Subprocess popen error

[jihoonl]

^CTraceback (most recent call last):
  File "/home/jihoonl/research/ros/rocon/devel/bin/rocon_launch", line 5, in <module>
    exec(fh.read())
  File "<string>", line 18, in <module>
  File "/home/jihoonl/research/ros/rocon/src/rocon_tools/rocon_launch/src/rocon_launch/launch.py", line 152, in main
    signal.pause()
  File "/home/jihoonl/research/ros/rocon/src/rocon_tools/rocon_launch/src/rocon_launch/launch.py", line 114, in signal_handler
    self.terminal.shutdown_roslaunch_windows(self.processes, self.hold)
  File "/home/jihoonl/research/ros/rocon/src/rocon_tools/rocon_launch/src/rocon_launch/terminals.py", line 73, in shutdown_roslaunch_windows
    roslaunch_pids.extend(utils.get_roslaunch_pids(process.pid))
  File "/home/jihoonl/research/ros/rocon/src/rocon_tools/rocon_launch/src/rocon_launch/utils.py", line 98, in get_roslaunch_pids
    ps_command = subprocess.Popen("ps -o pid -o comm --ppid %d --noheaders" % parent_pid, shell=True, stdout=subprocess.PIPE, stderr=subprocess.STDOUT)
TypeError: %d format: a number is required, not NoneType

[stonier]

@jihoonl are you able to consistently reproduce this error?

[jihoonl]

No. I have seen this once so far.

[stonier]

Did it happen when ctrl-c'ing the windows as they were starting up? There is a small window there where the pid value can be None.

[jihoonl]

I will update the issue when I see this again.

[stonier]

I bypassed it in the commit above and added a warning log so we can see if it happens incorrectly in future. I believe it can only happen when killing the windows while they are still trying to spawn, which is not a bug. If the user decides to shutdown so early, then he can clean it up. Adding infrastructure to wait around to check and cleanup is too much overhead right now.

[stonier]

Reopen this if you see that log warning outside the startup-immediate shutdown sequence.

[stonier]

Saw again, slightly different message this time...no NoneType.

^Caborting call to find child roslaunches of a non-existant parent pid (can happen if cancelling spawned processes while they are still establishing).
Terminating roslaunch [pid: 20305]
Terminating roslaunch [pid: 20292]
Terminating roslaunch [pid: 20280]
Traceback (most recent call last):
  File "/mnt/zaphod/ros/rocon/devel/bin/rocon_launch", line 5, in <module>
    exec(fh.read())
  File "<string>", line 18, in <module>
  File "/mnt/zaphod/ros/rocon/src/rocon_tools/rocon_launch/src/rocon_launch/launch.py", line 153, in main
    signal.pause()
  File "/mnt/zaphod/ros/rocon/src/rocon_tools/rocon_launch/src/rocon_launch/launch.py", line 115, in signal_handler
    self.terminal.shutdown_roslaunch_windows(self.processes, self.hold)
  File "/mnt/zaphod/ros/rocon/src/rocon_tools/rocon_launch/src/rocon_launch/terminals.py", line 93, in shutdown_roslaunch_windows
    os.killpg(process.pid, signal.SIGTERM)
TypeError: an integer is required

[jihoonl]

yujin@isaac:~/demos/rocon_demo_201405/src/cafe_concert/cafe_concert/services/cafe_database$ rocon_launch cafe_concert cafe_sim.concert --screen
Launching /home/yujin/demos/rocon_demo_201405/src/cafe_concert/cafe_concert/solution/concert.launch on port 11311
Launching /home/yujin/demos/rocon_demo_201405/src/cafe_concert/cafe_concert/launch/cafe_backend.launch on port 11312
Launching /home/yujin/demos/rocon_demo_201405/src/cafe_concert/cafe_concert/launch/cafe_backend.launch on port 11313
Launching /home/yujin/demos/rocon_demo_201405/src/cafe_concert/cafe_concert/launch/cafe_backend.launch on port 11314
^Caborting call to find child roslaunches of a non-existant parent pid (can happen if cancelling spawned processes while they are still establishing).
Terminating roslaunch [pid: 2655]
Terminating roslaunch [pid: 2648]
Terminating roslaunch [pid: 2657]
Invalid pid value [None][an integer is required]

[stonier]

Really stumped on this one. The above was reproduced with a two minute gap between starting and stopping. Could easily check this and ignore the problem, but doesn't help understand why the problem is there.

[jihoonl]

Is this related?

(gnome-terminal:32221): GLib-GIO-CRITICAL **: g_settings_get: the format string may not contain '&' (key 'monospace-font-name' from schema 'org.gnome.desktop.interface'). This call will probably stop working with a future version of glib.

Kill signal failed to reach the terminal - typically this means the terminal has already shut down.
Traceback (most recent call last):
  File "/home/jihoonl/research/ros/gazeborocon/devel/bin/rocon_launch", line 5, in <module>
    exec(fh.read())
  File "<string>", line 18, in <module>
  File "/home/jihoonl/research/ros/gazeborocon/src/rocon_tools/rocon_launch/src/rocon_launch/launch.py", line 153, in main
    signal.pause()
  File "/home/jihoonl/research/ros/gazeborocon/src/rocon_tools/rocon_launch/src/rocon_launch/launch.py", line 115, in signal_handler
    self.terminal.shutdown_roslaunch_windows(self.processes, self.hold)
  File "/home/jihoonl/research/ros/gazeborocon/src/rocon_tools/rocon_launch/src/rocon_launch/terminals.py", line 82, in shutdown_roslaunch_windows
    rocon_python_utils.system.wait_pid(pid)
  File "/home/jihoonl/research/ros/gazeborocon/src/rocon_tools/rocon_python_utils/src/rocon_python_utils/system/pid.py", line 99, in wait_pid
    delay = check_timeout(delay)
  File "/home/jihoonl/research/ros/gazeborocon/src/rocon_tools/rocon_python_utils/src/rocon_python_utils/system/pid.py", line 73, in check_timeout
    time.sleep(delay)
  File "/home/jihoonl/research/ros/gazeborocon/src/rocon_tools/rocon_launch/src/rocon_launch/launch.py", line 115, in signal_handler
    self.terminal.shutdown_roslaunch_windows(self.processes, self.hold)
  File "/home/jihoonl/research/ros/gazeborocon/src/rocon_tools/rocon_launch/src/rocon_launch/terminals.py", line 95, in shutdown_roslaunch_windows
    console.warning("Kill signal failed to reach the terminal - typically this means the terminal has already shut down.")
  File "/home/jihoonl/research/ros/gazeborocon/src/rocon_tools/rocon_console/src/rocon_console/console.py", line 113, in warning
    print(yellow + msg + reset)
  File "/home/jihoonl/research/ros/gazeborocon/src/rocon_tools/rocon_launch/src/rocon_launch/launch.py", line 115, in signal_handler
    self.terminal.shutdown_roslaunch_windows(self.processes, self.hold)
  File "/home/jihoonl/research/ros/gazeborocon/src/rocon_tools/rocon_launch/src/rocon_launch/terminals.py", line 95, in shutdown_roslaunch_windows
    console.warning("Kill signal failed to reach the terminal - typically this means the terminal has already shut down.")
  File "/home/jihoonl/research/ros/gazeborocon/src/rocon_tools/rocon_console/src/rocon_console/console.py", line 113, in warning
    print(yellow + msg + reset)
  File "/home/jihoonl/research/ros/gazeborocon/src/rocon_tools/rocon_launch/src/rocon_launch/launch.py", line 115, in signal_handler
    self.terminal.shutdown_roslaunch_windows(self.processes, self.hold)
  File "/home/jihoonl/research/ros/gazeborocon/src/rocon_tools/rocon_launch/src/rocon_launch/terminals.py", line 95, in shutdown_roslaunch_windows
    console.warning("Kill signal failed to reach the terminal - typically this means the terminal has already shut down.")
  File "/home/jihoonl/research/ros/gazeborocon/src/rocon_tools/rocon_console/src/rocon_console/console.py", line 113, in warning
    print(yellow + msg + reset)
  File "/home/jihoonl/research/ros/gazeborocon/src/rocon_tools/rocon_launch/src/rocon_launch/launch.py", line 115, in signal_handler
    self.terminal.shutdown_roslaunch_windows(self.processes, self.hold)
  File "/home/jihoonl/research/ros/gazeborocon/src/rocon_tools/rocon_launch/src/rocon_launch/terminals.py", line 95, in shutdown_roslaunch_windows
    console.warning("Kill signal failed to reach the terminal - typically this means the terminal has already shut down.")
  File "/home/jihoonl/research/ros/gazeborocon/src/rocon_tools/rocon_console/src/rocon_console/console.py", line 113, in warning
    print(yellow + msg + reset)
  File "/home/jihoonl/research/ros/gazeborocon/src/rocon_tools/rocon_launch/src/rocon_launch/launch.py", line 121, in signal_handler
    os.unlink(f.name)
OSError: [Errno 2] No such file or directory: '/tmp/tmpqRyttt'

[stonier]

I doubt it, but can't qualify that 100%. Probably something 'cleaned it' from the tmp directory before we got to it.

[stonier]

Long time since seen...closing. Reopen if new reports come in.