Skip to content

YARN-11706. Make yarn rmadmin tool support failover function #6957

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

Closed
wants to merge 1 commit into from
Closed

YARN-11706. Make yarn rmadmin tool support failover function #6957

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
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
59 changes: 52 additions & 7 deletions ...p-yarn/hadoop-yarn-client/src/main/java/org/apache/hadoop/yarn/client/cli/RMAdminCLI.java
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -43,6 +43,9 @@
import org.apache.hadoop.fs.CommonConfigurationKeys;
import org.apache.hadoop.ha.HAAdmin;
import org.apache.hadoop.ha.HAServiceTarget;
import org.apache.hadoop.ha.ServiceFailedException;
import org.apache.hadoop.ha.HAServiceProtocol;
import org.apache.hadoop.ha.HAServiceProtocolHelper;
import org.apache.hadoop.ipc.RemoteException;
import org.apache.hadoop.security.UserGroupInformation;
import org.apache.hadoop.util.ToolRunner;
Expand Down Expand Up @@ -172,6 +175,14 @@ public class RMAdminCLI extends HAAdmin {
"Update resource on specific node."))
.build();

private final static Map<String, UsageInfo> YARN_HA_USAGE =
ImmutableMap.<String, UsageInfo> builder()
.put("-failover", new UsageInfo(
"<serviceId> <serviceId>",
Copy link
Contributor

Choose a reason for hiding this comment

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

What does means?

Copy link
Contributor Author

Choose a reason for hiding this comment

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

Thanks for review ,
USAGE from HAAdmin.java does not contain usage and help information about failover,so YARN_HA_USAGE structure is added here for appending these information

"Failover from the first service to the second.\n"))
.putAll(USAGE)
.build();

public RMAdminCLI() {
super();
}
Expand All @@ -189,7 +200,7 @@ protected void setOut(PrintStream out) {
}

private static void appendHAUsage(final StringBuilder usageBuilder) {
for (Map.Entry<String,UsageInfo> cmdEntry : USAGE.entrySet()) {
for (Map.Entry<String, UsageInfo> cmdEntry : YARN_HA_USAGE.entrySet()) {
if (cmdEntry.getKey().equals("-help")) {
continue;
}
Expand All @@ -206,7 +217,7 @@ private static void appendHAUsage(final StringBuilder usageBuilder) {
private static void buildHelpMsg(String cmd, StringBuilder builder) {
UsageInfo usageInfo = ADMIN_USAGE.get(cmd);
if (usageInfo == null) {
usageInfo = USAGE.get(cmd);
usageInfo = YARN_HA_USAGE.get(cmd);
if (usageInfo == null) {
return;
}
Expand All @@ -225,7 +236,7 @@ private static void buildIndividualUsageMsg(String cmd,
boolean isHACommand = false;
UsageInfo usageInfo = ADMIN_USAGE.get(cmd);
if (usageInfo == null) {
usageInfo = USAGE.get(cmd);
usageInfo = YARN_HA_USAGE.get(cmd);
if (usageInfo == null) {
return;
}
Expand All @@ -246,12 +257,12 @@ private static void buildIndividualUsageMsg(String cmd,
private static void buildUsageMsg(StringBuilder builder,
boolean isHAEnabled) {
builder.append("Usage: yarn rmadmin\n");
for (Map.Entry<String,UsageInfo> cmdEntry : ADMIN_USAGE.entrySet()) {
for (Map.Entry<String, UsageInfo> cmdEntry : ADMIN_USAGE.entrySet()) {
UsageInfo usageInfo = cmdEntry.getValue();
builder.append(" " + cmdEntry.getKey() + " " + usageInfo.args + "\n");
}
if (isHAEnabled) {
for (Map.Entry<String,UsageInfo> cmdEntry : USAGE.entrySet()) {
for (Map.Entry<String, UsageInfo> cmdEntry : YARN_HA_USAGE.entrySet()) {
String cmdKey = cmdEntry.getKey();
if (!cmdKey.equals("-help")) {
UsageInfo usageInfo = cmdEntry.getValue();
Expand Down Expand Up @@ -304,7 +315,7 @@ private static void printHelp(String cmd, boolean isHAEnabled) {
helpBuilder.append("\n");
}
if (isHAEnabled) {
for (String cmdKey : USAGE.keySet()) {
for (String cmdKey : YARN_HA_USAGE.keySet()) {
if (!cmdKey.equals("-help")) {
buildHelpMsg(cmdKey, helpBuilder);
helpBuilder.append("\n");
Expand All @@ -324,7 +335,7 @@ private static void printHelp(String cmd, boolean isHAEnabled) {
*/
private static void printUsage(String cmd, boolean isHAEnabled) {
StringBuilder usageBuilder = new StringBuilder();
if (ADMIN_USAGE.containsKey(cmd) || USAGE.containsKey(cmd)) {
if (ADMIN_USAGE.containsKey(cmd) || YARN_HA_USAGE.containsKey(cmd)) {
buildIndividualUsageMsg(cmd, usageBuilder);
} else {
buildUsageMsg(usageBuilder, isHAEnabled);
Expand Down Expand Up @@ -757,6 +768,28 @@ private int replaceLabelsOnNodes(Map<NodeId, Set<String>> map,
return 0;
}

private int failover(CommandLine cmd)
throws IOException, ServiceFailedException {
int numOpts = cmd.getOptions() == null ? 0 : cmd.getOptions().length;
final String[] args = cmd.getArgs();
if (numOpts > 1 || args.length != 2) {
errOut.println("failover: incorrect arguments");
printUsage(errOut, "-failover", YARN_HA_USAGE);
return -1;
}
HAServiceTarget fromNode = resolveTarget(args[0]);
HAServiceTarget toNode = resolveTarget(args[1]);
setRequestSource(HAServiceProtocol.RequestSource.REQUEST_BY_USER_FORCED);
HAServiceProtocol proto = fromNode.getProxy(
getConf(), 0);
HAServiceProtocolHelper.transitionToStandby(proto, createReqInfo());
HAServiceProtocol proto1 = toNode.getProxy(
getConf(), 0);
HAServiceProtocolHelper.transitionToActive(proto1, createReqInfo());
Copy link
Contributor

Choose a reason for hiding this comment

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

Do we need to check whether args[1] is the primary? If it is the primary, there is no need to switch.

Copy link
Contributor Author

Choose a reason for hiding this comment

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

If args[1] is the primary,it will return ：
HAServiceProtocolHelper.transitionToActive ->AdminService.java(transitionToActive):
"
@OverRide
public synchronized void transitionToActive(
HAServiceProtocol.StateChangeRequestInfo reqInfo) throws IOException {
if (isRMActive()) {
return;
}
....
"

out.println("Failover from "+args[0]+" to "+args[1]+" successful");
return 0;
}

@Override
public int run(String[] args) throws Exception {
YarnConfiguration yarnConf =
Expand Down Expand Up @@ -792,6 +825,18 @@ public int run(String[] args) throws Exception {
System.out.println("Cannot run " + cmd
+ " when ResourceManager HA is not enabled");
return -1;
}else if("-failover".equals(cmd)) {
if (isHAEnabled) {
Options opts = new Options();
CommandLine cmdLine = parseOpts(cmd, opts, args, YARN_HA_USAGE);
if (cmdLine == null) {
return -1;
}
return failover(cmdLine);
}
System.out.println("Cannot run " + cmd
+ " when ResourceManager HA is not enabled");
return -1;
}

//
Expand Down
23 changes: 13 additions & 10 deletions ...rn/hadoop-yarn-client/src/test/java/org/apache/hadoop/yarn/client/cli/TestRMAdminCLI.java
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -624,6 +624,16 @@ public void testTransitionToStandby() throws Exception {
any(HAServiceProtocol.StateChangeRequestInfo.class));
}

@Test
public void testFailover() throws Exception {
String[] args = {"-failover", "rm1", "rm2"};
// RM HA is disabled.
// failover should not be executed
assertEquals(-1, rmAdminCLI.run(args));
// failover should be executed
assertEquals(0, rmAdminCLIWithHAEnabled.run(args));
}

@Test
public void testGetServiceState() throws Exception {
String[] args = {"-getServiceState", "rm1"};
Expand Down Expand Up @@ -768,6 +778,8 @@ public void testHelp() throws Exception {
"Usage: yarn rmadmin [-refreshServiceAcl]", dataErr, 0);
testError(new String[] { "-help", "-getGroups" },
"Usage: yarn rmadmin [-getGroups [username]]", dataErr, 0);
testError(new String[] { "-help", "-failover" },
"Usage: yarn rmadmin [-failover <serviceId> <serviceId>]", dataErr, 0);
testError(new String[] { "-help", "-transitionToActive" },
"Usage: yarn rmadmin [-transitionToActive [--forceactive]" +
" <serviceId>]", dataErr, 0);
Expand Down Expand Up @@ -802,6 +814,7 @@ public void testHelp() throws Exception {
+ "([OvercommitTimeout]) "
+ "or -updateNodeResource [NodeID] [ResourceTypes] "
+ "([OvercommitTimeout])] "
+ "[-failover <serviceId> <serviceId>] "
+ "[-transitionToActive [--forceactive] <serviceId>] "
+ "[-transitionToStandby <serviceId>] "
+ "[-getServiceState <serviceId>] [-getAllServiceState] "
Expand Down Expand Up @@ -1083,21 +1096,11 @@ public void testNoUnsupportedHACommandsInHelp() throws Exception {
errOut = dataErr.toString();
assertTrue(errOut.contains("transitionToObserver: Unknown command"));
dataErr.reset();
args1[0] = "-failover";
assertEquals(-1, rmAdminCLIWithHAEnabled.run(args1));
errOut = dataErr.toString();
assertTrue(errOut.contains("failover: Unknown command"));
dataErr.reset();
String[] args2 = {"-help", "-transitionToObserver"};
assertEquals(0, rmAdminCLIWithHAEnabled.run(args2));
errOut = dataErr.toString();
assertFalse(errOut.contains("-transitionToObserver"));
dataErr.reset();
args2[1] = "-failover";
assertEquals(0, rmAdminCLIWithHAEnabled.run(args2));
errOut = dataErr.toString();
assertFalse(errOut.contains("-failover"));
dataErr.reset();
}

@Test
Expand Down