Bug #43

Shared memory leak in python bindings

Added by Redmine Admin over 10 years ago. Updated about 8 years ago.

Status:FixedStart date:
Priority:HighDue date:
Assignee:Redmine Admin% Done:

0%

Category:python
Target version:0.64.4

Description

The python bindings leak shared memory. This is a problem when you have resident scripts handling drqueue stuff periodically.

Repro steps:

import drqueue.base.libdrqueue as drqueue

comp = drqueue.request_computer_list(drqueue.CLIENT)

sys.exit(0)

Now if you do an ipcs, you will see a new shared memory item for each computer on the network. When you have a network of, say, 100 computers, this will exhaust the shared memory very quickly and the script will begin to fail :(

History

#2 Updated by Andreas Schröder almost 10 years ago

  • Status changed from New to In progress

#4 Updated by Andreas Schröder over 9 years ago

  • Status changed from In progress to New

#5 Updated by Andreas Schröder about 9 years ago

I've added a script to look for shared memory leaks: https://ssl.drqueue.org/project/browser/branches/0.64.x/python/python_leak_test.py

So far, no leaks for me with these tests.

#6 Updated by Andreas Schröder about 9 years ago

Running 'compinfo.Darwin.i386 -l' still leaks shared memory. It doesn't seem to depend on the number of connected slaves. The number of used SHM segments increases by 1 every time you run compinfo.

#7 Updated by Andreas Schröder over 8 years ago

Memory leak problems of other DrQueue components will be addressed in own tickets.

Please test the Python issue with the script posted above and extend it if it's not complete enough.

#8 Updated by Andreas Schröder about 8 years ago

  • Status changed from New to Fixed
  • Resolution set to fixed

Closed for now. Python bindings don't seem to leak memory anymore.

Also available in: Atom PDF