With a very big configuration during the detection with a Gate you may experience an out-of-memory.
I guess that the Gate sends a start_detection but doesn't consume it, because it's using polling. If detection needs more messages than the MAX_MSG_NB, the oldest message (start_detection) collides with the new one and Luos_engine assert.
Context and environment
I see this with a configuration with approximately 70 services over 6 boards.
By debugging it I see that the oldest message is a broadcast one, could be a start_detection or end_detection...
How to reproduce the bug
Get a big configuration and reduce the MAX_MSG_NBR
Details
Which version the bug has been detected on
*Luos engine 3.0.0
Description of the bug
With a very big configuration during the detection with a Gate you may experience an out-of-memory. I guess that the Gate sends a
start_detectionbut doesn't consume it, because it's using polling. If detection needs more messages than the MAX_MSG_NB, the oldest message (start_detection) collides with the new one and Luos_engine assert.Context and environment
I see this with a configuration with approximately 70 services over 6 boards. By debugging it I see that the oldest message is a broadcast one, could be a start_detection or end_detection...
How to reproduce the bug
Get a big configuration and reduce the MAX_MSG_NBR
Possible solution
Find a way to consume the oldest message.