Resolution:
Issues with multi-core engines or shared filesystem engines

In cases where multiple engines are using the same resource synchronization destination, engines can get into an unstable state while synchronizing large (> 80 MB) grid libraries. This behaviour can be overcome by increasing the Synchronization Time (seconds). This setting can be found on all brokers. For instance, in GridServer 5 it is located in Resource Deployment -> Broker Settings.

To change this setting:
- log in to the broker;
- go to Manager Configuration-&gtResource Deployment
- look for the field named Synchronization Timeout (seconds) and change it to a larger number.

To estimate the best timeout value, you can use the Estimated File Download Times Calculator from the Knowledge Base (DSKBID #303).

Issues between director and brokers

When deploying a large grid library from director to broker, it can take some time to deliver the complete file, especially if the two systems are separated by a WAN. In the meanwhile, the director might instruct the engines to pick up the new resource from the broker, as soon as the engine finishes its current task. If the large grid library has not fully downloaded to the broker, the engine is turned away due to incomplete synchronization, and may then be unavailable to take any further tasks.

To overcome this condition, you can:

1) Setup your managers to monitor resources less often than the default value of 30 seconds.

To change the monitoring value go to Manager | Manager Configuration | Resource Deployment -> Monitor Interval (in Minutes). You can estimate a value by using the Estimated File Download Times Calculator from the Knowledge Base (DSKBID #303).

2) Deploy your resource manually on all the Brokers and Director, using a two phase approach:

Deploy the new Resouces into a temporary directory on all the Managers;

Move the file from the temporary directory to the correct location of your resource directory
(\webapps\livecluster\deploy\resources)