\r
import java.io.File;\r
import java.io.FileFilter;\r
-import java.io.FileWriter;\r
import java.io.IOException;\r
import java.sql.SQLException;\r
import java.text.SimpleDateFormat;\r
import java.util.ArrayList;\r
+import java.util.Date;\r
import java.util.HashMap;\r
import java.util.HashSet;\r
import java.util.List;\r
\r
import org.apache.log4j.Logger;\r
\r
-import compbio.engine.client.ConfExecutable;\r
import compbio.engine.client.Executable;\r
-import compbio.engine.conf.PropertyHelperManager;\r
+import compbio.engine.client.PathValidator;\r
import compbio.metadata.JobStatus;\r
-import compbio.runner.msa.ClustalW;\r
import compbio.util.FileUtil;\r
-import compbio.util.PropertyHelper;\r
import compbio.ws.client.Services;\r
\r
/**\r
* @author pvtroshin\r
* \r
*/\r
-public class ExecutionStatCollector {\r
+public class ExecutionStatCollector implements Runnable {\r
\r
static final int UNDEFINED = -1;\r
\r
\r
static SimpleDateFormat DF = new SimpleDateFormat("dd/MM/yyyy hh:mm:ss");\r
\r
- static PropertyHelper ph = PropertyHelperManager.getPropertyHelper();\r
-\r
+ final private File workDirectory;\r
final private List<JobStat> stats;\r
+ /**\r
+ * Consider the job that has been working for longer than timeOutInHours\r
+ * completed, whatever the outcome\r
+ */\r
+ final private int timeOutInHours;\r
\r
- public ExecutionStatCollector(String workDirectory) {\r
- File[] files = FileUtil.getFiles(workDirectory, directories);\r
+ /**\r
+ * List subdirectories in the job directory\r
+ * \r
+ * @param workDirectory\r
+ * @param timeOutInHours\r
+ */\r
+ public ExecutionStatCollector(String workDirectory, int timeOutInHours) {\r
+ log.info("Starting stat collector for directory: " + workDirectory);\r
+ log.info("Maximum allowed runtime(h): " + timeOutInHours);\r
+ if (!PathValidator.isValidDirectory(workDirectory)) {\r
+ throw new IllegalArgumentException("workDirectory '"\r
+ + workDirectory + "' does not exist!");\r
+ }\r
+ this.workDirectory = new File(workDirectory);\r
stats = new ArrayList<JobStat>();\r
- for (File file : files) {\r
- JobDirectory jd = new JobDirectory(file);\r
- stats.add(jd.getJobStat());\r
- // System.out.println(jd.getJobStat().getJobReportTabulated());\r
+ if (timeOutInHours <= 0) {\r
+ throw new IllegalArgumentException(\r
+ "Timeout value must be greater than 0! Given value: "\r
+ + timeOutInHours);\r
}\r
+ this.timeOutInHours = timeOutInHours;\r
}\r
\r
- public StatProcessor getStats() {\r
+ boolean hasCompleted(JobDirectory jd) {\r
+ JobStat jstat = jd.getJobStat();\r
+ if (jstat.hasResult() || jstat.getIsCancelled()\r
+ || jstat.getIsFinished() || hasTimedOut(jd)) {\r
+ return true;\r
+ }\r
+ return false;\r
+ }\r
+\r
+ boolean hasTimedOut(JobDirectory jd) {\r
+ return ((System.currentTimeMillis() - jd.jobdir.lastModified()) / (1000 * 60 * 60)) > timeOutInHours;\r
+ }\r
+\r
+ StatProcessor getStats() {\r
return new StatProcessor(stats);\r
}\r
\r
- public void writeStatToDB() throws SQLException {\r
+ void writeStatToDB() throws SQLException {\r
Set<JobStat> rjobs = new HashSet<JobStat>(stats);\r
StatDB statdb = new StatDB();\r
+ log.debug("Removing records that has already been recorded");\r
+\r
statdb.removeRecordedJobs(rjobs);\r
+ log.debug("New records left: " + rjobs.size());\r
statdb.insertData(rjobs);\r
- statdb.conn.close();\r
- }\r
-\r
- static String getClusterJobDir() {\r
- String clusterdir = ph.getProperty("cluster.tmp.directory");\r
- if (clusterdir != null) {\r
- clusterdir.trim();\r
- }\r
- return clusterdir;\r
- }\r
-\r
- static void updateTime(File statFile) throws IOException {\r
- long lastMod = statFile.lastModified();\r
- FileWriter fw = new FileWriter(statFile);\r
- fw.write(new Long(lastMod).toString());\r
- fw.close();\r
}\r
\r
- static String getLocalJobDir() {\r
- String locdir = ph.getProperty("local.tmp.directory");\r
- if (locdir != null) {\r
- locdir.trim();\r
- }\r
- return locdir;\r
- }\r
+ /*\r
+ * static void updateTime(File statFile) throws IOException { long lastMod =\r
+ * statFile.lastModified(); FileWriter fw = new FileWriter(statFile);\r
+ * fw.write(new Long(lastMod).toString()); fw.close(); }\r
+ */\r
\r
/**\r
- * \r
- * @param args\r
- * @throws IOException\r
- * @throws SQLException\r
+ * Not in use\r
*/\r
public static void main(String[] args) throws IOException, SQLException {\r
\r
// updateTime(new File(\r
// "D:\\workspace\\JABA2\\jobsout\\AACon#170462904473672\\STARTED"));\r
\r
- String workDir = PropertyHelperManager.getLocalPath()\r
- + getLocalJobDir().trim();\r
- System.out.println(workDir);\r
File[] files = FileUtil.getFiles("Y:\\fc\\www-jws2\\jaba\\jobsout",\r
directories);\r
List<JobStat> stats = new ArrayList<JobStat>();\r
static FileFilter directories = new FileFilter() {\r
@Override\r
public boolean accept(File pathname) {\r
- return pathname.isDirectory();\r
+ return pathname.isDirectory()\r
+ && !pathname.getName().startsWith(".");\r
}\r
};\r
\r
File jobdir;\r
Map<String, File> files = new HashMap<String, File>();\r
\r
- public JobDirectory(File directory) {\r
+ JobDirectory(File directory) {\r
this.jobdir = directory;\r
for (File f : jobdir.listFiles()) {\r
files.put(f.getName(), f);\r
}\r
}\r
\r
- public boolean hasStatus(JobStatus status) {\r
+ boolean hasStatus(JobStatus status) {\r
return files.containsKey(status.toString());\r
}\r
\r
clustjobId = FileUtil.readFileToString(jobid);\r
}\r
} catch (IOException ioe) {\r
- ioe.printStackTrace();\r
- // TODO LOG\r
+ log.error(\r
+ "IO Exception while reading the content of JOBID file for job "\r
+ + jobid, ioe);\r
}\r
return clustjobId.trim();\r
}\r
return ftime;\r
}\r
\r
- @SuppressWarnings("unchecked")\r
- Class<Executable<?>> getWSRunnerName() {\r
- String name = jobdir.getName().split("#")[0];\r
- try {\r
- if (name.startsWith(ConfExecutable.CLUSTER_TASK_ID_PREFIX)) {\r
- assert ConfExecutable.CLUSTER_TASK_ID_PREFIX.length() == 1;\r
- name = name.substring(1);\r
- }\r
- name = ClustalW.class.getPackage().getName() + "." + name;\r
- return (Class<Executable<?>>) Class.forName(name);\r
- } catch (ClassNotFoundException e) {\r
- e.printStackTrace();\r
- throw new RuntimeException(\r
- "Cannot match the directory name to the executable! Executable name is "\r
- + name);\r
- }\r
- }\r
-\r
private Services getService() {\r
- return Services.getService(getWSRunnerName());\r
+ return Services.getServiceByJobDirectory(jobdir);\r
}\r
\r
// Mafft, Muscle, Tcoffee, Clustal task:fasta.in result:fasta.out\r
* TODO replace with Universal names for WS!\r
*/\r
long getResultSize() {\r
- Class<Executable<?>> name = getWSRunnerName();\r
+ Class<? extends Executable<?>> name = Services\r
+ .getRunnerByJobDirectory(jobdir);\r
+\r
File f = null;\r
if (name.getSimpleName().equalsIgnoreCase("Probcons")) {\r
f = files.get("alignment.out");\r
return false;\r
return true;\r
}\r
+ }\r
+\r
+ private void collectStatistics() {\r
+ File[] files = workDirectory.listFiles(directories);\r
+ for (File file : files) {\r
+ JobDirectory jd = new JobDirectory(file);\r
+ JobStat jstat = jd.getJobStat();\r
+ // Do not record stats on the job that has not completed yet\r
+ if (hasCompleted(jd)) {\r
+ stats.add(jstat);\r
+ } else {\r
+ log.debug("Skipping the job: " + jstat);\r
+ log.debug("As it has not completed yet");\r
+ }\r
+ // System.out.println(jd.getJobStat().getJobReportTabulated());\r
+ }\r
+ }\r
+\r
+ @Override\r
+ public void run() {\r
+ log.info("Started updating statistics at " + new Date());\r
\r
+ collectStatistics();\r
+\r
+ StatProcessor local_stats = getStats();\r
+ log.info("Found " + local_stats.getJobNumber() + " jobs!");\r
+ try {\r
+ writeStatToDB();\r
+ } catch (SQLException e) {\r
+ log.error("Fails to update jobs statistics database!");\r
+ log.error(e.getLocalizedMessage(), e);\r
+ }\r
+ log.info("Finished updating statistics at " + new Date());\r
}\r
}\r