Handle HW errors on read and write in CM by stopping controllers #742
Conversation
github-actions
bot
requested review from
bmagyar,
DasRoteSkelett,
harderthan,
jackcenter,
mhubii and
UsamaHamayun1
|
This pull request is in conflict. Could you fix it @destogl? |
|
If a hardware stopped working you would stop the controller. |
I think (and hope) you missed something 😄 There is a service And yes, unfortunately, this is manual process for now. I hope to get some external components soon that would be a state-machines for ros2_control. Adding this into framework would it make even more cluttered and complex. |
|
I'll split this into two PRs |
| @@ -1496,6 +1541,20 @@ void ControllerManager::list_hardware_components_srv_cb( | |||
| hwi.name = interface; | |||
| hwi.is_available = resource_manager_->command_interface_is_available(interface); | |||
| hwi.is_claimed = resource_manager_->command_interface_is_claimed(interface); | |||
| // TODO(destogl): Add here mapping to controller that has claimed or | |||
destogl
force-pushed
the
handle-hardware-errors-in-cm
branch
from
c6a0e1e to
0970d01
Compare
|
This pull request is in conflict. Could you fix it @destogl? |
…ead and write. Fix misleading variable name in the tests. Error handling in RM of read/write is tested. Remove all interface from available list for hardware when an error happens. Working verison of stopping controller on read error. Write error also works but the tests are missing. Finalized tests for deactivating controllers when write failes. Do some more variable renaming to the new nomenclature.
destogl
force-pushed
the
handle-hardware-errors-in-cm
branch
from
0970d01 to
64ec9e2
Compare
|
@Mergifyio backport humble |
Add code for deactivating controller when hardware gets an error on read and write. Fix misleading variable name in the tests. Remove all interface from available list for hardware when an error happens. Do some more variable renaming to the new nomenclature. (cherry picked from commit ecf6c69)
✅ Backports have been created
|
Add code for deactivating controller when hardware gets an error on read and write. Fix misleading variable name in the tests. Remove all interface from available list for hardware when an error happens. Do some more variable renaming to the new nomenclature. (cherry picked from commit ecf6c69)
Add code for deactivating controller when hardware gets an error on read and write. Fix misleading variable name in the tests. Remove all interface from available list for hardware when an error happens. Do some more variable renaming to the new nomenclature. (cherry picked from commit ecf6c69)
Add code for deactivating controller when hardware gets an error on read and write. Fix misleading variable name in the tests. Remove all interface from available list for hardware when an error happens. Do some more variable renaming to the new nomenclature. (cherry picked from commit ecf6c69)
(cherry picked from commit 5178503) Co-authored-by: Christoph Fröhlich <[email protected]>
In short: when a hardware fails, i.e., error happens on read and write and hardware goes to "UNCONFIGURED" or "FINALIZED" state, controllers should be stopped automatically.
this addresses few recent issues.
Add code for deactivating controller when hardware gets an error on read and write.
Remove all interface from available list for hardware when an error happens.
To add in follow up PR: automatically restart Broadcasters (and trigger publishing before stopping)