//fetch input data
func MoveInputData(work *core.Workunit) (size int64, err error) {
for inputname, io := range work.Inputs {
// skip if NoFile == true
if !io.NoFile { // is file !
var dataUrl string
inputFilePath := fmt.Sprintf("%s/%s", work.Path(), inputname)
if work.Rank == 0 {
if conf.CACHE_ENABLED && io.Node != "" {
if file_path, err := StatCacheFilePath(io.Node); err == nil {
//make a link in work dir from cached file
linkname := fmt.Sprintf("%s/%s", work.Path(), inputname)
fmt.Printf("input found in cache, making link: " + file_path + " -> " + linkname + "\n")
err = os.Symlink(file_path, linkname)
if err == nil {
logger.Event(event.FILE_READY, "workid="+work.Id+";url="+dataUrl)
}
return 0, err
}
}
dataUrl = io.DataUrl()
} else {
dataUrl = fmt.Sprintf("%s&index=%s&part=%s", io.DataUrl(), work.IndexType(), work.Part())
}
logger.Debug(2, "mover: fetching input file from url:"+dataUrl)
logger.Event(event.FILE_IN, "workid="+work.Id+";url="+dataUrl)
// download file
if datamoved, err := shock.FetchFile(inputFilePath, dataUrl, work.Info.DataToken, io.Uncompress); err != nil {
return size, err
} else {
size += datamoved
}
logger.Event(event.FILE_READY, "workid="+work.Id+";url="+dataUrl)
}
// download node attributes if requested
if io.AttrFile != "" {
// get node
node, err := shock.ShockGet(io.Host, io.Node, work.Info.DataToken)
if err != nil {
return size, err
}
logger.Debug(2, "mover: fetching input attributes from node:"+node.Id)
logger.Event(event.ATTR_IN, "workid="+work.Id+";node="+node.Id)
// print node attributes
attrFilePath := fmt.Sprintf("%s/%s", work.Path(), io.AttrFile)
attr_json, _ := json.Marshal(node.Attributes)
if err := ioutil.WriteFile(attrFilePath, attr_json, 0644); err != nil {
return size, err
}
logger.Event(event.ATTR_READY, "workid="+work.Id+";path="+attrFilePath)
}
}
return
}
//fetch input data
func moveInputData(work *core.Workunit) (size int64, err error) {
for _, io := range work.Inputs {
inputname := io.FileName
dataUrl, uerr := io.DataUrl()
if uerr != nil {
return 0, uerr
}
if work.Rank > 0 {
dataUrl = fmt.Sprintf("%s&index=%s&part=%s", dataUrl, work.IndexType(), work.Part())
}
inputFilePath := path.Join(work.Path(), inputname)
logger.Debug(2, "mover: fetching input from url:"+dataUrl)
logger.Event(event.FILE_IN, "workid="+work.Id+" url="+dataUrl)
// this gets file from any downloadable url, not just shock
if datamoved, _, err := shock.FetchFile(inputFilePath, dataUrl, work.Info.DataToken, io.Uncompress, false); err != nil {
return size, err
} else {
size += datamoved
}
logger.Event(event.FILE_READY, "workid="+work.Id+";url="+dataUrl)
}
return
}
//parse workunit, fetch input data, compose command arguments
func ParseWorkunitArgs(work *core.Workunit) (args []string, err error) {
argstr := work.Cmd.Args
if argstr == "" {
return
}
argList := strings.Fields(argstr)
inputsMap := work.Inputs
for _, arg := range argList {
if strings.Contains(arg, "@") { //parse input/output to accessible local file
segs := strings.Split(arg, "@")
if len(segs) > 2 {
return []string{}, errors.New("invalid format in command args, multiple @ within one arg")
}
inputname := segs[1]
if inputsMap.Has(inputname) {
io := inputsMap[inputname]
var dataUrl string
if work.Rank == 0 {
dataUrl = io.DataUrl()
} else {
dataUrl = fmt.Sprintf("%s&index=%s&part=%s", io.DataUrl(), work.IndexType(), work.Part())
}
inputFilePath := fmt.Sprintf("%s/%s", work.Path(), inputname)
logger.Debug(2, "mover: fetching input from url:"+dataUrl)
logger.Event(event.FILE_IN, "workid="+work.Id+" url="+dataUrl)
if err := fetchFile(inputFilePath, dataUrl, work.Info.DataToken); err != nil { //get file from Shock
return []string{}, err
}
logger.Event(event.FILE_READY, "workid="+work.Id+" url="+dataUrl)
parsedArg := fmt.Sprintf("%s%s", segs[0], inputFilePath)
args = append(args, parsedArg)
}
} else { //no @, has nothing to do with input/output, append directly
args = append(args, arg)
}
}
return args, nil
}
//fetch input data
func moveInputData(work *core.Workunit) (size int64, err error) {
for inputname, io := range work.Inputs {
var dataUrl string
if work.Rank == 0 {
dataUrl = io.DataUrl()
} else {
dataUrl = fmt.Sprintf("%s&index=%s&part=%s", io.DataUrl(), work.IndexType(), work.Part())
}
inputFilePath := path.Join(work.Path(), inputname)
logger.Debug(2, "mover: fetching input from url:"+dataUrl)
logger.Event(event.FILE_IN, "workid="+work.Id+" url="+dataUrl)
if datamoved, err := shock.FetchFile(inputFilePath, dataUrl, work.Info.DataToken, io.Uncompress); err != nil {
return size, err
} else {
size += datamoved
}
logger.Event(event.FILE_READY, "workid="+work.Id+";url="+dataUrl)
}
return
}
//fetch prerequisite data (e.g. reference dbs)
func movePreData(workunit *core.Workunit) (size int64, err error) {
for _, io := range workunit.Predata {
name := io.FileName
predata_directory := path.Join(conf.DATA_PATH, "predata")
err = os.MkdirAll(predata_directory, 755)
if err != nil {
return 0, errors.New("error creating predata_directory: " + err.Error())
}
file_path := path.Join(predata_directory, name)
dataUrl, uerr := io.DataUrl()
if uerr != nil {
return 0, uerr
}
// get shock and local md5sums
isShockPredata := true
node_md5 := ""
if io.Node == "-" {
isShockPredata = false
} else {
node, err := shock.ShockGet(io.Host, io.Node, workunit.Info.DataToken)
if err != nil {
return 0, errors.New("error in ShockGet: " + err.Error())
}
// rename file to be md5sum
node_md5 = node.File.Checksum["md5"]
file_path = path.Join(predata_directory, node_md5)
}
// file does not exist or its md5sum is wrong
if !isFileExisting(file_path) {
logger.Debug(2, "mover: fetching predata from url: "+dataUrl)
logger.Event(event.PRE_IN, "workid="+workunit.Id+" url="+dataUrl)
var md5sum string
file_path_part := file_path + ".part" // temporary name
// this gets file from any downloadable url, not just shock
size, md5sum, err = shock.FetchFile(file_path_part, dataUrl, workunit.Info.DataToken, io.Uncompress, isShockPredata)
if err != nil {
return 0, errors.New("error in fetchFile: " + err.Error())
}
os.Rename(file_path_part, file_path)
if err != nil {
return 0, errors.New("error renaming after download of preData: " + err.Error())
}
if isShockPredata {
if node_md5 != md5sum {
return 0, errors.New("error downloaded file md5 does not mach shock md5, node: " + io.Node)
} else {
logger.Debug(2, "mover: predata "+name+" has md5sum "+md5sum)
}
}
} else {
logger.Debug(2, "mover: predata already exists: "+name)
}
// timstamp for last access - future caching
accessfile, err := os.Create(file_path + ".access")
if err != nil {
return 0, errors.New("error creating predata access file: " + err.Error())
}
defer accessfile.Close()
accessfile.WriteString(time.Now().String())
// determine if running with docker
wants_docker := false
if workunit.Cmd.Dockerimage != "" || workunit.App != nil { // TODO need more save way to detect use of docker
wants_docker = true
}
if wants_docker && conf.USE_DOCKER == "no" {
return 0, errors.New("error: use of docker images has been disabled by administrator")
}
if wants_docker == false && conf.USE_DOCKER == "only" {
return 0, errors.New("error: use of docker images is enforced by administrator")
}
// copy or create symlink in work dir
linkname := path.Join(workunit.Path(), name)
if conf.NO_SYMLINK {
// some programs do not accept symlinks (e.g. emirge), need to copy the file into the work directory
logger.Debug(1, "copy predata: "+file_path+" -> "+linkname)
_, err := shock.CopyFile(file_path, linkname)
if err != nil {
return 0, fmt.Errorf("error copying file from %s to % s: ", file_path, linkname, err.Error())
}
} else {
if wants_docker {
// new filepath for predata dir in container
var docker_file_path string
if isShockPredata {
docker_file_path = path.Join(conf.DOCKER_WORKUNIT_PREDATA_DIR, node_md5)
} else {
docker_file_path = path.Join(conf.DOCKER_WORKUNIT_PREDATA_DIR, name)
}
logger.Debug(1, "creating dangling symlink: "+linkname+" -> "+docker_file_path)
// dangling link will give error, we ignore that here
_ = os.Symlink(docker_file_path, linkname)
} else {
logger.Debug(1, "symlink:"+linkname+" -> "+file_path)
err = os.Symlink(file_path, linkname)
if err != nil {
return 0, errors.New("error creating predata file symlink: " + err.Error())
}
}
}
logger.Event(event.PRE_READY, "workid="+workunit.Id+";url="+dataUrl)
}
return
}