/**
* Creates or force updates an unassigned node to the OFFLINE state for the specified region.
*
* <p>Attempts to create the node but if it exists will force it to transition to and OFFLINE
* state.
*
* <p>Sets a watcher on the unassigned region node if the method is successful.
*
* <p>This method should be used when assigning a region.
*
* @param zkw zk reference
* @param region region to be created as offline
* @param serverName server event originates from
* @throws KeeperException if unexpected zookeeper exception
* @throws KeeperException.NodeExistsException if node already exists
*/
public static boolean createOrForceNodeOffline(
ZooKeeperWatcher zkw, HRegionInfo region, String serverName) throws KeeperException {
LOG.debug(
zkw.prefix(
"Creating (or updating) unassigned node for "
+ region.getEncodedName()
+ " with OFFLINE state"));
RegionTransitionData data =
new RegionTransitionData(EventType.M_ZK_REGION_OFFLINE, region.getRegionName(), serverName);
synchronized (zkw.getNodes()) {
String node = getNodeName(zkw, region.getEncodedName());
zkw.sync(node);
zkw.getNodes().add(node);
int version = ZKUtil.checkExists(zkw, node);
if (version == -1) {
ZKUtil.createAndWatch(zkw, node, data.getBytes());
} else {
if (!ZKUtil.setData(zkw, node, data.getBytes(), version)) {
return false;
} else {
// We successfully forced to OFFLINE, reset watch and handle if
// the state changed in between our set and the watch
RegionTransitionData curData = ZKAssign.getData(zkw, region.getEncodedName());
if (curData.getEventType() != data.getEventType()) {
// state changed, need to process
return false;
}
}
}
}
return true;
}
public static void createNodeOffline(
ZooKeeperWatcher zkw, HRegionInfo region, String serverName, final EventType event)
throws KeeperException, KeeperException.NodeExistsException {
LOG.debug(
zkw.prefix(
"Creating unassigned node for " + region.getEncodedName() + " in OFFLINE state"));
RegionTransitionData data = new RegionTransitionData(event, region.getRegionName(), serverName);
synchronized (zkw.getNodes()) {
String node = getNodeName(zkw, region.getEncodedName());
zkw.getNodes().add(node);
ZKUtil.createAndWatch(zkw, node, data.getBytes());
}
}
/**
* Forces an existing unassigned node to the OFFLINE state for the specified region.
*
* <p>Does not create a new node. If a node does not already exist for this region, a {@link
* NoNodeException} will be thrown.
*
* <p>Sets a watcher on the unassigned region node if the method is successful.
*
* <p>This method should only be used during recovery of regionserver failure.
*
* @param zkw zk reference
* @param region region to be forced as offline
* @param serverName server event originates from
* @throws KeeperException if unexpected zookeeper exception
* @throws KeeperException.NoNodeException if node does not exist
*/
public static void forceNodeOffline(ZooKeeperWatcher zkw, HRegionInfo region, String serverName)
throws KeeperException, KeeperException.NoNodeException {
LOG.debug(
zkw.prefix(
"Forcing existing unassigned node for "
+ region.getEncodedName()
+ " to OFFLINE state"));
RegionTransitionData data =
new RegionTransitionData(EventType.M_ZK_REGION_OFFLINE, region.getRegionName(), serverName);
synchronized (zkw.getNodes()) {
String node = getNodeName(zkw, region.getEncodedName());
zkw.getNodes().add(node);
ZKUtil.setData(zkw, node, data.getBytes());
}
}
/**
* Creates a new unassigned node in the CLOSING state for the specified region.
*
* <p>Does not transition nodes from any states. If a node already exists for this region, a
* {@link NodeExistsException} will be thrown.
*
* <p>If creation is successful, returns the version number of the CLOSING node created.
*
* <p>Does not set any watches.
*
* <p>This method should only be used by a RegionServer when initiating a close of a region after
* receiving a CLOSE RPC from the Master.
*
* @param zkw zk reference
* @param region region to be created as closing
* @param serverName server event originates from
* @return version of node after transition, -1 if unsuccessful transition
* @throws KeeperException if unexpected zookeeper exception
* @throws KeeperException.NodeExistsException if node already exists
*/
public static int createNodeClosing(ZooKeeperWatcher zkw, HRegionInfo region, String serverName)
throws KeeperException, KeeperException.NodeExistsException {
LOG.debug(
zkw.prefix(
"Creating unassigned node for " + region.getEncodedName() + " in a CLOSING state"));
RegionTransitionData data =
new RegionTransitionData(
EventType.RS_ZK_REGION_CLOSING, region.getRegionName(), serverName);
synchronized (zkw.getNodes()) {
String node = getNodeName(zkw, region.getEncodedName());
zkw.getNodes().add(node);
return ZKUtil.createAndWatch(zkw, node, data.getBytes());
}
}
/**
* Creates an unassigned node in the OFFLINE state for the specified region.
*
* <p>Runs asynchronously. Depends on no pre-existing znode.
*
* <p>Sets a watcher on the unassigned region node.
*
* @param zkw zk reference
* @param region region to be created as offline
* @param serverName server event originates from
* @param cb
* @param ctx
* @throws KeeperException if unexpected zookeeper exception
* @throws KeeperException.NodeExistsException if node already exists
*/
public static void asyncCreateNodeOffline(
ZooKeeperWatcher zkw,
HRegionInfo region,
String serverName,
final AsyncCallback.StringCallback cb,
final Object ctx)
throws KeeperException {
LOG.debug(
zkw.prefix(
"Async create of unassigned node for "
+ region.getEncodedName()
+ " with OFFLINE state"));
RegionTransitionData data =
new RegionTransitionData(EventType.M_ZK_REGION_OFFLINE, region.getRegionName(), serverName);
synchronized (zkw.getNodes()) {
String node = getNodeName(zkw, region.getEncodedName());
zkw.getNodes().add(node);
ZKUtil.asyncCreate(zkw, node, data.getBytes(), cb, ctx);
}
}
/**
* Private method that actually performs unassigned node transitions.
*
* <p>Attempts to transition the unassigned node for the specified region from the expected state
* to the state in the specified transition data.
*
* <p>Method first reads existing data and verifies it is in the expected state. If the node does
* not exist or the node is not in the expected state, the method returns -1. If the transition is
* successful, the version number of the node following the transition is returned.
*
* <p>If the read state is what is expected, it attempts to write the new state and data into the
* node. When doing this, it includes the expected version (determined when the existing state was
* verified) to ensure that only one transition is successful. If there is a version mismatch, the
* method returns -1.
*
* <p>If the write is successful, no watch is set and the method returns true.
*
* @param zkw zk reference
* @param region region to be transitioned to opened
* @param serverName server event originates from
* @param endState state to transition node to if all checks pass
* @param beginState state the node must currently be in to do transition
* @param expectedVersion expected version of data before modification, or -1
* @return version of node after transition, -1 if unsuccessful transition
* @throws KeeperException if unexpected zookeeper exception
*/
public static int transitionNode(
ZooKeeperWatcher zkw,
HRegionInfo region,
String serverName,
EventType beginState,
EventType endState,
int expectedVersion)
throws KeeperException {
String encoded = region.getEncodedName();
if (LOG.isDebugEnabled()) {
LOG.debug(
zkw.prefix(
"Attempting to transition node "
+ HRegionInfo.prettyPrint(encoded)
+ " from "
+ beginState.toString()
+ " to "
+ endState.toString()));
}
String node = getNodeName(zkw, encoded);
zkw.sync(node);
// Read existing data of the node
Stat stat = new Stat();
byte[] existingBytes = ZKUtil.getDataNoWatch(zkw, node, stat);
if (existingBytes == null) {
// Node no longer exists. Return -1. It means unsuccessful transition.
return -1;
}
RegionTransitionData existingData = RegionTransitionData.fromBytes(existingBytes);
// Verify it is the expected version
if (expectedVersion != -1 && stat.getVersion() != expectedVersion) {
LOG.warn(
zkw.prefix(
"Attempt to transition the "
+ "unassigned node for "
+ encoded
+ " from "
+ beginState
+ " to "
+ endState
+ " failed, "
+ "the node existed but was version "
+ stat.getVersion()
+ " not the expected version "
+ expectedVersion));
return -1;
}
// Verify it is in expected state
if (!existingData.getEventType().equals(beginState)) {
LOG.warn(
zkw.prefix(
"Attempt to transition the "
+ "unassigned node for "
+ encoded
+ " from "
+ beginState
+ " to "
+ endState
+ " failed, "
+ "the node existed but was in the state "
+ existingData.getEventType()
+ " set by the server "
+ existingData.getServerName()));
return -1;
}
// Write new data, ensuring data has not changed since we last read it
try {
RegionTransitionData data =
new RegionTransitionData(endState, region.getRegionName(), serverName);
if (!ZKUtil.setData(zkw, node, data.getBytes(), stat.getVersion())) {
LOG.warn(
zkw.prefix(
"Attempt to transition the "
+ "unassigned node for "
+ encoded
+ " from "
+ beginState
+ " to "
+ endState
+ " failed, "
+ "the node existed and was in the expected state but then when "
+ "setting data we got a version mismatch"));
return -1;
}
if (LOG.isDebugEnabled()) {
LOG.debug(
zkw.prefix(
"Successfully transitioned node "
+ encoded
+ " from "
+ beginState
+ " to "
+ endState));
}
return stat.getVersion() + 1;
} catch (KeeperException.NoNodeException nne) {
LOG.warn(
zkw.prefix(
"Attempt to transition the "
+ "unassigned node for "
+ encoded
+ " from "
+ beginState
+ " to "
+ endState
+ " failed, "
+ "the node existed and was in the expected state but then when "
+ "setting data it no longer existed"));
return -1;
}
}
