GitLab

Added a --config_file option in runWorker.pl, which allows beekeeper.pl to propagate its own parameters

Inverted the if statements so that people are not told about rebalancing semaphores until it is really needed

No more warnings when compiling the manual !

When the worker table gets busy, the UPDATEs can fail because of timeout (see
below), so we "protect" them with our beloved protected_prepare_execute.

Beekeeper : loop #20 ======================================================
GarbageCollector:       Checking for lost Workers...
GarbageCollector:       [Queen:] out of 2 Workers that haven't checked in during the last 5 seconds...
DBD::mysql::st execute failed: Lock wait timeout exceeded; try restarting transaction at /homes/muffato/workspace/src/hive/master/modules/Bio/EnsEMBL/Hive/DBSQL/StatementHandle.pm line 119.
DBD::mysql::st execute failed: Lock wait timeout exceeded; try restarting transaction at /homes/muffato/workspace/src/hive/master/modules/Bio/EnsEMBL/Hive/DBSQL/StatementHandle.pm line 119.
 at /homes/muffato/workspace/src/hive/master/modules/Bio/EnsEMBL/Hive/DBSQL/StatementHandle.pm line 145.
        Bio::EnsEMBL::Hive::DBSQL::StatementHandle::__ANON__(Bio::EnsEMBL::Hive::DBSQL::StatementHandle=HASH(0x3c294f0), 14) called at /homes/muffato/workspace/src/hive/master/modules/Bio/EnsEMBL/Hive/Queen.pm line 485
        Bio::EnsEMBL::Hive::Queen::check_for_dead_workers(Bio::EnsEMBL::Hive::Queen=HASH(0x3916fc0), Bio::EnsEMBL::Hive::Valley=HASH(0x35e7c08), 0) called at /homes/muffato/workspace/src/hive/master/scripts/beekeeper.pl line 501
        main::run_autonomously(HASH(0x222ce18), Bio::EnsEMBL::Hive::HivePipeline=HASH(0x34e1aa0), -1, "FOREVER", Bio::EnsEMBL::Hive::Valley=HASH(0x35e7c08), ARRAY(0x3a6c0c8), undef, undef, ...) called at /homes/muffato/workspace/src/hive/master/scripts/beekeeper.pl line 378
        main::main() called at /homes/muffato/workspace/src/hive/master/scripts/beekeeper.pl line 30

* version/2.4:
  Revert "Update the done_job_count counter after each job."

* version/2.3:
  Revert "Update the done_job_count counter after each job."

Workers were in fact designed not to update the analysis_stats table too
frequently. The "done_job_count" counter is updated every two minutes.

This reverts commit 258cbf59.

* version/2.4:
  Update the done_job_count counter after each job.