Project

General

Profile

Bug #7602

Re-evaluation of NodeList blocks re-harvesting

Added by Robert Waltz about 8 years ago. Updated about 8 years ago.

Status:
Closed
Priority:
High
Assignee:
Robert Waltz
Category:
d1_log_aggregation
Start date:
2016-01-22
Due date:
% Done:

100%

Story Points:
Sprint:

Description

LogAggregation hangs during re-evaluation of the nodelist for harvest scheduling adjustments.

LogAggregation checks for new nodes or deleted nodes once a day. The scheduled job that performs the service hangs while attempting to determine if any jobs are executing. In order for changes to be made to the scheduler, the scheduler will pause or place in standby all triggers for the job. After checking if all jobs are complete, then the scheduler will add/remove jobs and start up again. The problem is that during the pause, all jobs are never paused because the code for managing the harvest is itself a job.

The resolution will be to check that every job is halted except for the calling job.

Associated revisions

Revision 17407
Added by Robert Waltz about 8 years ago

refs #7602

Modified LogAggregationScheduleManager to ignore the LogAggregationManageScheduleJob when re-evaluating MemberNode harvest scheduling. Switched to Log4j in most of the code for ease of debugging.Changed the name of LogAggregationHarvestJob to LogAggregationManageScheduleJob as a more accurate description of the purpose of the class.

Revision 17407
Added by Robert Waltz about 8 years ago

refs #7602

Modified LogAggregationScheduleManager to ignore the LogAggregationManageScheduleJob when re-evaluating MemberNode harvest scheduling. Switched to Log4j in most of the code for ease of debugging.Changed the name of LogAggregationHarvestJob to LogAggregationManageScheduleJob as a more accurate description of the purpose of the class.

Revision 17408
Added by Robert Waltz about 8 years ago

refs #7602

Modified LogAggregationScheduleManager to ignore the LogAggregationManageScheduleJob when re-evaluating MemberNode harvest scheduling. Switched to Log4j in most of the code for ease of debugging.Changed the name of LogAggregationHarvestJob to LogAggregationManageScheduleJob as a more accurate description of the purpose of the class.

Revision 17408
Added by Robert Waltz about 8 years ago

refs #7602

Modified LogAggregationScheduleManager to ignore the LogAggregationManageScheduleJob when re-evaluating MemberNode harvest scheduling. Switched to Log4j in most of the code for ease of debugging.Changed the name of LogAggregationHarvestJob to LogAggregationManageScheduleJob as a more accurate description of the purpose of the class.

Revision 17409
Added by Robert Waltz about 8 years ago

refs #7602

Modified LogAggregationScheduleManager to increase the timeframe when the ManageScheduleJob will run initially.

Revision 17409
Added by Robert Waltz about 8 years ago

refs #7602

Modified LogAggregationScheduleManager to increase the timeframe when the ManageScheduleJob will run initially.

Revision 17420
Added by Robert Waltz about 8 years ago

refs #7602

Modified LogAggregationScheduleManager removed the pause for due to hazelcast (hazelcast is no longer a factor in job scheduling) before triggering jobs.

Modified LogAggregationScheduleManager to shortcircuit rescheduling if there are no jobs to be scheduled or deleted.

Modified LogAggregationScheduleManager to only consider nodes whose state is UP when performing collection math to determine jobs to schedule or delete.

Revision 17420
Added by Robert Waltz about 8 years ago

refs #7602

Modified LogAggregationScheduleManager removed the pause for due to hazelcast (hazelcast is no longer a factor in job scheduling) before triggering jobs.

Modified LogAggregationScheduleManager to shortcircuit rescheduling if there are no jobs to be scheduled or deleted.

Modified LogAggregationScheduleManager to only consider nodes whose state is UP when performing collection math to determine jobs to schedule or delete.

Revision 17421
Added by Robert Waltz about 8 years ago

refs #7602

decrease the delay offset limit to 5 minutes

Revision 17421
Added by Robert Waltz about 8 years ago

refs #7602

decrease the delay offset limit to 5 minutes

Revision 17429
Added by Robert Waltz about 8 years ago

refs #7602

Modified LogHarvesterTask to fix an annoying log message.

Revision 17429
Added by Robert Waltz about 8 years ago

refs #7602

Modified LogHarvesterTask to fix an annoying log message.

Revision 17431
Added by Robert Waltz about 8 years ago

fixes #7602: Re-evaluation of NodeList blocks re-harvesting

Revision 17431
Added by Robert Waltz about 8 years ago

fixes #7602: Re-evaluation of NodeList blocks re-harvesting

Revision 17432
Added by Robert Waltz about 8 years ago

refs #7602: Re-evaluation of NodeList blocks re-harvesting

Revision 17432
Added by Robert Waltz about 8 years ago

refs #7602: Re-evaluation of NodeList blocks re-harvesting

Revision 17433
Added by Robert Waltz about 8 years ago

refs #7602: Re-evaluation of NodeList blocks re-harvesting of log aggregation

Revision 17433
Added by Robert Waltz about 8 years ago

refs #7602: Re-evaluation of NodeList blocks re-harvesting of log aggregation

History

#1 Updated by Robert Waltz about 8 years ago

  • Status changed from In Progress to Testing
  • % Done changed from 30 to 50

#2 Updated by Robert Waltz about 8 years ago

  • Status changed from Testing to Closed
  • % Done changed from 50 to 100

Applied in changeset d1-python:d1_python|r17431.

Also available in: Atom PDF

Add picture from clipboard (Maximum size: 14.8 MB)