private Map<JobId, Deployment> getTasks(final ZooKeeperClient client, final String host) {
final Map<JobId, Deployment> jobs = Maps.newHashMap();
try {
final String folder = Paths.configHostJobs(host);
final List<String> jobIds;
try {
jobIds = client.getChildren(folder);
} catch (KeeperException.NoNodeException e) {
return null;
}
for (final String jobIdString : jobIds) {
final JobId jobId = JobId.fromString(jobIdString);
final String containerPath = Paths.configHostJob(host, jobId);
try {
final byte[] data = client.getData(containerPath);
final Task task = parse(data, Task.class);
jobs.put(jobId, Deployment.of(jobId, task.getGoal()));
} catch (KeeperException.NoNodeException ignored) {
log.debug("deployment config node disappeared: {}", jobIdString);
}
}
} catch (KeeperException | IOException e) {
throw new HeliosRuntimeException("getting deployment config failed", e);
}
return jobs;
}
/** Returns a {@link Map} of {@link JobId} to {@link Job} objects for all of the jobs known. */
@Override
public Map<JobId, Job> getJobs() {
log.debug("getting jobs");
final String folder = Paths.configJobs();
final ZooKeeperClient client = provider.get("getJobs");
try {
final List<String> ids;
try {
ids = client.getChildren(folder);
} catch (NoNodeException e) {
return Maps.newHashMap();
}
final Map<JobId, Job> descriptors = Maps.newHashMap();
for (final String id : ids) {
final JobId jobId = JobId.fromString(id);
final String path = Paths.configJob(jobId);
final byte[] data = client.getData(path);
final Job descriptor = parse(data, Job.class);
descriptors.put(descriptor.getId(), descriptor);
}
return descriptors;
} catch (KeeperException | IOException e) {
throw new HeliosRuntimeException("getting jobs failed", e);
}
}
private void assertTaskExists(final ZooKeeperClient client, final String host, final JobId jobId)
throws JobNotDeployedException {
try {
client.getData(Paths.configHostJob(host, jobId));
} catch (NoNodeException e) {
throw new JobNotDeployedException(host, jobId);
} catch (KeeperException e) {
throw new HeliosRuntimeException(e);
}
}
private void assertHostExists(final ZooKeeperClient client, final String host)
throws HostNotFoundException {
try {
client.getData(Paths.configHost(host));
} catch (NoNodeException e) {
throw new HostNotFoundException(host, e);
} catch (KeeperException e) {
throw new HeliosRuntimeException(e);
}
}
private <T> T tryGetEntity(
final ZooKeeperClient client, String path, TypeReference<T> type, String name) {
try {
final byte[] data = client.getData(path);
return Json.read(data, type);
} catch (NoNodeException e) {
return null;
} catch (KeeperException | IOException e) {
throw new HeliosRuntimeException("reading " + name + " info failed", e);
}
}
private Job getJob(final ZooKeeperClient client, final JobId id) {
final String path = Paths.configJob(id);
try {
final byte[] data = client.getData(path);
return Json.read(data, Job.class);
} catch (NoNodeException e) {
// Return null to indicate that the job does not exist
return null;
} catch (KeeperException | IOException e) {
throw new HeliosRuntimeException("getting job " + id + " failed", e);
}
}
@Nullable
private TaskStatus getTaskStatus(
final ZooKeeperClient client, final String host, final JobId jobId) {
final String containerPath = Paths.statusHostJob(host, jobId);
try {
final byte[] data = client.getData(containerPath);
return parse(data, TaskStatus.class);
} catch (NoNodeException ignored) {
return null;
} catch (KeeperException | IOException e) {
throw new HeliosRuntimeException(
"Getting task " + jobId + " status " + "for host " + host + " failed", e);
}
}
/** Returns the current deployment state of {@code jobId} on {@code host}. */
@Override
public Deployment getDeployment(final String host, final JobId jobId) {
final String path = Paths.configHostJob(host, jobId);
final ZooKeeperClient client = provider.get("getDeployment");
try {
final byte[] data = client.getData(path);
final Task task = parse(data, Task.class);
return Deployment.of(jobId, task.getGoal());
} catch (KeeperException.NoNodeException e) {
return null;
} catch (KeeperException | IOException e) {
throw new HeliosRuntimeException("getting deployment failed", e);
}
}
/** Given a jobId, returns the N most recent events in it's history in the cluster. */
@Override
public List<TaskStatusEvent> getJobHistory(final JobId jobId) throws JobDoesNotExistException {
final Job descriptor = getJob(jobId);
if (descriptor == null) {
throw new JobDoesNotExistException(jobId);
}
final ZooKeeperClient client = provider.get("getJobHistory");
final List<String> hosts;
try {
hosts = client.getChildren(Paths.historyJobHosts(jobId));
} catch (NoNodeException e) {
return emptyList();
} catch (KeeperException e) {
throw Throwables.propagate(e);
}
final List<TaskStatusEvent> jsEvents = Lists.newArrayList();
for (String host : hosts) {
final List<String> events;
try {
events = client.getChildren(Paths.historyJobHostEvents(jobId, host));
} catch (KeeperException e) {
throw Throwables.propagate(e);
}
for (String event : events) {
try {
byte[] data =
client.getData(Paths.historyJobHostEventsTimestamp(jobId, host, Long.valueOf(event)));
final TaskStatus status = Json.read(data, TaskStatus.class);
jsEvents.add(new TaskStatusEvent(status, Long.valueOf(event), host));
} catch (NoNodeException e) { // ignore, it went away before we read it
} catch (KeeperException | IOException e) {
throw Throwables.propagate(e);
}
}
}
return Ordering.from(EVENT_COMPARATOR).sortedCopy(jsEvents);
}
// TODO(drewc): this kinda screams "long method"
private void deployJobRetry(
final ZooKeeperClient client, final String host, final Deployment deployment, int count)
throws JobDoesNotExistException, JobAlreadyDeployedException, HostNotFoundException,
JobPortAllocationConflictException {
if (count == 3) {
throw new HeliosRuntimeException(
"3 failures (possibly concurrent modifications) while " + "deploying. Giving up.");
}
log.info("deploying {}: {} (retry={})", deployment, host, count);
final JobId id = deployment.getJobId();
final Job job = getJob(id);
if (job == null) {
throw new JobDoesNotExistException(id);
}
final UUID operationId = UUID.randomUUID();
final String jobPath = Paths.configJob(id);
final String taskPath = Paths.configHostJob(host, id);
final String taskCreationPath = Paths.configHostJobCreation(host, id, operationId);
final List<Integer> staticPorts = staticPorts(job);
final Map<String, byte[]> portNodes = Maps.newHashMap();
final byte[] idJson = id.toJsonBytes();
for (final int port : staticPorts) {
final String path = Paths.configHostPort(host, port);
portNodes.put(path, idJson);
}
final Task task = new Task(job, deployment.getGoal());
final List<ZooKeeperOperation> operations =
Lists.newArrayList(
check(jobPath), create(portNodes), create(Paths.configJobHost(id, host)));
// Attempt to read a task here. If it's goal is UNDEPLOY, it's as good as not existing
try {
final Node existing = client.getNode(taskPath);
byte[] bytes = existing.getBytes();
Task readTask = Json.read(bytes, Task.class);
if (readTask.getGoal() != Goal.UNDEPLOY) {
throw new JobAlreadyDeployedException(host, id);
}
operations.add(check(taskPath, existing.getStat().getVersion()));
operations.add(set(taskPath, task));
} catch (NoNodeException e) {
operations.add(create(taskPath, task));
operations.add(create(taskCreationPath));
} catch (IOException | KeeperException e) {
throw new HeliosRuntimeException("reading existing task description failed", e);
}
// TODO (dano): Failure handling is racy wrt agent and job modifications.
try {
client.transaction(operations);
log.info("deployed {}: {} (retry={})", deployment, host, count);
} catch (NoNodeException e) {
// Either the job, the host or the task went away
assertJobExists(client, id);
assertHostExists(client, host);
// If the job and host still exists, we likely tried to redeploy a job that had an UNDEPLOY
// goal and lost the race with the agent removing the task before we could set it. Retry.
deployJobRetry(client, host, deployment, count + 1);
} catch (NodeExistsException e) {
// Check for conflict due to transaction retry
try {
if (client.exists(taskCreationPath) != null) {
// Our creation operation node existed, we're done here
return;
}
} catch (KeeperException ex) {
throw new HeliosRuntimeException("checking job deployment failed", ex);
}
try {
// Check if the job was already deployed
if (client.stat(taskPath) != null) {
throw new JobAlreadyDeployedException(host, id);
}
} catch (KeeperException ex) {
throw new HeliosRuntimeException("checking job deployment failed", e);
}
// Check for static port collisions
for (final int port : staticPorts) {
final String path = Paths.configHostPort(host, port);
try {
if (client.stat(path) == null) {
continue;
}
final byte[] b = client.getData(path);
final JobId existingJobId = parse(b, JobId.class);
throw new JobPortAllocationConflictException(id, existingJobId, host, port);
} catch (KeeperException | IOException ex) {
throw new HeliosRuntimeException("checking port allocations failed", e);
}
}
// Catch all for logic and ephemeral issues
throw new HeliosRuntimeException("deploying job failed", e);
} catch (KeeperException e) {
throw new HeliosRuntimeException("deploying job failed", e);
}
}