private static Map<String, Set<Path>> getPermMap(Configuration conf) {
String permLoc = conf.get("hdfsproxy.user.permissions.file.location", "user-permissions.xml");
if (conf.getResource(permLoc) == null) {
LOG.warn("HdfsProxy user permissions file not found");
return null;
}
Configuration permConf = new Configuration(false);
permConf.addResource(permLoc);
Map<String, Set<Path>> map = new HashMap<String, Set<Path>>();
for (Map.Entry<String, String> e : permConf) {
String k = e.getKey();
String v = e.getValue();
if (k != null && k.length() != 0 && v != null && v.length() != 0) {
Set<Path> pathSet = new HashSet<Path>();
String[] paths = v.split(",\\s*");
for (String p : paths) {
if (p.length() != 0) {
pathSet.add(new Path(p));
}
}
map.put(k, pathSet);
}
}
return map;
}
private static Map<String, Set<BigInteger>> getCertsMap(Configuration conf) {
String certsLoc = conf.get("hdfsproxy.user.certs.file.location", "user-certs.xml");
if (conf.getResource(certsLoc) == null) {
LOG.warn("HdfsProxy user certs file not found");
return null;
}
Configuration certsConf = new Configuration(false);
certsConf.addResource(certsLoc);
Map<String, Set<BigInteger>> map = new HashMap<String, Set<BigInteger>>();
for (Map.Entry<String, String> e : certsConf) {
String k = e.getKey();
String v = e.getValue().trim();
if (k != null && k.length() != 0 && v != null && v.length() != 0) {
Set<BigInteger> numSet = new HashSet<BigInteger>();
String[] serialnumbers = v.split("\\s*,\\s*");
for (String num : serialnumbers) {
if (num.length() != 0) {
numSet.add(new BigInteger(num, 16));
}
}
map.put(k, numSet);
}
}
return map;
}
public void setConf(Configuration conf) {
this.conf = conf;
this.tikaConfig = null;
// do we want a custom Tika configuration file
// deprecated since Tika 0.7 which is based on
// a service provider based configuration
String customConfFile = conf.get("tika.config.file");
if (customConfFile != null) {
try {
// see if a Tika config file can be found in the job file
URL customTikaConfig = conf.getResource(customConfFile);
if (customTikaConfig != null) tikaConfig = new TikaConfig(customTikaConfig);
} catch (Exception e1) {
String message = "Problem loading custom Tika configuration from " + customConfFile;
LOG.error(message, e1);
}
} else {
try {
tikaConfig = new TikaConfig(this.getClass().getClassLoader());
} catch (Exception e2) {
String message = "Problem loading default Tika configuration";
LOG.error(message, e2);
}
}
this.htmlParseFilters = new HtmlParseFilters(getConf());
this.utils = new DOMContentUtils(conf);
this.cachingPolicy =
getConf().get("parser.caching.forbidden.policy", Nutch.CACHING_FORBIDDEN_CONTENT);
}
/**
* Setup ssl configuration on the job configuration to enable hsftp access from map job. Also copy
* the ssl configuration file to Distributed cache
*
* @param configuration - Reference to job's configruation handle
* @throws java.io.IOException - Exception if unable to locate ssl config file
*/
private void setupSSLConfig(Configuration configuration) throws IOException {
Path sslConfigPath =
new Path(configuration.getResource(inputOptions.getSslConfigurationFile()).toString());
addSSLFilesToDistCache(configuration, sslConfigPath);
configuration.set(DistCpConstants.CONF_LABEL_SSL_CONF, sslConfigPath.getName());
configuration.set(DistCpConstants.CONF_LABEL_SSL_KEYSTORE, sslConfigPath.getName());
}
private void copyConfig(LlapOptions options, FileSystem lfs, Path confPath, String f)
throws IOException {
if (f.equals("llap-daemon-site.xml")) {
FSDataOutputStream confStream = lfs.create(new Path(confPath, f));
Configuration copy = resolve(conf, "llap-daemon-site.xml");
for (Entry<Object, Object> props : options.getConfig().entrySet()) {
// overrides
copy.set((String) props.getKey(), (String) props.getValue());
}
copy.writeXml(confStream);
confStream.close();
} else {
// they will be file:// URLs
lfs.copyFromLocalFile(new Path(conf.getResource(f).toString()), confPath);
}
}
private void run(String[] args) throws Exception {
LlapOptionsProcessor optionsProcessor = new LlapOptionsProcessor();
LlapOptions options = optionsProcessor.processOptions(args);
if (options == null) {
// help
return;
}
Path tmpDir = new Path(options.getDirectory());
if (conf == null) {
throw new Exception("Cannot load any configuration to run command");
}
FileSystem fs = FileSystem.get(conf);
FileSystem lfs = FileSystem.getLocal(conf).getRawFileSystem();
// needed so that the file is actually loaded into configuration.
for (String f : NEEDED_CONFIGS) {
conf.addResource(f);
if (conf.getResource(f) == null) {
throw new Exception("Unable to find required config file: " + f);
}
}
for (String f : OPTIONAL_CONFIGS) {
conf.addResource(f);
}
conf.reloadConfiguration();
if (options.getName() != null) {
// update service registry configs - caveat: this has nothing to do with the actual settings
// as read by the AM
// if needed, use --hiveconf llap.daemon.service.hosts=@llap0 to dynamically switch between
// instances
conf.set(ConfVars.LLAP_DAEMON_SERVICE_HOSTS.varname, "@" + options.getName());
}
if (options.getSize() != -1) {
if (options.getCache() != -1) {
Preconditions.checkArgument(
options.getCache() < options.getSize(),
"Cache has to be smaller than the container sizing");
}
if (options.getXmx() != -1) {
Preconditions.checkArgument(
options.getXmx() < options.getSize(),
"Working memory has to be smaller than the container sizing");
}
if (HiveConf.getBoolVar(conf, HiveConf.ConfVars.LLAP_ALLOCATOR_DIRECT)) {
Preconditions.checkArgument(
options.getXmx() + options.getCache() < options.getSize(),
"Working memory + cache has to be smaller than the containing sizing ");
}
}
final long minAlloc = conf.getInt(YarnConfiguration.RM_SCHEDULER_MINIMUM_ALLOCATION_MB, -1);
if (options.getSize() != -1) {
final long containerSize = options.getSize() / (1024 * 1024);
Preconditions.checkArgument(
containerSize >= minAlloc,
"Container size should be greater than minimum allocation(%s)",
minAlloc + "m");
conf.setLong(ConfVars.LLAP_DAEMON_YARN_CONTAINER_MB.varname, containerSize);
}
if (options.getExecutors() != -1) {
conf.setLong(ConfVars.LLAP_DAEMON_NUM_EXECUTORS.varname, options.getExecutors());
// TODO: vcpu settings - possibly when DRFA works right
}
if (options.getCache() != -1) {
conf.setLong(HiveConf.ConfVars.LLAP_IO_MEMORY_MAX_SIZE.varname, options.getCache());
}
if (options.getXmx() != -1) {
// Needs more explanation here
// Xmx is not the max heap value in JDK8
// You need to subtract 50% of the survivor fraction from this, to get actual usable memory
// before it goes into GC
conf.setLong(
ConfVars.LLAP_DAEMON_MEMORY_PER_INSTANCE_MB.varname,
(long) (options.getXmx()) / (1024 * 1024));
}
for (Entry<Object, Object> props : options.getConfig().entrySet()) {
conf.set((String) props.getKey(), (String) props.getValue());
}
URL logger = conf.getResource("llap-daemon-log4j2.properties");
if (null == logger) {
throw new Exception("Unable to find required config file: llap-daemon-log4j2.properties");
}
Path home = new Path(System.getenv("HIVE_HOME"));
Path scripts = new Path(new Path(new Path(home, "scripts"), "llap"), "bin");
if (!lfs.exists(home)) {
throw new Exception("Unable to find HIVE_HOME:" + home);
} else if (!lfs.exists(scripts)) {
LOG.warn("Unable to find llap scripts:" + scripts);
}
Path libDir = new Path(tmpDir, "lib");
String tezLibs = conf.get("tez.lib.uris");
if (tezLibs == null) {
LOG.warn("Missing tez.lib.uris in tez-site.xml");
}
if (LOG.isDebugEnabled()) {
LOG.debug("Copying tez libs from " + tezLibs);
}
lfs.mkdirs(libDir);
fs.copyToLocalFile(new Path(tezLibs), new Path(libDir, "tez.tar.gz"));
CompressionUtils.unTar(new Path(libDir, "tez.tar.gz").toString(), libDir.toString(), true);
lfs.delete(new Path(libDir, "tez.tar.gz"), false);
lfs.copyFromLocalFile(new Path(Utilities.jarFinderGetJar(LlapInputFormat.class)), libDir);
lfs.copyFromLocalFile(new Path(Utilities.jarFinderGetJar(HiveInputFormat.class)), libDir);
// copy default aux classes (json/hbase)
for (String className : DEFAULT_AUX_CLASSES) {
localizeJarForClass(lfs, libDir, className, false);
}
if (options.getIsHBase()) {
try {
localizeJarForClass(lfs, libDir, HBASE_SERDE_CLASS, true);
Job fakeJob = new Job(new JobConf()); // HBase API is convoluted.
TableMapReduceUtil.addDependencyJars(fakeJob);
Collection<String> hbaseJars = fakeJob.getConfiguration().getStringCollection("tmpjars");
for (String jarPath : hbaseJars) {
if (!jarPath.isEmpty()) {
lfs.copyFromLocalFile(new Path(jarPath), libDir);
}
}
} catch (Throwable t) {
String err = "Failed to add HBase jars. Use --auxhbase=false to avoid localizing them";
LOG.error(err);
System.err.println(err);
throw new RuntimeException(t);
}
}
String auxJars = options.getAuxJars();
if (auxJars != null && !auxJars.isEmpty()) {
// TODO: transitive dependencies warning?
String[] jarPaths = auxJars.split(",");
for (String jarPath : jarPaths) {
if (!jarPath.isEmpty()) {
lfs.copyFromLocalFile(new Path(jarPath), libDir);
}
}
}
Path confPath = new Path(tmpDir, "conf");
lfs.mkdirs(confPath);
for (String f : NEEDED_CONFIGS) {
copyConfig(options, lfs, confPath, f);
}
for (String f : OPTIONAL_CONFIGS) {
try {
copyConfig(options, lfs, confPath, f);
} catch (Throwable t) {
LOG.info("Error getting an optional config " + f + "; ignoring: " + t.getMessage());
}
}
lfs.copyFromLocalFile(new Path(logger.toString()), confPath);
// extract configs for processing by the python fragments in Slider
JSONObject configs = new JSONObject();
configs.put(
ConfVars.LLAP_DAEMON_YARN_CONTAINER_MB.varname,
HiveConf.getIntVar(conf, ConfVars.LLAP_DAEMON_YARN_CONTAINER_MB));
configs.put(
HiveConf.ConfVars.LLAP_IO_MEMORY_MAX_SIZE.varname,
HiveConf.getLongVar(conf, HiveConf.ConfVars.LLAP_IO_MEMORY_MAX_SIZE));
configs.put(
HiveConf.ConfVars.LLAP_ALLOCATOR_DIRECT.varname,
HiveConf.getBoolVar(conf, HiveConf.ConfVars.LLAP_ALLOCATOR_DIRECT));
configs.put(
ConfVars.LLAP_DAEMON_MEMORY_PER_INSTANCE_MB.varname,
HiveConf.getIntVar(conf, ConfVars.LLAP_DAEMON_MEMORY_PER_INSTANCE_MB));
configs.put(
ConfVars.LLAP_DAEMON_VCPUS_PER_INSTANCE.varname,
HiveConf.getIntVar(conf, ConfVars.LLAP_DAEMON_VCPUS_PER_INSTANCE));
configs.put(
ConfVars.LLAP_DAEMON_NUM_EXECUTORS.varname,
HiveConf.getIntVar(conf, ConfVars.LLAP_DAEMON_NUM_EXECUTORS));
configs.put(
YarnConfiguration.RM_SCHEDULER_MINIMUM_ALLOCATION_MB,
conf.getInt(YarnConfiguration.RM_SCHEDULER_MINIMUM_ALLOCATION_MB, -1));
configs.put(
YarnConfiguration.RM_SCHEDULER_MINIMUM_ALLOCATION_VCORES,
conf.getInt(YarnConfiguration.RM_SCHEDULER_MINIMUM_ALLOCATION_VCORES, -1));
FSDataOutputStream os = lfs.create(new Path(tmpDir, "config.json"));
OutputStreamWriter w = new OutputStreamWriter(os);
configs.write(w);
w.close();
os.close();
lfs.close();
fs.close();
if (LOG.isDebugEnabled()) {
LOG.debug("Exiting successfully");
}
}