processing");
LOG.debug("Server {} done with server shutdown processing", serverName);
}
}