@Marwan99 please review. Its not quite working yet (one last quirk) & need to add tests, but look over the basics of what's going on

tested:

• Mess with timeout values parameterization working
• Clean exit
• 0 valued timeout disables
• is properly brought down if a server crashes while operating or breaks bone
• All up, then transition down and back up again 2 times without issues
• All up and stable for 1 hour
• comes up 100% of the time without crashing or timing out

Needs testing / working

• test coverage (manager: fake node work up and down; param set to 0, lifecycle wrapper: covered by integration tests)

Codecov Report

Merging #1869 into master will decrease coverage by 0.02%.
The diff coverage is 97.33%.

@@            Coverage Diff             @@
##           master    #1869      +/-   ##
==========================================
- Coverage   70.14%   70.11%   -0.03%     
==========================================
  Files         218      218              
  Lines       10586    10649      +63     
==========================================
+ Hits         7426     7467      +41     
- Misses       3160     3182      +22     

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 69977cd...19e1dcc. Read the comment docs.

Looks good to me!

I'm super irritated - so I have things generally working (if a server dies or breaks the bond, the lifecycle manager hears about it and triggers things down).

Now the issue is that I can't keep things up stably. I have printouts saying that both directions are sending regular heartbeats but then after the timeout period, every single time one of the servers triggers a failed heartbeat connection to the lifecycle manager checking if they're broken and the lifecycle manager brings down all the nodes.

I think bondcpp is broken or something really weird happening. I see the active regular heartbeats coming in at 4x the rate of the heartbeat timeout so everything should be working fine. I think it has to do with the ROS1 to ROS2 port trying to be fancy and ended up messing alot of things up.

@naiveHobo mind taking a look? See the repos file for my fork of bondcpp to enable lifecycle. Take particular attention at the heartbeat things & the ROS1 versions. I see the Heartbeat() method being triggered by the state machine every time we have a new message from the bond status callback (it calls SisterAlive() or something which in turns calls Heartbeat()). In Heartbeat() for ROS1 it resets the timer without triggering the heartbeat timeout callback but in this one for ROS2, it pulls a flag and resets the timer which calls it and triggers the timeout procedure to bring things down I think.

I think the timeout.cpp from ROS1 is not equivalent to the weird timer reset / cancel functions they added.

Yup sure, on it

Awesome! I also filed a ticket in Bond itself since I got deep enough into it to believe that bond itself is broken. Today I'm going to scour the interwebs for another ROS2 application using it, but I'm not hopeful, to see if I'm just messing something up and its working for anyone else.

new PR to replace later today