PDA

View Full Version : Assemblies & Start Order Attribute


tmart
07-23-2008, 12:14 PM
This is tricky wording here... so I'm trying to be precise:

I know that I could test this with some simple apps to find out, but I thought that it might be useful both to get the question answered here... and to get the docs updated to be inclusive of this matter. I apologize if it's already in the docs, but I did not find it there. I also want to make sure that this is deterministic and that I'm not just seeing some biased artifacts of what is actually arbitrary behavior.

Now, without further delay, in two parts, the question:

Do the start orders of components within an assembly collate directly with other components outside of that assembly (eg. components directly within "main")?

Or instead, do the start orders of components within an assembly collate only with other components of the same assembly. And furthermore, the assembly has its own start order that collates with other components/assemblies at the same hierarchical level. Thus, start orders of components within an assembly serve as orders for a secondary collation. ?

To illustrate with a simple case... assume the following Components: Start Order

main.LINUX:40
main.assembly1: 50
main.assembly1.WEB: 30

The question then is this: is the component named "main.assembly1.WEB" started deterministically before or after the component named "main.LINUX" ??

PeterNic
07-23-2008, 12:58 PM
Tim,

Good question. See http://doc.3tera.com/AppLogic23/RefEditorInstanceSettings.html, "Start order". The start order within the assembly is opaque; the assembly starts as a whole, with the appliances inside it starting in their own order.

So, in your example, the appliance start sequence will be:

main.LINUX
main.assembly1.WEB



Regards,
-- Peter

tmart
07-23-2008, 01:02 PM
Thanks... I'm still on the 2.1 production release, so I guess I should manually start looking at the 2.3 docs to answer these questions.

PeterNic
07-23-2008, 01:10 PM
No problem. This behavior is the same in 1.2 and 2.1/2.2 -- but please do try 2.3 as well :)