Revert "cta/CTA#186 CTA should cope with idle database connections being dropped by the database server"

This reverts commit 1be86a70.

Refactored RdbmsCatalogue::getTapesForWriting() to be like OracleCatalogue::filesWrittenToTape()

OracleCatalogue::filesWrittenToTape() should now transparently re-connect to the database.

RdbmsCatalogue::getTapesForWriting() should now transparently re-connect to the database.

Admin commands should now transparently re-connect to the database.

Fixed the following offending method:

    RdbmsCatalogue::prepareForNewFile()

Added the following unit-test that explicitly tests for a UserError exception:

    cta_catalogue_CatalogueTest.prepareForNewFile_no_archive_routes

Done

Reduced to the minium the number of database connections used by createSameSchemaInTwoSeparateInMemoryDatabases

Partially implemented.

The Catalogue::getTapePools() method called by the CTA front-end
now returns an std::list of TapePool structures, where each
structure now has the following additional "summary"
member-variables:

  /**
   * The total number of tapes in the pool.
   */
  uint64_t nbTapes;

  /**
   * The total capacity of all the tapes in the pool in gigabytes
   * (10^9 bytes).
   */
  uint64_t capacityGigabytes;

  /**
   * The total amount of compressed data written to all the tapes
   * in the pool in gigabytes * (10^9 bytes).
   */
  uint64_t dataGigabytes;

These three additional member-variables can now be used by the
CTA front-end to provide the summary information requested by
issue #136 in the appropriate format / presentation.  For example,
the CTA front-end could subtract dataGigabytes from
capacityGigabytes in order to calculate the approximate amount of
free space.  Please note that this calculation could potentially
result in a negative number due to the fact that capacityGigabytes
is an approximate number entered by a tape operator for each tape
when they register those tapes into CTA.  In such a case the CTA
front-end should display 0 and not the negative result.

Moved TapePool from common::dataStructures:: to catalogue:: because it belongs to the Catalogue interface

Replaced set_target_properties( .. .. INSTALL_RPATH ..) with set_property (TARGET .. APPEND PROPERTY INSTALL_RPATH ..)

If the whole contents of the ARCHIVE_FILE and TAPE_FILE tables are
listed without a search criteria and with an ORDER BY clause then the
first row back from the database will take a relatively long period of
time to be returned.  This is because the database will have to sort
every row before returning the first.  This long period of time will
cause the current (non SSI) cta command-line tool to time out.

Without this index we got errors like:

[root@p06253947b39467 ~]# cta af ls --instance cta --path /eos/dev/test/gcancio/10072017/file_2167
Failed to execute the command. Reason: Failed to open root://10.16.135.81:10955//Y3Rh&YWY=&bHM=&LS1pbnN0YW5jZQ==&Y3Rh&LS1wYXRo&L2Vvcy9kZXYvdGVzdC9nY2FuY2lvLzEwMDcyMDE3L2ZpbGVfMjE2Nw==: [ERROR] Operation expired
[root@p06253947b39467 ~]#

With this index we execute the same query in less than a second:

[root@p06253947b39467 ~]# time cta af ls --instance cta --path /eos/dev/test/gcancio/10072017/file_2167
 289440       1  I20170    1381    51096      cta   18965         1006       ADLER32     0XFBA23F1D    german_raw  gcancio       c3    1499697578 /eos/dev/test/gcancio/10072017/file_2167

real    0m0.590s
user    0m0.005s
sys    0m0.006s
[root@p06253947b39467 ~]#

The reponse rate of the CTA catalogue to "eos rm"
commands was too slow.  The response rate was
around 20Hz when using the CTA/EOS production
database configured with 10 database connections.
The rate has now been increased to 30Hz by adding
the TAPE_FILE_ARCHIVE_FILE_ID_FK index.

This reverts commit 3b1855a8.

The 'SELECT FOR UPDATE' was not the bottle neck.  I am putting it
back for now.