fkie / multimaster_fkie

ROS stack with FKIE packages for multi-robot (discovering, synchronizing and management GUI)

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

master_discovery memory leak

amilcarlucas opened this issue · comments

I'm using the latest master version compiled on NVidia Jetson TX2, TX2i and NX.
All three boards run on Ubuntu 18.04 LTS.

The longer the master_discovery service runs the more memory it needs.
I suspect it has a memory leak!

I run some tests with five ROS-Master's (five master_discovery, five master_sync and one node_manager) on Ubuntu 18.04.

I record the memory usage with psrecord.

With no changes on ROS-Master the memory usage increases slowly (almost an hour) until an upper limit (60MB)
md_no_changes_in_ros_master

With a ROS-Node stop/start every 5 seconds the limit of 60 MB is reached faster and stay at this level.
md_new_nodes_every_app_5s

I don't know exactly what the reason for the increase is at the beginning and whether it can be fixed at all.

I now upgraded to Ubuntu 20.04 LTS and ROS noetic
It was using python2, now is using python3
After 10 days it increases from 40MiB to 140MiB

Any progress on this?
Please re-open it.
After 40 days it increases from 40MiB to 423MiB