Publisher cleanup via service call results in deadlock on Jazzy
Nobody has claimed this yet.
- Dominant language
- C++
- Stars
- 805
- Forks
- 564
- Avg merge
- 1d 17h
- Merged PRs (30d)
- 27
Description
Relevant issue here:
https://github.com/luxonis/depthai-ros/issues/657
Short description:
When using a service for resetting the driver's state (mostly cleaning up classes that have publishers inside them), the driver has a high chance to hang without response in rclcpp::Executor::wait_for_work (rcl_wait/rmw_wait). This was thought to be fixed by assigning Reentrant CB groups to services that correspond that call this cleanup, but it seems that it didn't help in the end.
The issue has been observed on both Cyclone and FastDDS RMW.
Humble and Noetic versions of the driver don't experience this issue.
Contributor guide
First steps
- Read the whole issue, then the project's contributing guide.
- Comment on the issue to say you are picking it up — it saves two people doing the same work.
- Fork the repository and make your change on a branch.
- Open a pull request that references the issue number.
Research direction
Start by reproducing the Jazzy-only hang described at rclcpp::Executor::wait_for_work, including the rcl_wait/rmw_wait path, and compare behavior with Humble and Noetic. Investigate whether Reentrant callback groups affect cleanup services, using both Cyclone and FastDDS; done means the service returns reliably without the executor deadlocking.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- cpp
- Domain
- distributed-systems
- Issue type
- Bug
- Difficulty
- 4/5
- Estimated time
- 3-5 days
- Activity status
- Stale
- Clarity
- Needs clarification
- Newbie friendliness
- 30/100