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                 if (status.equals("OK")) {
  64                         long njobsOK = row.getLong("TotalOK") - 1;
  65                         String com4 = "DELETE FROM ProteinRow WHERE JobID = '" + jobid + "';";
  66                         System.out.println("Command: " + com4);
  67                         CassandraQuery(com4);
  68
  69                         String com5 = "DELETE FROM ProteinData WHERE JobID = '" + jobid + "' AND jobtime = " + date + ";";
  70                         System.out.println("Command: " + com5);
  71                         CassandraQuery(com5);
  72                         UpdateJobDateInfo(date, "TotalOK", njobsOK);
  73                 } else {
  74                         String com6 = "DELETE FROM FailLog WHERE JobID = '" + jobid + "' AND jobtime = " + date + ";";
  75                         System.out.println("Command: " + com6);
  76                         CassandraQuery(com6);
  77                         if (status.equals("STOPPED")) {
  78                                 long njobsStopped = row.getLong("TotalStopped") - 1;
  79                                 UpdateJobDateInfo(date, "TotalStopped", njobsStopped);
  80                         } else if (status.equals("ERROR")) {
  81                                 long njobsError = row.getLong("TotalError") - 1;
  82                                 UpdateJobDateInfo(date, "TotalError", njobsError);
  83                         } else if (status.equals("TIMEDOUT")) {
  84                                 long njobsTimeOut = row.getLong("TotalTimeOut") - 1;
  85                                 UpdateJobDateInfo(date, "TotalTimeOut", njobsTimeOut);
  86                         }
  87                 }
  88                 System.out.println("Job " + jobid + " removed...");
  89                 return 1;
  90         }
  91
  92         /**
  93          * update a pariticular column in the JobDateInfo table
  94          *
  95          * @param jobid
  96          *            job ID
  97          *
  98          * @return nothing
  99          *
 100          */
 101         private void UpdateJobDateInfo(long date, String ColumnName, long totalcol) {
 102                 String com = "UPDATE JobDateInfo SET " + ColumnName + " = " + totalcol + " WHERE jobday = " + date + ";";
 103                 CassandraQuery(com);
 104         }
 105
 106         /**
 107          * external method for deleting job with given job ID (strategy 1)
 108          *
 109          * @param jobid
 110          *            job ID
 111          *
 112          * @return a number of deleted jobs
 113          *
 114          */
 115         public int RemoveJobById(String jobid) {
 116                 if (jobid == null)
 117                         return 0;
 118                 long date = FindJobDate(jobid);
 119                 return RemoveJob(jobid, date);
 120         }
 121
 122         /**
 123          * external method for deleting jobs within a time range (strategy 4)
 124          *
 125          * @param date1
 126          *            starting date
 127          *
 128          * @param date2
 129          *            ending date
 130          *
 131          * @return a number of deleted jobs
 132          *
 133          */
 134         public int RemoveJobByDate(String date1, String date2) {
 135                 if (date1 == null || date2 == null)
 136                         return 0;
 137
 138                 int njobs = 0;
 139                 Long dateBegin = convertDate(date1);
 140                 Long dateEnd = convertDate(date2);
 141                 Calendar start = Calendar.getInstance();
 142                 start.setTime(new Date(dateBegin));
 143                 Calendar end = Calendar.getInstance();
 144                 end.setTime(new Date(dateEnd));
 145
 146                 for (Date date = start.getTime(); !start.after(end); start.add(Calendar.DATE, 1), date = start.getTime()) {
 147                         String com1 = "SELECT JobID FROM ProteinData WHERE jobtime = " + date.getTime() + ";";
 148                         System.out.println("Command: " + com1);
 149                         ResultSet results = CassandraQuery(com1);
 150                         if (!results.isExhausted()) {
 151                                 List<Row> rows = results.all();
 152                                 for (Row r : rows) {
 153                                         String jobid = r.getString("JobID");
 154                                         if (jobid != null) {
 155                                                 njobs += RemoveJob(jobid, date.getTime());
 156                                         }
 157                                 }
 158                         }
 159
 160                         String com2 = "SELECT JobID FROM FailLog WHERE jobtime = " + date.getTime() + ";";
 161                         ResultSet resultsfail = CassandraQuery(com2);
 162                         if (!resultsfail.isExhausted()) {
 163                                 List<Row> rows = resultsfail.all();
 164                                 for (Row r : rows) {
 165                                         String jobid = r.getString("JobID");
 166                                         if (jobid != null) {
 167                                                 njobs += RemoveJob(jobid, date.getTime());
 168                                         }
 169                                 }
 170                         }
 171                 }
 172                 return njobs;
 173         }
 174
 175         /**
 176          * external method for deleting jobs launched from a particular IP (strategy
 177          * 2)
 178          *
 179          * @param ip
 180          *            the IP
 181          *
 182          * @return a number of deleted jobs
 183          *
 184          */
 185         public int RemoveJobByIp(String ip) {
 186                 int njobs = 0;
 187                 if (ip == null)
 188                         return 0;
 189                 String com = "SELECT databegin, JobID FROM ProteinLog WHERE ip = '" + ip + "';";
 190                 ResultSet results = CassandraQuery(com);
 191                 if (!results.isExhausted()) {
 192                         List<Row> rows = results.all();
 193                         for (Row r : rows) {
 194                                 Long date = convertDate(r.getString("databegin"));
 195                                 String jobid = r.getString("JobID");
 196                                 if (date != null || jobid != null) {
 197                                         njobs += RemoveJob(jobid, date);
 198                                 }
 199                         }
 200                 }
 201                 return njobs;
 202         }
 203
 204         /**
 205          * external method for deleting jobs with a protein sequence (strategy 3)
 206          *
 207          * @param sequence
 208          *            the sequence
 209          *
 210          * @return a number of deleted jobs
 211          *
 212          */
 213         public int RemoveJobBySequence(String sequence) {
 214                 int njobs = 0;
 215                 if (sequence == null)
 216                         return 0;
 217                 String com = "SELECT JobID FROM ProteinRow WHERE Protein = '" + sequence + "';";
 218                 ResultSet results = CassandraQuery(com);
 219                 if (!results.isExhausted()) {
 220                         List<Row> rows = results.all();
 221                         for (Row r : rows) {
 222                                 String jobid = r.getString("JobID");
 223                                 long date = FindJobDate(jobid);
 224                                 njobs += RemoveJob(jobid, date);
 225                         }
 226                 }
 227                 return njobs;
 228         }
 229
 230         private long FindJobDate(String jobid) {
 231                 String com = "SELECT databegin FROM ProteinLog WHERE JobID = '" + jobid + "';";
 232                 ResultSet results = CassandraQuery(com);
 233                 if (!results.isExhausted()) {
 234                         return convertDate(results.one().getString("databegin"));
 235                 }
 236                 return -1L;
 237         }
 238
 239         private String FindStatus(String jobid) {
 240                 String com = "SELECT FinalStatus FROM ProteinLog WHERE JobID = '" + jobid + "';";
 241                 System.out.println("Command: " + com);
 242                 ResultSet results = CassandraQuery(com);
 243                 if (!results.isExhausted()) {
 244                         return results.one().getString("FinalStatus");
 245                 }
 246                 return "UNKNOWN";
 247         }
 248
 249         protected long convertDate(String d) {
 250                 try {
 251                         if (null != d) {
 252                                 Date startdate = dateformatter.parse(d);
 253                                 return startdate.getTime();
 254                         }
 255                 } catch (ParseException e) {
 256                         e.printStackTrace();
 257                 }
 258                 return 0L;
 259         }
 260
 261 }