Skip to content

HDFS-16373. Fix MiniDFSCluster restart in case of multiple namenodes. #3756

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 2 commits into from
Dec 14, 2021
Merged

HDFS-16373. Fix MiniDFSCluster restart in case of multiple namenodes. #3756

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
f00a2b8
HDFS-16373. Fix MiniDFSCluster restart in case of multiple namenodes.
ayushtkn Dec 7, 2021
c0a82f9
Fix Test.
ayushtkn Dec 7, 2021
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
43 changes: 22 additions & 21 deletions hadoop-hdfs-project/hadoop-hdfs/src/test/java/org/apache/hadoop/hdfs/MiniDFSCluster.java
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -2267,9 +2267,11 @@ public synchronized void restartNameNode(int nnIndex, boolean waitActive,
info.nameNode = nn;
info.setStartOpt(startOpt);
if (waitActive) {
waitClusterUp();
if (numDataNodes > 0) {
waitNameNodeUp(nnIndex);
}
LOG.info("Restarted the namenode");
waitActive();
waitActive(nnIndex);
}
}

Expand Down Expand Up @@ -2775,11 +2777,25 @@ public void waitActive(int nnIndex) throws IOException {
DFSClient client = new DFSClient(addr, conf);

// ensure all datanodes have registered and sent heartbeat to the namenode
while (shouldWait(client.datanodeReport(DatanodeReportType.LIVE), addr)) {
int failedCount = 0;
while (true) {
try {
LOG.info("Waiting for cluster to become active");
Thread.sleep(100);
while (shouldWait(client.datanodeReport(DatanodeReportType.LIVE), addr)) {
LOG.info("Waiting for cluster to become active");
Thread.sleep(100);
}
break;
} catch (IOException e) {
failedCount++;
// Cached RPC connection to namenode, if any, is expected to fail once
if (failedCount > 1) {
LOG.warn("Tried waitActive() " + failedCount
+ " time(s) and failed, giving up. " + StringUtils
.stringifyException(e));
throw e;
}
} catch (InterruptedException e) {
throw new IOException(e);
}
}

Expand Down Expand Up @@ -2815,22 +2831,7 @@ public Boolean get() {
*/
public void waitActive() throws IOException {
for (int index = 0; index < namenodes.size(); index++) {
int failedCount = 0;
while (true) {
try {
waitActive(index);
break;
} catch (IOException e) {
failedCount++;
// Cached RPC connection to namenode, if any, is expected to fail once
if (failedCount > 1) {
LOG.warn("Tried waitActive() " + failedCount
+ " time(s) and failed, giving up. "
+ StringUtils.stringifyException(e));
throw e;
}
}
}
waitActive(index);
}
LOG.info("Cluster is active");
}
Expand Down
8 changes: 8 additions & 0 deletions hadoop-hdfs-project/hadoop-hdfs/src/test/java/org/apache/hadoop/hdfs/TestMiniDFSCluster.java
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -309,6 +309,14 @@ public void testSetUpFederatedCluster() throws Exception {
DFSUtil.addKeySuffixes(
DFS_NAMENODE_HTTP_ADDRESS_KEY, "ns1", "nn1")));
}

// Shutdown namenodes individually.
cluster.shutdownNameNode(0);
cluster.shutdownNameNode(1);
Comment on lines +314 to +315
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Shall we just do restarts rather than shutdown + restart?

Copy link
Member Author

@ayushtkn ayushtkn Dec 7, 2021 •
edited
Loading

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I don't think that will repro the issue, Single restart shall pass only since all other namenodes are up. May be just one shutdown and the other one restarting will repro...
But that won't look good. My original use case was from a RBF patch, where I was restaring the namenodes like this only, so kept the logic same

Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I don't think that will repro the issue, Single restart shall pass only since all other namenodes are up. May be just one shutdown and the other one restarting will repro...
But that won't look good. My original use case was from a RBF patch, where I was restaring the namenodes like this only, so kept the logic same

Ah I see, yeah makes sense 👍


// Restart namenodes individually with wait active, both should be successful.
cluster.restartNameNode(0);
cluster.restartNameNode(1);
}
}
}