Lobster_data can handle load peaks through queuing. Profiles can be given different priorities (Lowest (-2), Low (-1), Normal (0), Higher (+1), Highest (+2)) for each of which there is a separate queue. In addition, there is a queue for profiles that can only run in one instance (Single Run) and for Response Routes that run as a separate thread.

If the underlying operating system permits it, threads are processed according to their priority with more or less processor power. So threads of the group Very Low (-2) get less CPU time than threads of the group Low (-1), etc.

The entries in this list contain profiles with input data already received by Lobster_data. These are jobs with phase 1 completed, but phase 2 has not yet begun. For more details see section Phase 1.

images/download/attachments/21305742/Thread_Queues_2_EN-version-1-modificationdate-1529922398000-api-v2.png

(1) Selection of the thread queue by priority.

(2) The name of the profile to the thread.

(3) The name of the input file and the job number to the thread.

(4) The selected entry in the list is moved to the last or first position in the list.

(5) The selected entry is completely removed from the list (or all jobs of the selected profile (2) of the entry). This will terminate the started job before the data is being processed.

Behaviour after Profile Change

If the priority of a profile is changed (and saved), and if jobs of the profiles are in a queue, these jobs will be moved from the current queue to the queue with the new priority. This also applies to a load balance system.

So if a queue is about to overflow, for example, because a partner sends a huge amount of data, you can change the priority of the profile that is handling the partner's data and therefore all the related job move to another queue that does not contain any other jobs.