kill all pids when killing pash via signal

[dkarnikis]

This should fix #452

Signed-off-by: Dimitris Karnikis [email protected]

[dkarnikis]

The only issue here is that I do not know any way to silence the produced output by the killed processes:

temp-analytics.sh:^CTraceback (most recent call last):                                              
  File "/home/dkarnikis/pash/compiler/pash.py", line 258, in <module>                               
    main()                                                                                          
  File "/home/dkarnikis/pash/compiler/pash.py", line 35, in main                                    
    return_code = preprocess_and_execute_asts(ast_objects, args, input_script_arguments, shell_name)
  File "/home/dkarnikis/pash/compiler/pash.py", line 73, in preprocess_and_execute_asts             
    return_code = execute_script(fname, args.command, input_script_arguments, shell_name)           
  File "/home/dkarnikis/pash/compiler/pash.py", line 254, in execute_script                         
    exec_obj = subprocess.run(subprocess_args, env=new_env, close_fds=False)                        
  File "/usr/lib/python3.8/subprocess.py", line 495, in run                                         
    stdout, stderr = process.communicate(input, timeout=timeout)                                    
  File "/usr/lib/python3.8/subprocess.py", line 1020, in communicate                                
    self.wait()
  File "/usr/lib/python3.8/subprocess.py", line 1083, in wait
    return self._wait(timeout=timeout)
  File "/usr/lib/python3.8/subprocess.py", line 1806, in _wait
    (pid, sts) = self._try_wait(0) 
  File "/usr/lib/python3.8/subprocess.py", line 1764, in _try_wait
    (pid, sts) = os.waitpid(self.pid, wait_flags)
KeyboardInterrupt
Killed
1.477
Terminated
25.665

gzip: stdin: unexpected end of file
cat: /tmp/pash_d3DWqNu/4a72fcee018249e68d6af4f1b1d0fc45/#fifo248: No such file or directory
cat: /tmp/pash_d3DWqNu/4a72fcee018249e68d6af4f1b1d0fc45/#fifo249: No such file or directory
cat: /tmp/pash_d3DWqNu/4a72fcee018249e68d6af4f1b1d0fc45/#fifo250: No such file or directory
cat: /tmp/pash_d3DWqNu/4a72fcee018249e68d6af4f1b1d0fc45/#fifo251: No such file or directory
cat: /tmp/pash_d3DWqNu/4a72fcee018249e68d6af4f1b1d0fc45/#fifo252: No such file or directory
cat: /tmp/pash_d3DWqNu/4a72fcee018249e68d6af4f1b1d0fc45/#fifo253: No such file or directory
cat: /tmp/pash_d3DWqNu/4a72fcee018249e68d6af4f1b1d0fc45/#fifo254: No such file or directory
cat: /tmp/pash_d3DWqNu/4a72fcee018249e68d6af4f1b1d0fc45/#fifo255: No such file or directory
cat: /tmp/pash_d3DWqNu/4a72fcee018249e68d6af4f1b1d0fc45/#fifo256: No such file or directory
cat: /tmp/pash_d3DWqNu/4a72fcee018249e68d6af4f1b1d0fc45/#fifo257: No such file or directory
cat: /tmp/pash_d3DWqNu/4a72fcee018249e68d6af4f1b1d0fc45/#fifo258: No such file or directory
cat: /tmp/pash_d3DWqNu/4a72fcee018249e68d6af4f1b1d0fc45/#fifo259: No such file or directory
cat: /tmp/pash_d3DWqNu/4a72fcee018249e68d6af4f1b1d0fc45/#fifo260: No such file or directory
cat: /tmp/pash_d3DWqNu/4a72fcee018249e68d6af4f1b1d0fc45/#fifo261: No such file or directory

All the processes are killed however :)

[angelhof]

There could be processes that run commands like grep and sort right? I don't think this kills those. Also, could this kill the processes of another pash instance?

I think that in order to kill all processes we need to do something inside this: https://github.com/binpash/pash/blob/main/runtime/wait_for_output_and_sigpipe_rest.sh in addition to whatever we catch here.

Also, whenever killing processes, we also need to wait for them.

Also, this should not go to main but to future since it is an experimental change.

[dkarnikis]

It actually kills everything!
In the current state of the script yeah, it will kill other pash instances. However, don't we assume that only a single instance of pash is running on machine?

[angelhof]

It actually kills everything!

I don't see why tmp/pash would match a grep that is left running. Maybe I am missing something though

In the current state of the script yeah, it will kill other pash instances. However, don't we assume that only a single instance of pash is running on machine?

I don't think this is a valid assumption. We did a lot of modifications to allow multiple instances of pash to be run by the same user.

[dkarnikis]

If I am not mistaken, all the pash commands operate on /tmp/pash_XXXXX/blah, that's why the pkill -f -9 tmp/pash works for us.

Is there even a point to optimize the script to only kill the children of the current pash instance?

[angelhof]

If I am not mistaken, all the pash commands operate on /tmp/pash_XXXXX/blah, that's why the pkill -f -9 tmp/pash works for us.

A grep in the dataflow might read and write to two completely arbitrary files, therefore I am not sure if /tmp/pash would appear anywhere in its invocation.

Is there even a point to optimize the script to only kill the children of the current pash instance?

This is not an optimizations but rather correctness! If two instances of pash are running we don't want to kill both because one is killed.

[dkarnikis]

I quite confident that the issue with the failing tests on the CI are already fixed on the fixes PR

[dkarnikis]

Closing this PR. Moved to #458