static int hdfsSingleNameNodeConnect(struct NativeMiniDfsCluster *cl, hdfsFS *fs,
const char *username)
{
int ret;
tPort port;
hdfsFS hdfs;
struct hdfsBuilder *bld;
port = (tPort)nmdGetNameNodePort(cl);
if (port < 0) {
fprintf(stderr, "hdfsSingleNameNodeConnect: nmdGetNameNodePort "
"returned error %d\n", port);
return port;
}
bld = hdfsNewBuilder();
if (!bld)
return -ENOMEM;
hdfsBuilderSetForceNewInstance(bld);
hdfsBuilderSetNameNode(bld, "localhost");
hdfsBuilderSetNameNodePort(bld, port);
hdfsBuilderConfSetStr(bld, "dfs.block.size",
TO_STR(TLH_DEFAULT_BLOCK_SIZE));
hdfsBuilderConfSetStr(bld, "dfs.blocksize",
TO_STR(TLH_DEFAULT_BLOCK_SIZE));
if (username) {
hdfsBuilderSetUserName(bld, username);
}
hdfs = hdfsBuilderConnect(bld);
if (!hdfs) {
ret = -errno;
return ret;
}
*fs = hdfs;
return 0;
}
hdfsFS hdfsConnectAsUser(const char* nn, tPort port, const char *user)
{
struct hdfsBuilder* bld = hdfsNewBuilder();
if (!bld) {
return NULL;
}
hdfsBuilderSetNameNode(bld, nn);
hdfsBuilderSetNameNodePort(bld, port);
hdfsBuilderSetUserName(bld, user);
return hdfsBuilderConnect(bld);
}
FileManagerHdfs::FileManagerHdfs(const string &storage_path)
: FileManager(storage_path) {
DEBUG_ASSERT(hdfs_namenode_port_dummy);
DEBUG_ASSERT(hdfs_num_replications_dummy);
struct hdfsBuilder *builder = hdfsNewBuilder();
hdfsBuilderSetNameNode(builder, FLAGS_hdfs_namenode_host.c_str());
hdfsBuilderSetNameNodePort(builder, FLAGS_hdfs_namenode_port);
// hdfsBuilderConnect releases builder.
hdfs_ = hdfsBuilderConnect(builder);
DEBUG_ASSERT(hdfs_ != nullptr);
}
hdfsFS hdfsConnectAsUserNewInstance(const char* host, tPort port,
const char *user)
{
struct hdfsBuilder *bld = hdfsNewBuilder();
if (!bld)
return NULL;
hdfsBuilderSetNameNode(bld, host);
hdfsBuilderSetNameNodePort(bld, port);
hdfsBuilderSetUserName(bld, user);
hdfsBuilderSetForceNewInstance(bld);
return hdfsBuilderConnect(bld);
}
hdfsFS Hdfs3FindConnection(const std::string& hostport) {
std::unique_lock<std::mutex> lock(s_hdfs_mutex);
auto it = s_hdfs_map.find(hostport);
if (it != s_hdfs_map.end())
return it->second;
// split host:port
std::vector<std::string> splitted = common::Split(hostport, ':', 2);
uint16_t port;
if (splitted.size() == 1) {
port = 8020;
}
else {
if (!common::from_str<uint16_t>(splitted[1], port))
die("Could not parse port in host:port \"" << hostport << "\"");
}
// split user@host
std::vector<std::string> user_split = common::Split(splitted[0], '@', 2);
const char* host, * user;
if (user_split.size() == 1) {
host = user_split[0].c_str();
user = nullptr;
}
else {
user = user_split[0].c_str();
host = user_split[1].c_str();
}
hdfsBuilder* builder = hdfsNewBuilder();
hdfsBuilderSetNameNode(builder, host);
hdfsBuilderSetNameNodePort(builder, port);
if (user)
hdfsBuilderSetUserName(builder, user);
hdfsFS hdfs = hdfsBuilderConnect(builder);
if (!hdfs)
die("Could not connect to HDFS server \"" << hostport << "\""
": " << hdfsGetLastError());
s_hdfs_map[hostport] = hdfs;
return hdfs;
}
int main(int argc, char* argv[]) {
if (argc < 4) {
printf("usage: hdfs_get <name node address> <name node port> <input file>\n");
return 1;
}
// Sleep for 100ms.
usleep(100 * 1000);
struct hdfsBuilder* hdfs_builder = hdfsNewBuilder();
if (!hdfs_builder) {
printf("Could not create HDFS builder");
return 1;
}
hdfsBuilderSetNameNode(hdfs_builder, argv[1]);
int port = atoi(argv[2]);
hdfsBuilderSetNameNodePort(hdfs_builder, port);
hdfsBuilderConfSetStr(hdfs_builder, "dfs.client.read.shortcircuit", "false");
hdfsFS fs = hdfsBuilderConnect(hdfs_builder);
hdfsFreeBuilder(hdfs_builder);
if (!fs) {
printf("Could not connect to HDFS");
return 1;
}
hdfsFile file_in = hdfsOpenFile(fs, argv[3], O_RDONLY, 0, 0, 0);
char buffer[1048576];
int done = 0;
do {
done = hdfsRead(fs, file_in, &buffer, 1048576);
} while (done > 0);
if (done < 0) {
printf("Failed to read file: %s", hdfsGetLastError());
return 1;
}
hdfsCloseFile(fs, file_in);
hdfsDisconnect(fs);
return 0;
}
