datadb/compbio/cassandra/CassandraRemover.java

   1 package compbio.cassandra;
   2
   3 import java.text.ParseException;
   4 import java.text.SimpleDateFormat;
   5 import java.util.Calendar;
   6 import java.util.Date;
   7 import java.util.List;
   8
   9 import com.datastax.driver.core.ResultSet;
  10 import com.datastax.driver.core.Row;
  11 import compbio.cassandra.readers.CassandraReader;
  12
  13 /**
  14  * The class removes jobs from the cassandra database. 4 different strategies
  15  * are possiable: 1. remove 1 job with given job ID 2. remove jobs launched from
  16  * an IP 3. remove jobs with particular protein sequence 4. remove jobs launched
  17  * within a time range (date1, data2)
  18  *
  19  * @author Alexander Sherstnev
  20  * @author Natasha Sherstneva
  21  * @version 1.0
  22  * @since Nov 2013
  23  */
  24 public class CassandraRemover extends CassandraReader {
  25         static SimpleDateFormat dateformatter = new SimpleDateFormat("yyyy/MM/dd");
  26
  27         /**
  28          * private method for real deleting one job
  29          *
  30          * @param jobid
  31          *            job ID
  32          * @param date
  33          *            job execution date
  34          *
  35          * @return nothing
  36          */
  37         private int RemoveJob(String jobid, long date) {
  38
  39                 if (date < 0L) {
  40                         log.error("CassandraRemover error: job " + jobid + " with date " + date
  41                                         + " can not be deleted in JobDateInfo. Daily statistics is inconsistent");
  42                         return 0;
  43                 }
  44
  45                 String status = FindStatus(jobid);
  46                 String com1 = "DELETE FROM ProteinLog WHERE JobID = '" + jobid + "';";
  47                 System.out.println("Command: " + com1);
  48                 CassandraQuery(com1);
  49
  50                 String com2 = "UPDATE jpredarchive SET finalstatus = 'DELETED'  WHERE JobID = '" + jobid + "' ;";
  51                 System.out.println("Command: " + com2);
  52                 CassandraQuery(com2);
  53
  54                 String com3 = "SELECT * FROM JobDateInfo WHERE jobday = " + date + ";";
  55                 System.out.println("Command: " + com3);
  56                 ResultSet results = CassandraQuery(com3);
  57                 if (results.isExhausted()) {
  58                         log.error("CassandraRemover error: job " + jobid + " with date " + date
  59                                         + " can not be deleted in JobDateInfo. Daily statistics is inconsistent");
  60                         return 0;
  61                 }
  62                 Row row = results.one();
  63                 long njobs = row.getLong("Total") - 1;
  64                 if (status.equals("OK")) {
  65                         long njobsOK = row.getLong("TotalOK") - 1;
  66                         String com4 = "DELETE FROM ProteinRow WHERE JobID = '" + jobid + "';";
  67                         System.out.println("Command: " + com4);
  68                         CassandraQuery(com4);
  69
  70                         String com5 = "DELETE FROM ProteinData WHERE JobID = '" + jobid + "' AND jobtime = " + date + ";";
  71                         System.out.println("Command: " + com5);
  72                         CassandraQuery(com5);
  73                         UpdateJobDateInfo(date, "TotalOK", njobsOK, njobs);
  74                 } else {
  75                         String com6 = "DELETE FROM FailLog WHERE JobID = '" + jobid + "' AND jobtime = " + date + ";";
  76                         System.out.println("Command: " + com6);
  77                         CassandraQuery(com6);
  78                         if (status.equals("STOPPED")) {
  79                                 long njobsStopped = row.getLong("TotalStopped") - 1;
  80                                 UpdateJobDateInfo(date, "TotalStopped", njobsStopped, njobs);
  81                         } else if (status.equals("ERROR")) {
  82                                 long njobsError = row.getLong("TotalError") - 1;
  83                                 UpdateJobDateInfo(date, "TotalError", njobsError, njobs);
  84                         } else if (status.equals("TIMEDOUT")) {
  85                                 long njobsTimeOut = row.getLong("TotalTimeOut") - 1;
  86                                 UpdateJobDateInfo(date, "TotalTimeOut", njobsTimeOut, njobs);
  87                         }
  88                 }
  89                 System.out.println("Job " + jobid + " removed...");
  90                 return 1;
  91         }
  92
  93         /**
  94          * update a pariticular column in the JobDateInfo table
  95          *
  96          * @param jobid
  97          *            job ID
  98          *
  99          * @return nothing
 100          *
 101          */
 102         private void UpdateJobDateInfo(long date, String ColumnName, long totalcol, long total) {
 103                 String com = "UPDATE JobDateInfo SET " + ColumnName + " = " + totalcol + ", Total = " + total + " WHERE jobday = " + date + ";";
 104                 System.out.println("Command: " + com);
 105                 CassandraQuery(com);
 106         }
 107
 108         /**
 109          * external method for deleting job with given job ID (strategy 1)
 110          *
 111          * @param jobid
 112          *            job ID
 113          *
 114          * @return a number of deleted jobs
 115          *
 116          */
 117         public int RemoveJobById(String jobid) {
 118                 if (jobid == null)
 119                         return 0;
 120                 long date = FindJobDate(jobid);
 121                 return RemoveJob(jobid, date);
 122         }
 123
 124         /**
 125          * external method for deleting jobs within a time range (strategy 4)
 126          *
 127          * @param date1
 128          *            starting date
 129          *
 130          * @param date2
 131          *            ending date
 132          *
 133          * @return a number of deleted jobs
 134          *
 135          */
 136         public int RemoveJobByDate(String date1, String date2) {
 137                 if (date1 == null || date2 == null)
 138                         return 0;
 139
 140                 int njobs = 0;
 141                 Long dateBegin = convertDate(date1);
 142                 Long dateEnd = convertDate(date2);
 143                 Calendar start = Calendar.getInstance();
 144                 start.setTime(new Date(dateBegin));
 145                 Calendar end = Calendar.getInstance();
 146                 end.setTime(new Date(dateEnd));
 147
 148                 for (Date date = start.getTime(); !start.after(end); start.add(Calendar.DATE, 1), date = start.getTime()) {
 149                         String com1 = "SELECT JobID FROM ProteinData WHERE jobtime = " + date.getTime() + ";";
 150                         System.out.println("Command: " + com1);
 151                         ResultSet results = CassandraQuery(com1);
 152                         if (!results.isExhausted()) {
 153                                 List<Row> rows = results.all();
 154                                 for (Row r : rows) {
 155                                         String jobid = r.getString("JobID");
 156                                         if (jobid != null) {
 157                                                 njobs += RemoveJob(jobid, date.getTime());
 158                                         }
 159                                 }
 160                         }
 161
 162                         String com2 = "SELECT JobID FROM FailLog WHERE jobtime = " + date.getTime() + ";";
 163                         ResultSet resultsfail = CassandraQuery(com2);
 164                         if (!resultsfail.isExhausted()) {
 165                                 List<Row> rows = resultsfail.all();
 166                                 for (Row r : rows) {
 167                                         String jobid = r.getString("JobID");
 168                                         if (jobid != null) {
 169                                                 njobs += RemoveJob(jobid, date.getTime());
 170                                         }
 171                                 }
 172                         }
 173                 }
 174                 return njobs;
 175         }
 176
 177         /**
 178          * external method for deleting jobs launched from a particular IP (strategy
 179          * 2)
 180          *
 181          * @param ip
 182          *            the IP
 183          *
 184          * @return a number of deleted jobs
 185          *
 186          */
 187         public int RemoveJobByIp(String ip) {
 188                 int njobs = 0;
 189                 if (ip == null)
 190                         return 0;
 191                 String com = "SELECT databegin, JobID FROM ProteinLog WHERE ip = '" + ip + "';";
 192                 ResultSet results = CassandraQuery(com);
 193                 if (!results.isExhausted()) {
 194                         List<Row> rows = results.all();
 195                         for (Row r : rows) {
 196                                 Long date = convertDate(r.getString("databegin"));
 197                                 String jobid = r.getString("JobID");
 198                                 if (date != null || jobid != null) {
 199                                         njobs += RemoveJob(jobid, date);
 200                                 }
 201                         }
 202                 }
 203                 return njobs;
 204         }
 205
 206         /**
 207          * external method for deleting jobs with a protein sequence (strategy 3)
 208          *
 209          * @param sequence
 210          *            the sequence
 211          *
 212          * @return a number of deleted jobs
 213          *
 214          */
 215         public int RemoveJobBySequence(String sequence) {
 216                 int njobs = 0;
 217                 if (sequence == null)
 218                         return 0;
 219                 String com = "SELECT JobID FROM ProteinRow WHERE Protein = '" + sequence + "';";
 220                 ResultSet results = CassandraQuery(com);
 221                 if (!results.isExhausted()) {
 222                         List<Row> rows = results.all();
 223                         for (Row r : rows) {
 224                                 String jobid = r.getString("JobID");
 225                                 long date = FindJobDate(jobid);
 226                                 njobs += RemoveJob(jobid, date);
 227                         }
 228                 }
 229                 return njobs;
 230         }
 231
 232         private long FindJobDate(String jobid) {
 233                 String com = "SELECT databegin FROM ProteinLog WHERE JobID = '" + jobid + "';";
 234                 ResultSet results = CassandraQuery(com);
 235                 if (!results.isExhausted()) {
 236                         return convertDate(results.one().getString("databegin"));
 237                 }
 238                 return -1L;
 239         }
 240
 241         private String FindStatus(String jobid) {
 242                 String com = "SELECT FinalStatus FROM ProteinLog WHERE JobID = '" + jobid + "';";
 243                 System.out.println("Command: " + com);
 244                 ResultSet results = CassandraQuery(com);
 245                 if (!results.isExhausted()) {
 246                         return results.one().getString("FinalStatus");
 247                 }
 248                 return "UNKNOWN";
 249         }
 250
 251         protected long convertDate(String d) {
 252                 try {
 253                         if (null != d) {
 254                                 Date startdate = dateformatter.parse(d);
 255                                 return startdate.getTime();
 256                         }
 257                 } catch (ParseException e) {
 258                         e.printStackTrace();
 259                 }
 260                 return 0L;
 261         }
 262
 263 }