How to use Post Mortem

Post Mortem describes the possibility to save data in case of system failure. System refers to the accelerator in this case. If a certain event is distributed relevant data for further analysis should be stored in the Post Mortem database.

At GSI no such system is in place yet (09/2015). FESA supports Post Mortem since version 3.0.0. Post Mortem data may be stored on the file system instead of a specialized Post Mortem database.

How to add Post Mortem to a design

To use Post Mortem export in a property add an 'export' node to the property. In this export node add a 'post-mortem' node.

The post-mortem node needs the following attributes:

• data-class: Data class ID (string). May not contain quotes, backslashes, or characters forbidden by the PM system.
• system: System ID (string). May not contain quotes, backslashes, or characters forbidden by the PM system.
• data-source-pattern: Data source pattern (string). Use $d as a placeholder for the device name. A double dollar sign is replaced by a single one ($$ becomes $). May not contain quotes, backslashes, or characters forbidden by the PM system.

Configuring Post Mortem to File

Using command-line parameters to the DU:

-pmfiles -pmdir path/for/file

Using configuration properties:

pm.dest = file 
pm.dir = path/for/file

The filename will be of the form: pm_data_SystemID

Configuring Post Mortem to Server

Specifics will be determined when once the post mortem system design is available.

Using configuration properties:

pm.dest = servername

How to Trigger Post Mortem

Post Mortem is triggered by notification of a property with an export/port-mortem node.

This requires a real-time action with a notified-property node referencing the property to export.

Info

https://wikis.cern.ch/display/FESA3/Post-mortem+integration+in+FESA

https://wikis.cern.ch/display/PMS/Guidelines+for+PM+modules

-- SolveighMatthies - 03 Sep 2015