//fetch prerequisite data (e.g. reference dbs)
func movePreData(workunit *core.Workunit) (size int64, err error) {
for name, io := range workunit.Predata {
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)
if !isFileExisting(file_path) {
size, err = shock.FetchFile(file_path, io.Url, workunit.Info.DataToken, io.Uncompress)
if err != nil {
return 0, errors.New("error in fetchFile:" + err.Error())
}
}
use_symlink := false
linkname := path.Join(workunit.Path(), name)
if workunit.Cmd.Dockerimage != "" || strings.HasPrefix(workunit.Cmd.Name, "app:") { // TODO need more save way to detect use of docker
use_symlink = false // TODO mechanism
if use_symlink {
file_path = path.Join(conf.DOCKER_WORKUNIT_PREDATA_DIR, name)
// some tasks want to write in predata dir, thus need symlink
logger.Debug(1, "dangling symlink:"+linkname+" -> "+file_path)
// creation of dangling symlinks is not possible with with os.Symlink, thus use system ln
link_out, err := exec.Command("ln", "-s", file_path, linkname).CombinedOutput()
logger.Debug(1, fmt.Sprintf("ln returned %s", link_out))
if err != nil {
return 0, errors.New("error creating predata file symlink (dangling version): " + err.Error())
}
} else {
// some programs do not accept symlinks (e.g. emirge), need to copy the file into the work directory
// linkname refers to target file now.
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 {
//linkname := path.Join(workunit.Path(), name)
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())
}
}
}
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
}
