Jool is acting weird after flushing pool6791 and blacklist

[crisdeleon]

After playing around with --pool6791 I found some strange behaviour after using the --flush option:

Adding an address, flushing, then adding the same address again

ip netns exec blue jool_siit --pool6791 198.1.2.3
ip netns exec blue jool_siit --pool6791 --flush
ip netns exec blue jool_siit --pool6791 198.1.2.3
Killed

After the Killed message:

• The jool_siit kernel module was unaffected, and it was still translating packets.
• Trying to use the userspace modules (jool and jool_siit) in the global netns worked properly, stating that they could not communicate with the kernel module (expected, since the module was running inside the blue netns)
• Trying to use the userspace modules inside the blue netns proved unsuccessful, leaving the console hanging. This happened both to jool_siit and jool userspace modules (hinting this might be a netlink problem).

  Trying to query --pool6791 after --flush

ip netns exec blue jool_siit --pool6791 198.1.2.3
ip netns exec blue jool_siit --pool6791 --flush
ip netns exec blue jool_siit --pool6791

When I tried this, the whole virtual machine hosting jool went unresponsive. I was forced to reset.

Testing with --blacklist was no better

Adding and address to blacklist, flushing then adding the same address again

ip netns exec blue jool_siit --blacklist 198.1.2.3
ip netns exec blue jool_siit --blacklist --flush
ip netns exec blue jool_siit --blacklist 198.1.2.3
Error: the requested entry intersects with pool entry 120.116.0.0/0
(Error code: 17)

I won't even try to draw any conclusion.

Trying to query --blacklist after --flush

• Exactly the same as the pool6791 scenario.

I'll keep testing for everything that uses the --flush option (i. e. eamt, pool4, pool6), but right now my VMware server is kind of wonky after all the resets.

[crisdeleon]

Seems to be working fine now. Also tested the same scenarios with --eamt and --pool6 and looks fine.

[ydahhrk]

Thanks!