/**
* call-seq:
* hdfs.new(options={}) -> hdfs
*
* Creates a new HDFS client connection, configured by options, returning a new
* HDFS::FileSystem object if successful. If this fails, raises a
* ConnectError.
*
* options can have the following keys:
*
* * *local*: whether to use the local filesystem instead of HDFS
* (default: false)
* * *host*: hostname or IP address of a Hadoop NameNode (default: '0.0.0.0')
* * *port*: port through which to connect to Hadoop NameNode (default: 8020)
* * *user*: user to connect to filesystem as (default: current user)
*/
VALUE HDFS_File_System_initialize(int argc, VALUE* argv, VALUE self) {
VALUE options;
rb_scan_args(argc, argv, "01", &options);
// Sets default values for keyword args, type-checks supplied value.
options = NIL_P(options) ? rb_hash_new() : options;
if (TYPE(options) != T_HASH) {
rb_raise(rb_eArgError, "options must be of type Hash");
}
FSData* data = NULL;
Data_Get_Struct(self, FSData, data);
VALUE r_local = rb_hash_aref(options, rb_eval_string(":local"));
VALUE r_user = rb_hash_aref(options, rb_eval_string(":user"));
if (r_local == Qtrue) {
if (NIL_P(r_user)) {
data->fs = hdfsConnect(NULL, 0);
} else {
data->fs = hdfsConnectAsUser(NULL, 0, StringValuePtr(r_user));
}
rb_iv_set(self, "@local", Qtrue);
} else {
VALUE r_host = rb_hash_aref(options, rb_eval_string(":host"));
VALUE r_port = rb_hash_aref(options, rb_eval_string(":port"));
// Sets default values for host and port if not supplied by user.
char* hdfs_host = RTEST(r_host) ? StringValuePtr(r_host) :
(char*) HDFS_DEFAULT_HOST;
int hdfs_port = RTEST(r_port) ? NUM2INT(r_port) :
HDFS_DEFAULT_PORT;
if (NIL_P(r_user)) {
data->fs = hdfsConnect(hdfs_host, hdfs_port);
} else {
data->fs = hdfsConnectAsUser(hdfs_host, hdfs_port,
StringValuePtr(r_user));
rb_iv_set(self, "@user", rb_str_new2(StringValuePtr(r_user)));
}
rb_iv_set(self, "@local", Qfalse);
rb_iv_set(self, "@host", rb_str_new2(hdfs_host));
rb_iv_set(self, "@port", INT2NUM(hdfs_port));
}
if (data->fs == NULL) {
rb_raise(e_connect_error, "Failed to connect to HDFS: %s",
get_error(errno));
return Qnil;
}
return self;
}
/**
* hdfsConnect
*
* Left as part of the libhdfs API/ABI, but this will always fail with
* libhadoofus. Use hdfsConnectAsUser() instead.
*/
hdfsFS
hdfsConnect(const char* host, tPort port)
{
const char *user = getenv("HDFS_DEFAULT_USER");
if (!user) {
ERR(EINVAL, "Set HDFS_DEFAULT_USER if you must use "
"hdfsConnect()");
return NULL;
}
return hdfsConnectAsUser(host, port, user);
}
bool libhdfsconnector::connect ()
{
fs = NULL;
if (strlen(hdfsuser) > 0)
fs = hdfsConnectAsUser(hadoopHost, hadoopPort, hdfsuser);
else
fs = hdfsConnect(hadoopHost, hadoopPort);
if (!fs)
{
fprintf(stderr, "Error: Could not connect to hdfs via LIBHDFS on %s:%d\n", hadoopHost, hadoopPort);
return false;
}
return true;
};
static int hdfsSingleNameNodeConnect(struct NativeMiniDfsCluster *cluster,
hdfsFS *fs)
{
int nnPort;
const char *nnHost;
hdfsFS hdfs;
if (nmdGetNameNodeHttpAddress(cluster, &nnPort, &nnHost)) {
fprintf(stderr, "Error when retrieving namenode host address.\n");
return 1;
}
hdfs = hdfsConnectAsUser(nnHost, nnPort, user);
if(!hdfs) {
fprintf(stderr, "Oops! Failed to connect to hdfs!\n");
return 1;
}
*fs = hdfs;
return 0;
}
int main(int argc, char **argv) {
hdfsFS fs = hdfsConnect("default", 0);
if(!fs) {
fprintf(stderr, "Oops! Failed to connect to hdfs!\n");
exit(-1);
}
hdfsFS lfs = hdfsConnect(NULL, 0);
if(!lfs) {
fprintf(stderr, "Oops! Failed to connect to 'local' hdfs!\n");
exit(-1);
}
const char* writePath = "/tmp/testfile.txt";
{
//Write tests
hdfsFile writeFile = hdfsOpenFile(fs, writePath, O_WRONLY|O_CREAT, 0, 0, 0);
if(!writeFile) {
fprintf(stderr, "Failed to open %s for writing!\n", writePath);
exit(-1);
}
fprintf(stderr, "Opened %s for writing successfully...\n", writePath);
char* buffer = "Hello, World!";
tSize num_written_bytes = hdfsWrite(fs, writeFile, (void*)buffer, strlen(buffer)+1);
fprintf(stderr, "Wrote %d bytes\n", num_written_bytes);
tOffset currentPos = -1;
if ((currentPos = hdfsTell(fs, writeFile)) == -1) {
fprintf(stderr,
"Failed to get current file position correctly! Got %ld!\n",
currentPos);
exit(-1);
}
fprintf(stderr, "Current position: %ld\n", currentPos);
if (hdfsFlush(fs, writeFile)) {
fprintf(stderr, "Failed to 'flush' %s\n", writePath);
exit(-1);
}
fprintf(stderr, "Flushed %s successfully!\n", writePath);
hdfsCloseFile(fs, writeFile);
}
{
//Read tests
const char* readPath = "/tmp/testfile.txt";
int exists = hdfsExists(fs, readPath);
if (exists) {
fprintf(stderr, "Failed to validate existence of %s\n", readPath);
exit(-1);
}
hdfsFile readFile = hdfsOpenFile(fs, readPath, O_RDONLY, 0, 0, 0);
if (!readFile) {
fprintf(stderr, "Failed to open %s for reading!\n", readPath);
exit(-1);
}
fprintf(stderr, "hdfsAvailable: %d\n", hdfsAvailable(fs, readFile));
tOffset seekPos = 1;
if(hdfsSeek(fs, readFile, seekPos)) {
fprintf(stderr, "Failed to seek %s for reading!\n", readPath);
exit(-1);
}
tOffset currentPos = -1;
if((currentPos = hdfsTell(fs, readFile)) != seekPos) {
fprintf(stderr,
"Failed to get current file position correctly! Got %ld!\n",
currentPos);
exit(-1);
}
fprintf(stderr, "Current position: %ld\n", currentPos);
static char buffer[32];
tSize num_read_bytes = hdfsRead(fs, readFile, (void*)buffer,
sizeof(buffer));
fprintf(stderr, "Read following %d bytes:\n%s\n",
num_read_bytes, buffer);
num_read_bytes = hdfsPread(fs, readFile, 0, (void*)buffer,
sizeof(buffer));
fprintf(stderr, "Read following %d bytes:\n%s\n",
num_read_bytes, buffer);
hdfsCloseFile(fs, readFile);
}
int totalResult = 0;
int result = 0;
{
//Generic file-system operations
const char* srcPath = "/tmp/testfile.txt";
const char* dstPath = "/tmp/testfile2.txt";
fprintf(stderr, "hdfsCopy(remote-local): %s\n", ((result = hdfsCopy(fs, srcPath, lfs, srcPath)) ? "Failed!" : "Success!"));
totalResult += result;
fprintf(stderr, "hdfsCopy(remote-remote): %s\n", ((result = hdfsCopy(fs, srcPath, fs, dstPath)) ? "Failed!" : "Success!"));
totalResult += result;
fprintf(stderr, "hdfsMove(local-local): %s\n", ((result = hdfsMove(lfs, srcPath, lfs, dstPath)) ? "Failed!" : "Success!"));
totalResult += result;
fprintf(stderr, "hdfsMove(remote-local): %s\n", ((result = hdfsMove(fs, srcPath, lfs, srcPath)) ? "Failed!" : "Success!"));
totalResult += result;
fprintf(stderr, "hdfsRename: %s\n", ((result = hdfsRename(fs, dstPath, srcPath)) ? "Failed!" : "Success!"));
totalResult += result;
fprintf(stderr, "hdfsCopy(remote-remote): %s\n", ((result = hdfsCopy(fs, srcPath, fs, dstPath)) ? "Failed!" : "Success!"));
totalResult += result;
const char* slashTmp = "/tmp";
const char* newDirectory = "/tmp/newdir";
fprintf(stderr, "hdfsCreateDirectory: %s\n", ((result = hdfsCreateDirectory(fs, newDirectory)) ? "Failed!" : "Success!"));
totalResult += result;
fprintf(stderr, "hdfsSetReplication: %s\n", ((result = hdfsSetReplication(fs, srcPath, 2)) ? "Failed!" : "Success!"));
totalResult += result;
char buffer[256];
const char *resp;
fprintf(stderr, "hdfsGetWorkingDirectory: %s\n", ((resp = hdfsGetWorkingDirectory(fs, buffer, sizeof(buffer))) ? buffer : "Failed!"));
totalResult += (resp ? 0 : 1);
fprintf(stderr, "hdfsSetWorkingDirectory: %s\n", ((result = hdfsSetWorkingDirectory(fs, slashTmp)) ? "Failed!" : "Success!"));
totalResult += result;
fprintf(stderr, "hdfsGetWorkingDirectory: %s\n", ((resp = hdfsGetWorkingDirectory(fs, buffer, sizeof(buffer))) ? buffer : "Failed!"));
totalResult += (resp ? 0 : 1);
fprintf(stderr, "hdfsGetDefaultBlockSize: %ld\n", hdfsGetDefaultBlockSize(fs));
fprintf(stderr, "hdfsGetCapacity: %ld\n", hdfsGetCapacity(fs));
fprintf(stderr, "hdfsGetUsed: %ld\n", hdfsGetUsed(fs));
hdfsFileInfo *fileInfo = NULL;
if((fileInfo = hdfsGetPathInfo(fs, slashTmp)) != NULL) {
fprintf(stderr, "hdfsGetPathInfo - SUCCESS!\n");
fprintf(stderr, "Name: %s, ", fileInfo->mName);
fprintf(stderr, "Type: %c, ", (char)(fileInfo->mKind));
fprintf(stderr, "Replication: %d, ", fileInfo->mReplication);
fprintf(stderr, "BlockSize: %ld, ", fileInfo->mBlockSize);
fprintf(stderr, "Size: %ld, ", fileInfo->mSize);
fprintf(stderr, "LastMod: %s", ctime(&fileInfo->mLastMod));
fprintf(stderr, "Owner: %s, ", fileInfo->mOwner);
fprintf(stderr, "Group: %s, ", fileInfo->mGroup);
char permissions[10];
permission_disp(fileInfo->mPermissions, permissions);
fprintf(stderr, "Permissions: %d (%s)\n", fileInfo->mPermissions, permissions);
hdfsFreeFileInfo(fileInfo, 1);
} else {
totalResult++;
fprintf(stderr, "waah! hdfsGetPathInfo for %s - FAILED!\n", slashTmp);
}
hdfsFileInfo *fileList = 0;
int numEntries = 0;
if((fileList = hdfsListDirectory(fs, slashTmp, &numEntries)) != NULL) {
int i = 0;
for(i=0; i < numEntries; ++i) {
fprintf(stderr, "Name: %s, ", fileList[i].mName);
fprintf(stderr, "Type: %c, ", (char)fileList[i].mKind);
fprintf(stderr, "Replication: %d, ", fileList[i].mReplication);
fprintf(stderr, "BlockSize: %ld, ", fileList[i].mBlockSize);
fprintf(stderr, "Size: %ld, ", fileList[i].mSize);
fprintf(stderr, "LastMod: %s", ctime(&fileList[i].mLastMod));
fprintf(stderr, "Owner: %s, ", fileList[i].mOwner);
fprintf(stderr, "Group: %s, ", fileList[i].mGroup);
char permissions[10];
permission_disp(fileList[i].mPermissions, permissions);
fprintf(stderr, "Permissions: %d (%s)\n", fileList[i].mPermissions, permissions);
}
hdfsFreeFileInfo(fileList, numEntries);
} else {
if (errno) {
totalResult++;
fprintf(stderr, "waah! hdfsListDirectory - FAILED!\n");
} else {
fprintf(stderr, "Empty directory!\n");
}
}
char*** hosts = hdfsGetHosts(fs, srcPath, 0, 1);
if(hosts) {
fprintf(stderr, "hdfsGetHosts - SUCCESS! ... \n");
int i=0;
while(hosts[i]) {
int j = 0;
while(hosts[i][j]) {
fprintf(stderr,
"\thosts[%d][%d] - %s\n", i, j, hosts[i][j]);
++j;
}
++i;
}
} else {
totalResult++;
fprintf(stderr, "waah! hdfsGetHosts - FAILED!\n");
}
char *newOwner = "root";
// setting tmp dir to 777 so later when connectAsUser nobody, we can write to it
short newPerm = 0666;
// chown write
fprintf(stderr, "hdfsChown: %s\n", ((result = hdfsChown(fs, writePath, NULL, "users")) ? "Failed!" : "Success!"));
totalResult += result;
fprintf(stderr, "hdfsChown: %s\n", ((result = hdfsChown(fs, writePath, newOwner, NULL)) ? "Failed!" : "Success!"));
totalResult += result;
// chmod write
fprintf(stderr, "hdfsChmod: %s\n", ((result = hdfsChmod(fs, writePath, newPerm)) ? "Failed!" : "Success!"));
totalResult += result;
sleep(2);
tTime newMtime = time(NULL);
tTime newAtime = time(NULL);
// utime write
fprintf(stderr, "hdfsUtime: %s\n", ((result = hdfsUtime(fs, writePath, newMtime, newAtime)) ? "Failed!" : "Success!"));
totalResult += result;
// chown/chmod/utime read
hdfsFileInfo *finfo = hdfsGetPathInfo(fs, writePath);
fprintf(stderr, "hdfsChown read: %s\n", ((result = (strcmp(finfo->mOwner, newOwner) != 0)) ? "Failed!" : "Success!"));
totalResult += result;
fprintf(stderr, "hdfsChmod read: %s\n", ((result = (finfo->mPermissions != newPerm)) ? "Failed!" : "Success!"));
totalResult += result;
// will later use /tmp/ as a different user so enable it
fprintf(stderr, "hdfsChmod: %s\n", ((result = hdfsChmod(fs, "/tmp/", 0777)) ? "Failed!" : "Success!"));
totalResult += result;
fprintf(stderr,"newMTime=%ld\n",newMtime);
fprintf(stderr,"curMTime=%ld\n",finfo->mLastMod);
fprintf(stderr, "hdfsUtime read (mtime): %s\n", ((result = (finfo->mLastMod != newMtime)) ? "Failed!" : "Success!"));
totalResult += result;
// No easy way to turn on access times from hdfs_test right now
// fprintf(stderr, "hdfsUtime read (atime): %s\n", ((result = (finfo->mLastAccess != newAtime)) ? "Failed!" : "Success!"));
// totalResult += result;
hdfsFreeFileInfo(finfo, 1);
// Clean up
fprintf(stderr, "hdfsDelete: %s\n", ((result = hdfsDelete(fs, newDirectory)) ? "Failed!" : "Success!"));
totalResult += result;
fprintf(stderr, "hdfsDelete: %s\n", ((result = hdfsDelete(fs, srcPath)) ? "Failed!" : "Success!"));
totalResult += result;
fprintf(stderr, "hdfsDelete: %s\n", ((result = hdfsDelete(lfs, srcPath)) ? "Failed!" : "Success!"));
totalResult += result;
fprintf(stderr, "hdfsDelete: %s\n", ((result = hdfsDelete(lfs, dstPath)) ? "Failed!" : "Success!"));
totalResult += result;
fprintf(stderr, "hdfsExists: %s\n", ((result = hdfsExists(fs, newDirectory)) ? "Success!" : "Failed!"));
totalResult += (result ? 0 : 1);
}
totalResult += (hdfsDisconnect(fs) != 0);
{
//
// Now test as connecting as a specific user
// This is only meant to test that we connected as that user, not to test
// the actual fs user capabilities. Thus just create a file and read
// the owner is correct.
const char *tuser = "******";
const char* writePath = "/tmp/usertestfile.txt";
const char **groups = (const char**)malloc(sizeof(char*)* 2);
groups[0] = "users";
groups[1] = "nobody";
fs = hdfsConnectAsUser("default", 0, tuser, groups, 2);
if(!fs) {
fprintf(stderr, "Oops! Failed to connect to hdfs as user %s!\n",tuser);
exit(-1);
}
hdfsFile writeFile = hdfsOpenFile(fs, writePath, O_WRONLY|O_CREAT, 0, 0, 0);
if(!writeFile) {
fprintf(stderr, "Failed to open %s for writing!\n", writePath);
exit(-1);
}
fprintf(stderr, "Opened %s for writing successfully...\n", writePath);
char* buffer = "Hello, World!";
tSize num_written_bytes = hdfsWrite(fs, writeFile, (void*)buffer, strlen(buffer)+1);
fprintf(stderr, "Wrote %d bytes\n", num_written_bytes);
if (hdfsFlush(fs, writeFile)) {
fprintf(stderr, "Failed to 'flush' %s\n", writePath);
exit(-1);
}
fprintf(stderr, "Flushed %s successfully!\n", writePath);
hdfsCloseFile(fs, writeFile);
hdfsFileInfo *finfo = hdfsGetPathInfo(fs, writePath);
fprintf(stderr, "hdfs new file user is correct: %s\n", ((result = (strcmp(finfo->mOwner, tuser) != 0)) ? "Failed!" : "Success!"));
totalResult += result;
}
totalResult += (hdfsDisconnect(fs) != 0);
if (totalResult != 0) {
return -1;
} else {
return 0;
}
}
hdfsFS hdfsConnect(const char* nn, tPort port)
{
return hdfsConnectAsUser(nn, port, NULL);
}
