Node crashes with exit code 9 when recording big amount of data

[bofetadas]

Hi,

I was working with your bag recorder node recently to be able to remotely trigger bag recordings. It worked pretty well but I discovered that when the topics to record deliver big amount of data your node seems to crash. E.g. I tested recording a topic which yields very small amount of data without any problems. When I changed the topic to one with a type of PointCloud I could record data for roughly 1min until the bag recorder node crashes, saying "exited with error code 9". To give you a scale 1min of recording equals 3.5GB. The problem is that after the node has restarted I no longer can stop the bag recording because it then tells me that no bag recording is going on.
However, when I manually try to record the same heavy topic with rosbag record, it actually works without any problems.

I read that it might an issue with the buffer or chunk size. By looking into your source code I discovered that you set the chunk size to 1024*768. Can you explain why you chose that value? I am referencing to this line: https://github.com/joshs333/bag_recorder/blob/master/src/bag_recorder.cpp#L182

Regards