Aborting worker is not working

[saganshul]

Aborting worker on worker panel is not working.
Link to screenvideo:
https://www.dropbox.com/s/90atwivjpkhzfkq/out.mp4?dl=0

Expected Behavior

It should stop the current job.

Your Environment

• Installation method used: python install/install.py
• Version/branch used: Develop
• Operating System and version (like Kali, Debian, ArchLinux, etc): Ubuntu 16.04
• pip/setuptools version: 8.1.2
• sudo access: Yes

[DarKnight24]

@saganshul I am not able to reproduce the same.

[saganshul]

@DarKnight24 , try to test on plugin which takes time to run and try to abort the same.

[viyatb]

@saganshul @DarKnight24 the request to kill the worker is sent successfully but sometimes the worker does not die immediately. Need to check the function in https://github.com/owtf/owtf/blob/develop/framework/plugin/worker_manager.py#L242.

[jack17529]

I am working on this issue can someone please tell me what's the meaning of -
return True in [worker['busy'] for worker in self.workers]
https://github.com/owtf/owtf/blob/develop/framework/plugin/worker_manager.py#L141

[viyatb]

@jack17529 self.workers is a list of workers. The line for worker in self.workers iterates over every worker in that list and gets the status of the worker using worker['busy']. The function returns True if any worker is 'busy'.

The idiom used is a list comprehension - it is a syntactic way to write:

status = []
for worker in self.workers:
    status.append(worker['busy'])
return True in status

[jack17529]

Thank you @delta24

[jack17529]

Can someone explain meaning of "NA" in self.workers[k]["start_time"] = "NA" in context to code.
https://github.com/owtf/owtf/blob/develop/framework/plugin/worker_manager.py#L119

and also- del self.workers[pseudo_index - 1] in context to the code.
https://github.com/owtf/owtf/blob/develop/framework/plugin/worker_manager.py#L252

[DePierre]

@jack17529 I would assume that 'NA' is a start_time error value used when the worker is idling. Since start_time cannot have any meaningful value if the worker is not working, 'NA' is used instead.

About your second question, you can get more meaning by running git grep delete_worker. You'll see that the pseudo_index refers to the index provided via the REST API. When the user clicks on the red cross to delete a worker, the URL /api/workers/<id>/abort is fetched. This API will then call delete_worker(<id>) (therefore corresponding to pseudo_index, -1 to have the index starting from 0 and not 1). Then, as explained in the docstring, it Remove it from self.workers so that is is not restarted by manager cron.

Do you need more information? We'll be happy to give you more details ;)

[jack17529]

@delta24 I have already looked all the code ,can't find any problem with code. But I think can be a problem related to the api used.What to do now ?
furthermore can you provide your email please.

[DePierre]

@jack17529 I am not convinced that the API is at fault here and I would find it more likely that the issue is in our multiprocess system.

It is possible that the worker_manager.py is not the culprit here. Because a worker can run a shell command (to run w3af for instance), it is possible that when trying to send a SIGINT the worker does not receive it because w3af is freezing.

[viyatb]

Resolved by 2b2bd22.