tree 8cf42f35417cd465a69d47c4370290eaf9c8f1cd
parent 722055abdcedbda88152453733631f0d93d79d4a
author Flavio Fernandes <flaviof@redhat.com> 1593629833 -0400
committer Flavio Fernandes <flaviof@redhat.com> 1593705564 -0400

test_port_forwarding_to_2_servers: netcat listen stops too soon

This change is a continuation of https://review.opendev.org/#/c/713208/

Even with the changes from commit fd4141f2015d25f1b009d7cf2ebdd2907cd8e81a
our tests have intermittent failures where nc listen process terminates as
soon as the shell that spawns it closes. By (1) breaking the cmd script to
have an extra line, (2) adding a 0.1 sec delay and (3) setting stdin of
process the proposed changes ensures that this termination no longer
takes place.

In order to reproduce this issue, follow these set of steps as a reference:

From a separate shell session in the cirros vm, watch out for nc processes

$ watch -n1 -d "ps -elf | grep nc"

Then, start a python session to interact with that vm via ssh session:

$ python3

from neutron_tempest_plugin.common import ssh
cirros_vm_ip='10.0.0.10'
c=ssh.Client(cirros_vm_ip, 'cirros', 'gocubsgo', port=22)
c.execute_script('nc udp -p 12346 &')

If you are lucky (or unlucky), you will see that there are no nc
processes running, unless you use a variation to that command that
looks like this:

c.execute_script('nc udp -p 12349 &\nsleep 0')

Than then in the watch session you will see something that look like:

_<PID>_  cirros   nc udp -p 12349

And that will remain running even after the python session closes.

Closes-Bug: #1868100
Change-Id: I8192391d6f2e86f9e486805374f710444e770ea4