Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

[HOPSWORKS-843] TensorBoard cleans up certificates twice #95

Merged
merged 2 commits into from
Dec 18, 2018
Merged

[HOPSWORKS-843] TensorBoard cleans up certificates twice #95

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
b15c38a
[HOPSWORKS-843] TensorBoard cleans up certificates twice
robzor92 Dec 18, 2018
37e6955
remove jupyter test
robzor92 Dec 18, 2018
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
30 changes: 12 additions & 18 deletions ...n/src/main/java/io/hops/hopsworks/common/dao/tensorflow/config/TensorBoardProcessMgr.java
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -143,19 +143,17 @@ public TensorBoardDTO startTensorBoard(Project project, Users user, HdfsUsers hd

while(retries > 0) {

if(retries == 0) {
throw new IOException("Failed to start TensorBoard for project=" + project.getName() + ", user="
+ user.getUid());
}
try {

// use pidfile to kill any running servers
port = ThreadLocalRandom.current().nextInt(40000, 59999);
if(retries == 0) {
throw new IOException("Failed to start TensorBoard for project=" + project.getName() + ", user="
+ user.getUid());
}

String[] command = new String[]{"/usr/bin/sudo", prog, "start", hdfsUser.getName(), hdfsLogdir,
tbPath, port.toString(), anacondaEnvironmentPath, settings.getHadoopVersion(), certsPath,
settings.getJavaHome()};
// use pidfile to kill any running servers
port = ThreadLocalRandom.current().nextInt(40000, 59999);

ProcessDescriptor processDescriptor = new ProcessDescriptor.Builder()
ProcessDescriptor processDescriptor = new ProcessDescriptor.Builder()
.addCommand("/usr/bin/sudo")
.addCommand(prog)
.addCommand("start")
Expand All @@ -169,9 +167,8 @@ public TensorBoardDTO startTensorBoard(Project project, Users user, HdfsUsers hd
.addCommand(settings.getJavaHome())
.ignoreOutErrStreams(true)
.build();
LOGGER.log(Level.FINE, processDescriptor.toString());

try {
LOGGER.log(Level.FINE, processDescriptor.toString());

ProcessResult processResult = osProcessExecutor.execute(processDescriptor);
if (!processResult.processExited()) {
throw new IOException("Tensorboard start process timed out!");
Expand All @@ -192,7 +189,7 @@ public TensorBoardDTO startTensorBoard(Project project, Users user, HdfsUsers hd
try {
host = InetAddress.getLocalHost().getHostAddress();
} catch (UnknownHostException ex) {
Logger.getLogger(TensorBoardProcessMgr.class.getName()).log(Level.SEVERE, null, ex);
LOGGER.log(Level.SEVERE, null, ex);
}
tensorBoardDTO.setEndpoint(host + ":" + port);
tensorBoardDTO.setPid(pid);
Expand Down Expand Up @@ -262,7 +259,6 @@ public int killTensorBoard(BigInteger pid) {
ProcessResult processResult = osProcessExecutor.execute(processDescriptor);
if (!processResult.processExited()) {
LOGGER.log(Level.SEVERE,"Failed to kill TensorBoard");
exitValue = 2;
}
exitValue = processResult.getExitCode();
} catch (IOException ex) {
Expand All @@ -278,7 +274,7 @@ public int killTensorBoard(BigInteger pid) {
* @return
*/
@TransactionAttribute(TransactionAttributeType.NOT_SUPPORTED)
public int killTensorBoard(TensorBoard tb) throws ServiceException {
public int killTensorBoard(TensorBoard tb) {

String prog = settings.getHopsworksDomainDir() + "/bin/tensorboard.sh";
int exitValue;
Expand All @@ -295,10 +291,8 @@ public int killTensorBoard(TensorBoard tb) throws ServiceException {
ProcessResult processResult = osProcessExecutor.execute(processDescriptor);
if (!processResult.processExited()) {
LOGGER.log(Level.SEVERE, "Failed to kill TensorBoard, process time-out");
exitValue = 2;
}
exitValue = processResult.getExitCode();
cleanupLocalTBDir(tb);
} catch (IOException ex) {
exitValue=2;
LOGGER.log(Level.SEVERE,"Failed to kill TensorBoard" , ex);
Expand Down
1 change: 0 additions & 1 deletion ...works-common/src/main/java/io/hops/hopsworks/common/security/CertificateMaterializer.java
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1124,7 +1124,6 @@ private boolean removeRemoteInternal(MaterialKey key, String remoteDirectory, bo
remoteMaterialReferencesFacade.delete(materialRef.getIdentifier());
deletedMaterial = true;
} else {
materialRef.decrementReferences();
remoteMaterialReferencesFacade.update(materialRef);
}
} else {
Expand Down
4 changes: 0 additions & 4 deletions hopsworks-ear/test/spec/projects_spec.rb
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -90,11 +90,7 @@
project_id = json_body[:projectId]
get "#{ENV['HOPSWORKS_API']}/project/#{project_id}/dataset/getContent"
expect_status(200)
jupyter = json_body.detect { |e| e[:name] == "Jupyter" }
notebook = json_body.detect { |e| e[:name] == "notebook" }
expect(jupyter[:description]).to eq ("Contains Jupyter notebooks.")
expect(jupyter[:permission]).to eq ("rwxrwx--T")
expect(jupyter[:owner]).to eq ("#{@user[:fname]} #{@user[:lname]}")
expect(notebook[:description]).to eq ("Contains Zeppelin notebooks.")
expect(notebook[:permission]).to eq ("rwxrwx--T")
expect(notebook[:owner]).to eq ("#{@user[:fname]} #{@user[:lname]}")
Expand Down