Improve IngestJob/IngestScheduler interation

This commit is contained in:
Richard Cordovano 2014-05-16 15:41:17 -04:00
parent 5130353ef3
commit f8d26589e0
2 changed files with 71 additions and 45 deletions

View File

@ -68,7 +68,7 @@ final class IngestJob {
List<IngestModuleError> errors = job.start();
if (errors.isEmpty()) {
IngestManager.getInstance().fireIngestJobStarted(jobId);
taskScheduler.scheduleTasksForIngestJob(job, dataSource);
taskScheduler.addTasksForIngestJob(job, dataSource);
} else {
ingestJobsById.remove(jobId);
}
@ -185,11 +185,13 @@ final class IngestJob {
if (!errors.isEmpty()) {
logIngestModuleErrors(errors);
}
} else {
taskScheduler.removeTasksForIngestJob(id);
}
// Because there is only one data source task per job, it is o.k. to
// call ProgressHandle.finish() now that the data source ingest modules
// are through using it via the DataSourceIngestModuleProgress wrapper.
// are through using the progress bar via the DataSourceIngestModuleProgress wrapper.
// Calling ProgressHandle.finish() again in finish() will be harmless.
dataSourceTasksProgress.finish();
@ -216,7 +218,10 @@ final class IngestJob {
if (!errors.isEmpty()) {
logIngestModuleErrors(errors);
}
} else {
taskScheduler.removeTasksForIngestJob(id);
}
if (taskScheduler.isLastTaskForIngestJob(task)) {
finish();
}

View File

@ -56,22 +56,15 @@ final class IngestScheduler {
private IngestScheduler() {
}
synchronized void scheduleTasksForIngestJob(IngestJob job, Content dataSource) throws InterruptedException {
synchronized void addTasksForIngestJob(IngestJob job, Content dataSource) throws InterruptedException {
// Enqueue a data source ingest task for the data source.
// If the thread executing this code is interrupted, tasksInProgressIterator is because the
// the number of ingest threads has been decreased while ingest jobs are
// running. The calling thread will exit in an orderly fashion, but the
// task still needs to be enqueued rather than lost, hence the loop.
DataSourceIngestTask task = new DataSourceIngestTask(job, dataSource);
while (true) {
try {
dataSourceTasks.put(task);
break;
} catch (InterruptedException ex) {
// Reset the interrupted status of the thread so the orderly
// exit can occur in the intended place.
Thread.currentThread().interrupt();
}
logger.log(Level.FINE, "Task scheduling for ingest job interrupted", ex); //NON-NLS
return;
}
// Get the top level files of the data source.
@ -86,12 +79,12 @@ final class IngestScheduler {
try {
children = root.getChildren();
if (children.isEmpty()) {
// Add the root object itself, tasksInProgressIterator could be an unallocated space
// Add the root object itself, it could be an unallocated space
// file, or a child of a volume or an image.
toptLevelFiles.add(root);
} else {
// The root object is a file system root directory, get
// the files within tasksInProgressIterator.
// the files within it.
for (Content child : children) {
if (child instanceof AbstractFile) {
toptLevelFiles.add((AbstractFile) child);
@ -118,7 +111,40 @@ final class IngestScheduler {
void addFileTaskToIngestJob(IngestJob job, AbstractFile file) {
FileIngestTask task = new FileIngestTask(job, file);
if (shouldEnqueueFileTask(task)) {
addTaskToFileQueue(task);
try {
fileTasks.put(task);
} catch (InterruptedException ex) {
Thread.currentThread().interrupt();
logger.log(Level.FINE, "Task scheduling for ingest job interrupted", ex); //NON-NLS
}
}
}
synchronized void removeTasksForIngestJob(long ingestJobId) {
// Remove all tasks for this ingest job that are not in progress.
Iterator<FileIngestTask> fileTasksIterator = fileTasks.iterator();
while (fileTasksIterator.hasNext()) {
if (fileTasksIterator.next().getIngestJob().getId() == ingestJobId) {
fileTasksIterator.remove();
}
}
Iterator<FileIngestTask> directoryTasksIterator = directoryTasks.iterator();
while (directoryTasksIterator.hasNext()) {
if (directoryTasksIterator.next().getIngestJob().getId() == ingestJobId) {
directoryTasksIterator.remove();
}
}
Iterator<FileIngestTask> rootDirectoryTasksIterator = rootDirectoryTasks.iterator();
while (rootDirectoryTasksIterator.hasNext()) {
if (rootDirectoryTasksIterator.next().getIngestJob().getId() == ingestJobId) {
rootDirectoryTasksIterator.remove();
}
}
Iterator<DataSourceIngestTask> dataSourceTasksIterator = dataSourceTasks.iterator();
while (dataSourceTasksIterator.hasNext()) {
if (dataSourceTasksIterator.next().getIngestJob().getId() == ingestJobId) {
dataSourceTasksIterator.remove();
}
}
}
@ -134,7 +160,7 @@ final class IngestScheduler {
if (fileTasks.isEmpty() == false) {
return;
}
// fill in the directory queue if tasksInProgressIterator is empty.
// fill in the directory queue if it is empty.
if (this.directoryTasks.isEmpty()) {
// bail out if root is also empty -- we are done
if (rootDirectoryTasks.isEmpty()) {
@ -149,7 +175,13 @@ final class IngestScheduler {
final AbstractFile parentFile = parentTask.getFile();
// add itself to the file list
if (shouldEnqueueFileTask(parentTask)) {
addTaskToFileQueue(parentTask);
try {
fileTasks.put(parentTask);
} catch (InterruptedException ex) {
Thread.currentThread().interrupt();
logger.log(Level.FINE, "Task scheduling for ingest job interrupted", ex); //NON-NLS
return;
}
}
// add its children to the file and directory lists
try {
@ -161,7 +193,13 @@ final class IngestScheduler {
if (childFile.hasChildren()) {
directoryTasks.add(childTask);
} else if (shouldEnqueueFileTask(childTask)) {
addTaskToFileQueue(childTask);
try {
fileTasks.put(childTask);
} catch (InterruptedException ex) {
Thread.currentThread().interrupt();
logger.log(Level.FINE, "Task scheduling for ingest job interrupted", ex); //NON-NLS
return;
}
}
}
}
@ -171,26 +209,9 @@ final class IngestScheduler {
}
}
private void addTaskToFileQueue(FileIngestTask task) {
// If the thread executing this code is interrupted, tasksInProgressIterator is because the
// the number of ingest threads has been decreased while ingest jobs are
// running. The calling thread will exit in an orderly fashion, but the
// task still needs to be enqueued rather than lost.
while (true) {
try {
fileTasks.put(task);
break;
} catch (InterruptedException ex) {
// Reset the interrupted status of the thread so the orderly
// exit can occur in the intended place.
Thread.currentThread().interrupt();
}
}
}
private static boolean shouldEnqueueFileTask(final FileIngestTask processTask) {
final AbstractFile aFile = processTask.getFile();
//if tasksInProgressIterator's unalloc file, skip if so scheduled
//if it's unalloc file, skip if so scheduled
if (processTask.getIngestJob().shouldProcessUnallocatedSpace() == false && aFile.getType().equals(TskData.TSK_DB_FILES_TYPE_ENUM.UNALLOC_BLOCKS)) {
return false;
}