Implemented OStoreDB::RetrieveMount::getNextJob()

objectstore/ArchiveToFileRequest.hpp

@@ -49,15 +49,7 @@ public:
     uint16_t totalFailures;
   };
   FailuresCount addJobFailure(uint16_t copyNumber, uint64_t sessionId);
-  // Duplication of the protbuff statuses
-  enum class JobStatus {
-    AJS_LinkingToTapePool = 0,
-    AJS_Pending = 1,
-    AJS_Selected = 2,
-    AJS_Complete = 3,
-    AJS_Failed = 99
-  };
-  JobStatus getJobStatus(uint16_t copyNumber);
+  serializers::ArchiveJobStatus getJobStatus(uint16_t copyNumber);
   // Handling of the consequences of a job status change for the entire request.
   // This function returns true if the request got finished.
   bool finishIfNecessary();

objectstore/RetrieveToFileRequest.cpp

@@ -80,6 +80,11 @@ void cta::objectstore::RetrieveToFileRequest::setPriority(uint64_t priority) {
   m_payload.set_priority(priority);
 }
 
+uint64_t cta::objectstore::RetrieveToFileRequest::getPriority() {
+  checkPayloadReadable();
+  return m_payload.priority();
+}
+
 void cta::objectstore::RetrieveToFileRequest::setCreationLog(
   const objectstore::CreationLog& creationLog) {
   checkPayloadWritable();
@@ -120,4 +125,24 @@ uint64_t cta::objectstore::RetrieveToFileRequest::getSize() {
 void cta::objectstore::RetrieveToFileRequest::setSize(uint64_t size) {
   checkPayloadWritable();
   m_payload.set_size(size);
-}
\ No newline at end of file
+}
+
+auto  cta::objectstore::RetrieveToFileRequest::getJob(uint16_t copyNb) -> JobDump {
+  checkPayloadReadable();
+  // find the job
+  auto & jl = m_payload.jobs();
+  for (auto j=jl.begin(); j!=jl.end(); j++) {
+    if (j->copynb() == copyNb) {
+      JobDump ret;
+      ret.blockid = j->blockid();
+      ret.copyNb = j->copynb();
+      ret.fseq = j->fseq();
+      ret.tape = j->tape();
+      ret.tapeAddress = j->tapeaddress();
+      return ret;
+    }
+  }
+  throw NoSuchJob("In objectstore::RetrieveToFileRequest::getJob(): job not found for this copyNb");
+}
+
+

objectstore/RetrieveToFileRequest.hpp

@@ -35,26 +35,53 @@ public:
   RetrieveToFileRequest(const std::string & address, Backend & os);
   RetrieveToFileRequest(GenericObject & go);
   void initialize();
+  // Job management ============================================================
   void addJob(const cta::TapeFileLocation & tapeFileLocation,
     const std::string & tapeaddress);
+  void setJobFailureLimits(uint16_t copyNumber,
+    uint16_t maxRetiesWithinMount, uint16_t maxTotalRetries);
+  void setJobSelected(uint16_t copyNumber, const std::string & owner);
+  void setJobPending(uint16_t copyNumber);
+  class JobDump {
+  public:
+    uint16_t copyNb;
+    std::string tape;
+    std::string tapeAddress;
+    uint64_t fseq;
+    uint64_t blockid;
+  };
+  JobDump getJob(uint16_t copyNb);
+  struct FailuresCount {
+    uint16_t failuresWithinMount;
+    uint16_t totalFailures;
+  };
+  FailuresCount addJobFailure(uint16_t copyNumber, uint64_t sessionId);
+  serializers::RetrieveJobStatus getJobStatus(uint16_t copyNumber);
+  // Handling of the consequences of a job status. This is simpler that archival
+  // as one finish is enough.
+  void finish();
+  // Mark all jobs as pending mount (following their linking to a tape pool)
+  void setAllJobsLinkingToTapePool();
+  // Mark all the jobs as being deleted, in case of a cancellation
+  void setAllJobsFailed();
+  // Mark all the jobs as pending deletion from NS.
+  void setAllJobsPendingNSdeletion();
+  CTA_GENERATE_EXCEPTION_CLASS(NoSuchJob);
+  // Request management ========================================================
+  void setSuccessful();
+  void setFailed();
+  // ===========================================================================
   void setArchiveFile(const std::string & archiveFile);
   std::string getArchiveFile();
   void setRemoteFile (const std::string & remoteFile);
   std::string getRemoteFile();
   void setPriority (uint64_t priority);
+  uint64_t getPriority();
   void setCreationLog (const objectstore::CreationLog& creationLog);
   CreationLog getCreationLog();
   void setRetrieveToDirRequestAddress(const std::string & dirRequestAddress);
   void setSize(uint64_t size);
   uint64_t getSize();
-  class JobDump {
-  public:
-    uint16_t copyNb;
-    std::string tape;
-    std::string tapeAddress;
-    uint64_t fseq;
-    uint64_t blockid;
-  };
   std::list<JobDump> dumpJobs();
 };
 

objectstore/Tape.cpp

@@ -207,6 +207,20 @@ auto cta::objectstore::Tape::dumpAndFetchRetrieveRequests()
     } catch (cta::exception::Exception &) {}
   }
   return ret;
+  }
+
+auto cta::objectstore::Tape::dumpJobs() -> std::list<JobDump> {
+  checkPayloadReadable();
+  std::list<JobDump> ret;
+  auto & rjl = m_payload.retrievejobs();
+  for (auto rj=rjl.begin(); rj!=rjl.end(); rj++) {
+    ret.push_back(JobDump());
+    auto & b=ret.back();
+    b.copyNb = rj->copynb();
+    b.address = rj->address();
+    b.size = rj->size();
+  }
+  return ret;
 }
 
 
@@ -280,6 +294,30 @@ void cta::objectstore::Tape::releaseBusy() {
   m_payload.set_busy(false);
 }
 
+void cta::objectstore::Tape::removeJob(const std::string& retriveToFileAddress) {
+  checkPayloadWritable();
+  auto * jl = m_payload.mutable_retrievejobs();
+  bool found=false;
+  do {
+    found=false;
+    found = false;
+    // Push the found entry all the way to the end.
+    for (size_t i=0; i<(size_t)jl->size(); i++) {
+      if (jl->Get(i).address() == retriveToFileAddress) {
+        found = true;
+        while (i+1 < (size_t)jl->size()) {
+          jl->SwapElements(i, i+1);
+          i++;
+        }
+        break;
+      }
+    }
+    // and remove it
+    if (found)
+      jl->RemoveLast();
+  } while (found);
+}
+
 
 
 

objectstore/Tape.hpp

@@ -78,6 +78,14 @@ public:
   };
   JobsSummary getJobsSummary();
   std::list<RetrieveRequestDump> dumpAndFetchRetrieveRequests();
+  struct JobDump {
+    std::string address;
+    uint16_t copyNb;
+    uint64_t size;
+  };
+  std::list<JobDump> dumpJobs();
+  
+  void removeJob(const std::string & retriveToFileAddress);
   
   // -- Stored data counting ---------------------------------------------------
   uint64_t getStoredData();

scheduler/ArchiveMount.cpp

@@ -91,7 +91,7 @@ std::unique_ptr<cta::ArchiveJob> cta::ArchiveMount::getNextJob() {
   std::unique_ptr<cta::SchedulerDatabase::ArchiveJob> dbJob(m_dbMount->getNextJob().release());
   if (!dbJob.get())
     return std::unique_ptr<cta::ArchiveJob>(NULL);
-  // We have something to migrate: prepare the response
+  // We have something to archive: prepare the response
   std::unique_ptr<cta::ArchiveJob> ret(new ArchiveJob(*this, m_ns,
       dbJob->archiveFile, dbJob->remoteFile, dbJob->nameServerTapeFile));
   ret->m_dbJob.reset(dbJob.release());

scheduler/OStoreDB/OStoreDB.cpp

@@ -1351,9 +1351,9 @@ auto OStoreDB::ArchiveMount::getNextJob() -> std::unique_ptr<SchedulerDatabase::
   while (tp.dumpJobs().size()) {
     // Get the tape pool's jobs list, and pop the first
     auto jl = tp.dumpJobs();
-    // First take a lock on a download the job
+    // First take a lock on and download the job
     // If the request is not around anymore, we will just move the the next
-    // prepare the return value
+    // Prepare the return value
     std::unique_ptr<OStoreDB::ArchiveJob> privateRet(new OStoreDB::ArchiveJob(
       jl.front().address, m_objectStore, m_agent));
     privateRet->m_copyNb = jl.front().copyNb;
@@ -1439,12 +1439,90 @@ const OStoreDB::RetrieveMount::MountInfo& OStoreDB::RetrieveMount::getMountInfo(
   return mountInfo;
 }
 
-auto  OStoreDB::RetrieveMount::getNextJob() -> std::unique_ptr<RetrieveJob> {
-  throw NotImplemented("In OStoreDB::RetrieveMount::getNextJob: not implemented");
+auto OStoreDB::RetrieveMount::getNextJob() -> std::unique_ptr<SchedulerDatabase::RetrieveJob> {
+  // Find the next file to retrieve
+  // Get the tape pool and then tape
+  objectstore::RootEntry re(m_objectStore);
+  objectstore::ScopedSharedLock rel(re);
+  re.fetch();
+  auto tpl = re.dumpTapePools();
+  rel.release();
+  std::string tpAddress;
+  for (auto tpp = tpl.begin(); tpp != tpl.end(); tpp++) {
+    if (tpp->tapePool == mountInfo.tapePool)
+      tpAddress = tpp->address;
+  }
+  if (!tpAddress.size())
+    throw NoSuchTapePool("In OStoreDB::RetrieveMount::getNextJob(): tape pool not found");
+  objectstore::TapePool tp(tpAddress, m_objectStore);
+  objectstore::ScopedSharedLock tplock(tp);
+  tp.fetch();
+  auto tl = tp.dumpTapes();
+  tplock.release();
+  std::string tAddress;
+  for (auto tptr = tl.begin(); tptr != tl.end(); tptr++) {
+    if (tptr->vid == mountInfo.vid)
+      tAddress = tptr->address;
+  }
+  if (!tAddress.size())
+    throw NoSuchTape("In OStoreDB::RetrieveMount::getNextJob(): tape not found");
+  objectstore::Tape t(tAddress, m_objectStore);
+  objectstore::ScopedExclusiveLock tlock(t);
+  t.fetch();
+  while (t.dumpJobs().size()) {
+    // Get the tape pool's jobs list, and pop the first
+    auto jl=t.dumpJobs();
+    // First take a lock on and download the job
+    // If the request is not around anymore, we will just move the the next
+    // Prepare the return value
+    std::unique_ptr<OStoreDB::RetrieveJob> privateRet(new OStoreDB::RetrieveJob(
+      jl.front().address, m_objectStore, m_agent));
+    privateRet->m_copyNb = jl.front().copyNb;
+    objectstore::ScopedExclusiveLock rtfrl;
+    try {
+      rtfrl.lock(privateRet->m_rtfr);
+      privateRet->m_rtfr.fetch();
+    } catch (cta::exception::Exception &) {
+      // we failed to access the object. It might be missing.
+      // Just pop this job from the pool and move to the next.
+      t.removeJob(privateRet->m_rtfr.getAddressIfSet());
+      // Commit in case we do not pass by again.
+      t.commit();
+      continue;
+    }
+    // Take ownership of the job
+    // Add to ownership
+    objectstore::ScopedExclusiveLock al(m_agent);
+    m_agent.fetch();
+    m_agent.addToOwnership(privateRet->m_rtfr.getAddressIfSet());
+    m_agent.commit();
+    al.release();
+    // Make the ownership official (for the whole request in retrieves)
+    privateRet->m_rtfr.setOwner(m_agent.getAddressIfSet());
+    privateRet->m_rtfr.commit();
+    // Remove the job from the tape pool queue
+    t.removeJob(privateRet->m_rtfr.getAddressIfSet());
+    // We can commit and release the tape pool lock, we will only fill up
+    // memory structure from here on.
+    t.commit();
+    privateRet->archiveFile.path = privateRet->m_rtfr.getArchiveFile();
+    privateRet->remoteFile = privateRet->m_rtfr.getRemoteFile();
+    objectstore::RetrieveToFileRequest::JobDump jobDump=
+        privateRet->m_rtfr.getJob(privateRet->m_copyNb);
+    privateRet->nameServerTapeFile.tapeFileLocation.fSeq = jobDump.fseq;
+    privateRet->nameServerTapeFile.tapeFileLocation.blockId = jobDump.blockid;
+    privateRet->nameServerTapeFile.tapeFileLocation.copyNb = privateRet->m_copyNb;
+    privateRet->nameServerTapeFile.tapeFileLocation.vid = mountInfo.vid;
+
+      std::numeric_limits<decltype(privateRet->nameServerTapeFile.tapeFileLocation.blockId)>::max();
+    privateRet->m_jobOwned = true;
+    return std::unique_ptr<SchedulerDatabase::RetrieveJob> (privateRet.release());
+  }
+  return std::unique_ptr<SchedulerDatabase::RetrieveJob> (NULL);
 }
 
 void OStoreDB::RetrieveMount::complete(time_t completionTime) {
-  throw NotImplemented("In OStoreDB::RetrieveMount::getNextJob: not implemented");
+  throw NotImplemented("In OStoreDB::RetrieveMount::complete: not implemented");
 }
 
 
@@ -1529,5 +1607,88 @@ OStoreDB::ArchiveJob::~ArchiveJob() {
   }
 }
 
+OStoreDB::RetrieveJob::RetrieveJob(const std::string& jobAddress, 
+    objectstore::Backend& os, objectstore::Agent& ag): m_jobOwned(false),
+  m_objectStore(os), m_agent(ag), m_rtfr(jobAddress, os) { }
+
+void OStoreDB::RetrieveJob::fail() {
+  throw NotImplemented("");
+  }
+
+OStoreDB::RetrieveJob::~RetrieveJob() {
+  if (m_jobOwned) {
+    // Re-queue the job entirely if we failed to handle it.
+    try {
+      // We now need to select the tape from which we will migrate next. This should
+      // be the tape with the most jobs already queued.
+      // TODO: this will have to look at tape statuses on the long run as well
+      uint16_t selectedCopyNumber;
+      uint64_t bestTapeQueuedBytes;
+      std::string selectedVid;
+      std::string selectedTapeAddress;
+      objectstore::ScopedExclusiveLock rtfrl(m_rtfr);
+      m_rtfr.fetch();
+      auto jl=m_rtfr.dumpJobs();
+      {
+        // First tape copy is always better than nothing. 
+        auto tc=jl.begin();
+        selectedCopyNumber = tc->copyNb;
+        selectedVid = tc->tape;
+        selectedTapeAddress = tc->tapeAddress;
+        // Get info for the tape.
+        {
+          objectstore::Tape t(tc->tapeAddress, m_objectStore);
+          objectstore::ScopedSharedLock tl(t);
+          t.fetch();
+          bestTapeQueuedBytes = t.getJobsSummary().bytes;
+        }
+        tc++;
+        // Compare with the next ones
+        for (;tc!=jl.end(); tc++) {
+          objectstore::Tape t(tc->tapeAddress, m_objectStore);
+          objectstore::ScopedSharedLock tl(t);
+          t.fetch();
+          if (t.getJobsSummary().bytes > bestTapeQueuedBytes) {
+            bestTapeQueuedBytes = t.getJobsSummary().bytes;
+            selectedCopyNumber = tc->copyNb;
+            selectedVid = tc->tape;
+            selectedTapeAddress = tc->tapeAddress;
+          }
+        }
+      }
+      // We now can enqueue the request on this most promising tape.
+      {
+        objectstore::Tape tp(selectedTapeAddress, m_objectStore);
+        ScopedExclusiveLock tpl(tp);
+        tp.fetch();
+        objectstore::RetrieveToFileRequest::JobDump jd;
+        jd.copyNb = selectedCopyNumber;
+        jd.tape = selectedVid;
+        jd.tapeAddress = selectedTapeAddress;
+        tp.addJob(jd, m_rtfr.getAddressIfSet(), m_rtfr.getSize(), m_rtfr.getPriority(), m_rtfr.getCreationLog().time);
+        tp.commit();
+      }
+      // The request is now fully set. It belongs to the tape.
+      std::string previousOwner = m_rtfr.getOwner();
+      m_rtfr.setOwner(selectedTapeAddress);
+      m_rtfr.commit();
+      // And remove reference from the agent (if it was owned by an agent)
+      try {
+        if (!previousOwner.size())
+          return;
+        objectstore::Agent agent(previousOwner, m_objectStore);
+        objectstore::ScopedExclusiveLock al(agent);
+        agent.fetch();
+        agent.removeFromOwnership(m_rtfr.getAddressIfSet());
+        agent.commit();
+      } catch (...) {}
+    } catch (...) {}
+  }
+}
+
+void OStoreDB::RetrieveJob::succeed() {
+  throw NotImplemented("");
+}
+
 
 }

scheduler/OStoreDB/OStoreDB.hpp

@@ -21,6 +21,7 @@
 #include "scheduler/SchedulerDatabase.hpp"
 #include "objectstore/Agent.hpp"
 #include "objectstore/ArchiveToFileRequest.hpp"
+#include "objectstore/RetrieveToFileRequest.hpp"
 #include "objectstore/SchedulerGlobalLock.hpp"
 
 namespace cta {
@@ -118,6 +119,26 @@ public:
     virtual std::unique_ptr<RetrieveJob> getNextJob();
     virtual void complete(time_t completionTime);
   };
+  
+  /* === Retrieve Job handling ============================================== */
+  class RetrieveJob: public SchedulerDatabase::RetrieveJob {
+    friend class RetrieveMount;
+  public:
+    CTA_GENERATE_EXCEPTION_CLASS(JobNowOwned);
+    CTA_GENERATE_EXCEPTION_CLASS(NoSuchJob);
+    virtual void succeed();
+    virtual void fail();
+    virtual ~RetrieveJob();
+  private:
+    RetrieveJob(const std::string &, objectstore::Backend &, objectstore::Agent &);
+    bool m_jobOwned;
+    uint16_t m_copyNb;
+    objectstore::Backend & m_objectStore;
+    objectstore::Agent & m_agent;
+    objectstore::RetrieveToFileRequest m_rtfr;
+    std::map<std::string, std::string> m_vidToAddress; /**< Cache of tape objects
+                                                        *  addresses filled up at queuing time */
+  };
 
   /* === Admin host handling ================================================ */
   virtual void createAdminHost(const std::string& hostName, 

scheduler/RetrieveJob.cpp

@@ -27,15 +27,15 @@ cta::RetrieveJob::~RetrieveJob() throw() {
 //------------------------------------------------------------------------------
 // constructor
 //------------------------------------------------------------------------------
-cta::RetrieveJob::RetrieveJob(/*RetrieveMount &mount,*/
+cta::RetrieveJob::RetrieveJob(RetrieveMount &mount,
   const ArchiveFile &archiveFile,
-  const RemotePathAndStatus &remotePathAndStatus,
-  const TapeFileLocation &tapeFileLocation,
+  const std::string &remotePath,
+  const NameServerTapeFile &nameServerTapeFile,
   const PositioningMethod positioningMethod):
-  /*mount(mount),*/
+  m_mount(mount),
   archiveFile(archiveFile),
-  remotePathAndStatus(remotePathAndStatus),
-  tapeFileLocation(tapeFileLocation),
+  remotePath(remotePath),
+  nameServerTapeFile(nameServerTapeFile),
   positioningMethod(positioningMethod),
   transferredSize(std::numeric_limits<decltype(transferredSize)>::max()) {}
 

scheduler/RetrieveJob.hpp

@@ -23,9 +23,11 @@
 #include "common/exception/Exception.hpp"
 #include "common/remoteFS/RemotePathAndStatus.hpp"
 #include "scheduler/PositioningMethod.hpp"
+#include "scheduler/SchedulerDatabase.hpp"
 
 #include <string>
 #include <limits>
+#include <memory>
 
 namespace cta {
 
@@ -41,14 +43,8 @@ class RetrieveJob {
    * constructor of RetrieveJob.
    */
   friend class RetrieveMount;
-
-protected:
   
-  /**
-   * Empty constructor. TODO: to be removed in the future when we put in the reference to the owning mount;
-   */
-  RetrieveJob(): transferredSize(std::numeric_limits<decltype(transferredSize)>::max()) {}
-
+public:
   /**
    * Constructor. It is not public as it is generated by the RetrieveMount.
    *
@@ -58,18 +54,26 @@ protected:
    * @param tapeFileLocation the location of the tape file
    * @param positioningMethod the positioning method
    */
-  RetrieveJob(/*RetrieveMount &mount,*/
+  RetrieveJob(RetrieveMount &mount,
   const ArchiveFile &archiveFile,
-  const RemotePathAndStatus &remotePathAndStatus,
-  const TapeFileLocation &tapeFileLocation,
+  const std::string &remotePath,
+  const NameServerTapeFile &tapeFileLocation,
   const PositioningMethod positioningMethod);
+  
+private:
+  std::unique_ptr<cta::SchedulerDatabase::RetrieveJob> m_dbJob;
+  
+  /**
+   * The mount that generated this job
+   */
+  RetrieveMount &m_mount;
 
 public:
 
   /**
    * Destructor.
    */
-  virtual ~RetrieveJob() throw() = 0;
+  virtual ~RetrieveJob() throw();
 
   /**
    * Indicates that the job was successful. The checksum and the size of the 
@@ -102,14 +106,14 @@ public:
   ArchiveFile archiveFile;
   
   /**
-   * The remote file information
+   * The remote file path
    */
-  RemotePathAndStatus remotePathAndStatus; 
+  std::string remotePath; 
                 
   /**
    * The location of the tape file
    */
-  TapeFileLocation tapeFileLocation;
+  NameServerTapeFile nameServerTapeFile;
   
   /**
    * The positioning method

scheduler/RetrieveMount.cpp

@@ -21,14 +21,15 @@
 //------------------------------------------------------------------------------
 // constructor
 //------------------------------------------------------------------------------
-cta::RetrieveMount::RetrieveMount() {
-}
+cta::RetrieveMount::RetrieveMount():
+  m_sessionRunning(false) {}
 
 //------------------------------------------------------------------------------
 // constructor
 //------------------------------------------------------------------------------
 cta::RetrieveMount::RetrieveMount(
-  std::unique_ptr<SchedulerDatabase::RetrieveMount> dbMount) {
+  std::unique_ptr<SchedulerDatabase::RetrieveMount> dbMount): 
+  m_sessionRunning(false) {
   m_dbMount.reset(dbMount.release());
 }
 
@@ -64,7 +65,18 @@ std::string cta::RetrieveMount::getMountTransactionId() const throw(){
 // getNextJob
 //------------------------------------------------------------------------------
 std::unique_ptr<cta::RetrieveJob> cta::RetrieveMount::getNextJob() {
-  throw NotImplemented(std::string(__FUNCTION__) + ": Not implemented");
+  if (!m_sessionRunning)
+    throw SessionNotRunning("In RetrieveMount::getNextJob(): trying to get job from complete/not started session");
+  // Try and get a new job from the DB
+  std::unique_ptr<cta::SchedulerDatabase::RetrieveJob> dbJob(m_dbMount->getNextJob().release());
+  if (!dbJob.get())
+    return std::unique_ptr<cta::RetrieveJob>(NULL);
+  // We have something to retrieve: prepare the response
+  std::unique_ptr<cta::RetrieveJob> ret (new RetrieveJob(*this, 
+    dbJob->archiveFile, dbJob->remoteFile, dbJob->nameServerTapeFile, 
+    PositioningMethod::ByBlock));
+  ret->m_dbJob.reset(dbJob.release());
+  return ret;
 }
 
 //------------------------------------------------------------------------------

scheduler/RetrieveMount.hpp

@@ -36,12 +36,11 @@ namespace cta {
   class RetrieveMount: public TapeMount {
     friend class Scheduler;
   protected:
-
     /**
-     * Constructor.
+     * Trivial constructor
      */
     RetrieveMount();
-
+    
     /**
      * Constructor.
      *
@@ -92,6 +91,7 @@ namespace cta {
      */
     virtual void complete();
 
+    CTA_GENERATE_EXCEPTION_CLASS(SessionNotRunning);
     /**
      * Job factory
      *
@@ -111,6 +111,11 @@ namespace cta {
      * The database representation of this mount.
      */
     std::unique_ptr<cta::SchedulerDatabase::RetrieveMount> m_dbMount;
+    
+    /**
+     * Internal tracking of the session completion
+     */
+    bool m_sessionRunning;
 
   }; // class RetrieveMount
 

scheduler/Scheduler.cpp

@@ -894,14 +894,15 @@ std::unique_ptr<cta::TapeMount> cta::Scheduler::getNextMount(
       for (auto t=tapesList.begin(); t!=tapesList.end(); t++) {
         if (t->vid == m->vid && t->status.availableToRead()) {
           try {