/** Create a new checkpoint */
void doCheckpoint() throws IOException {
LOG.info("Checkpoint starting");
// Do the required initialization of the merge work area.
startCheckpoint();
// Tell the namenode to start logging transactions in a new edit file
// Returns a token that would be used to upload the merged image.
CheckpointSignature sig = (CheckpointSignature) namenode.rollEditLog();
// error simulation code for junit test
if (ErrorSimulator.getErrorSimulation(0)) {
throw new IOException("Simulating error0 " + "after creating edits.new");
}
boolean loadImage = downloadCheckpointFiles(sig); // Fetch fsimage and edits
doMerge(sig, loadImage); // Do the merge
//
// Upload the new image into the NameNode. Then tell the Namenode
// to make this new uploaded image as the most current image.
//
putFSImage(sig);
// error simulation code for junit test
if (ErrorSimulator.getErrorSimulation(1)) {
throw new IOException("Simulating error1 " + "after uploading new image to NameNode");
}
namenode.rollFsImage(new CheckpointSignature(checkpointImage));
checkpointImage.endCheckpoint();
LOG.info("Checkpoint done. New Image Size: " + checkpointImage.getFsImageName().length());
}
/** Merge downloaded image and edits and write the new image into current storage directory. */
private void doMerge(CheckpointSignature sig, boolean loadImage) throws IOException {
if (loadImage) { // create an empty namespace if new image
namesystem = new FSNamesystem(checkpointImage, conf);
}
assert namesystem.dir.fsImage == checkpointImage;
checkpointImage.doMerge(sig, loadImage);
}
/** Initialize SecondaryNameNode. */
private void initialize(Configuration conf) throws IOException {
// initiate Java VM metrics
JvmMetrics.init("SecondaryNameNode", conf.get("session.id"));
// Create connection to the namenode.
shouldRun = true;
nameNodeAddr = NameNode.getAddress(conf);
this.conf = conf;
this.namenode =
(NamenodeProtocol)
RPC.waitForProxy(
NamenodeProtocol.class, NamenodeProtocol.versionID, nameNodeAddr, conf);
// initialize checkpoint directories
fsName = getInfoServer();
checkpointDirs = FSImage.getCheckpointDirs(conf, "/tmp/hadoop/dfs/namesecondary");
checkpointEditsDirs = FSImage.getCheckpointEditsDirs(conf, "/tmp/hadoop/dfs/namesecondary");
checkpointImage = new CheckpointStorage(conf);
checkpointImage.recoverCreate(checkpointDirs, checkpointEditsDirs);
// Initialize other scheduling parameters from the configuration
checkpointPeriod = conf.getLong("fs.checkpoint.period", 3600);
checkpointSize = conf.getLong("fs.checkpoint.size", 4194304);
// initialize the webserver for uploading files.
String infoAddr =
NetUtils.getServerAddress(
conf,
"dfs.secondary.info.bindAddress",
"dfs.secondary.info.port",
"dfs.secondary.http.address");
InetSocketAddress infoSocAddr = NetUtils.createSocketAddr(infoAddr);
infoBindAddress = infoSocAddr.getHostName();
int tmpInfoPort = infoSocAddr.getPort();
infoServer = new HttpServer("secondary", infoBindAddress, tmpInfoPort, tmpInfoPort == 0, conf);
infoServer.setAttribute("name.system.image", checkpointImage);
this.infoServer.setAttribute("name.conf", conf);
infoServer.addInternalServlet("getimage", "/getimage", GetImageServlet.class);
infoServer.start();
// The web-server port can be ephemeral... ensure we have the correct info
infoPort = infoServer.getPort();
conf.set("dfs.secondary.http.address", infoBindAddress + ":" + infoPort);
LOG.info("Secondary Web-server up at: " + infoBindAddress + ":" + infoPort);
LOG.warn(
"Checkpoint Period :"
+ checkpointPeriod
+ " secs "
+ "("
+ checkpointPeriod / 60
+ " min)");
LOG.warn(
"Log Size Trigger :"
+ checkpointSize
+ " bytes "
+ "("
+ checkpointSize / 1024
+ " KB)");
}
/**
* Download <code>fsimage</code> and <code>edits</code> files from the name-node.
*
* @return true if a new image has been downloaded and needs to be loaded
* @throws IOException
*/
private boolean downloadCheckpointFiles(CheckpointSignature sig) throws IOException {
checkpointImage.cTime = sig.cTime;
checkpointImage.checkpointTime = sig.checkpointTime;
boolean downloadImage = true;
String fileid;
File[] srcNames;
if (sig.imageDigest.equals(checkpointImage.imageDigest)) {
downloadImage = false;
LOG.info("Image has not changed. Will not download image.");
} else {
// get fsimage
srcNames = checkpointImage.getImageFiles();
assert srcNames.length > 0 : "No checkpoint targets.";
fileid = "getimage=1";
TransferFsImage.getFileClient(fsName, fileid, srcNames, false);
checkpointImage.imageDigest = sig.imageDigest;
LOG.info(
"Downloaded file " + srcNames[0].getName() + " size " + srcNames[0].length() + " bytes.");
}
// get edits file
fileid = "getedit=1";
srcNames = checkpointImage.getEditsFiles();
assert srcNames.length > 0 : "No checkpoint targets.";
TransferFsImage.getFileClient(fsName, fileid, srcNames, false);
LOG.info(
"Downloaded file " + srcNames[0].getName() + " size " + srcNames[0].length() + " bytes.");
checkpointImage.checkpointUploadDone(null);
return downloadImage;
}
/** Shut down this instance of the datanode. Returns only after shutdown is complete. */
public void shutdown() {
shouldRun = false;
try {
if (infoServer != null) infoServer.stop();
} catch (Exception e) {
LOG.warn("Exception shutting down SecondaryNameNode", e);
}
try {
if (checkpointImage != null) checkpointImage.close();
} catch (IOException e) {
LOG.warn(StringUtils.stringifyException(e));
}
}
//
// The main work loop
//
public void run() {
//
// Poll the Namenode (once every 5 minutes) to find the size of the
// pending edit log.
//
long period = 5 * 60; // 5 minutes
long lastCheckpointTime = 0;
if (checkpointPeriod < period) {
period = checkpointPeriod;
}
while (shouldRun) {
try {
Thread.sleep(1000 * period);
} catch (InterruptedException ie) {
// do nothing
}
if (!shouldRun) {
break;
}
try {
long now = System.currentTimeMillis();
long size = namenode.getEditLogSize();
if (size >= checkpointSize || now >= lastCheckpointTime + 1000 * checkpointPeriod) {
doCheckpoint();
lastCheckpointTime = now;
}
} catch (IOException e) {
LOG.error("Exception in doCheckpoint: ");
LOG.error(StringUtils.stringifyException(e));
e.printStackTrace();
checkpointImage.imageDigest = null;
} catch (Throwable e) {
LOG.error("Throwable Exception in doCheckpoint: ");
LOG.error(StringUtils.stringifyException(e));
e.printStackTrace();
Runtime.getRuntime().exit(-1);
}
}
}
private void startCheckpoint() throws IOException {
checkpointImage.unlockAll();
checkpointImage.getEditLog().close();
checkpointImage.recoverCreate(checkpointDirs, checkpointEditsDirs);
checkpointImage.startCheckpoint();
}
