It would be nice to have a mechanism for cleanly shutting down a patchbot, albeit not immediately, such that it exits after finishing the current ticket it's working on. For example, we could use SIGUSR1 for this purpose--when received the patchbot would just set a flag to exit after finishing the current ticket it's working on.
The Docker patchbot, for example, gets rebuilt every now and then, and it would be nice to have a mechanism to launch a new patchbot from the new Docker image, while giving the current patchbot from the previous image a chance to finish what it's working on first, so that it's not wasted effort. This would probably be useful outside the context of Docker images as well.
It would be nice to have a mechanism for cleanly shutting down a patchbot, albeit not immediately, such that it exits after finishing the current ticket it's working on. For example, we could use
SIGUSR1
for this purpose--when received the patchbot would just set a flag to exit after finishing the current ticket it's working on.The Docker patchbot, for example, gets rebuilt every now and then, and it would be nice to have a mechanism to launch a new patchbot from the new Docker image, while giving the current patchbot from the previous image a chance to finish what it's working on first, so that it's not wasted effort. This would probably be useful outside the context of Docker images as well.