Odd Quirk (probably me being dumb)

[Jaq_Flash]

Firstly, Love Ansible made my life so much easier…

Up until this morning

We’ve been running a playbook for a while (under Ansible 1.5.5 on CentOS 6.5) that copies a zip file to a remote machine (Also CentOS 6.5) and then unzips it.

For reasons that aren’t entirely clear, we’ve recently updated the Control box to Ansible 1.7.1

[root@XXX ~]# ansible --version
ansible 1.7
[root@XXX ~]# yum list installed | grep ansible
ansible.noarch 1.7-1.el6 @epel

However, now when we run the job, it’s starting that task and just hanging.

I’ve checked on the target machine and it is creating a .ansible/tmp folder with 2 files in it

source
unarchive

The size of the source file matches the size of the zip file.

Checking the destination, the files seem to be unarchived fine.

At this point, the 2 files mentioned above vanish.

There are no errors reported in the ansible.log that I have setup.

However, the playbook never proceeds past this point. I was wondering (As these are production servers controlled by a Machiavellian change control system, They lag slightly on the package updates) is there some other dependancy change between Ansible 1.5 and Ansible 1.7?

I can attach all sorts of log files if you think that will help

Thanks in advance of any form of vaguely helpful response.

[James_Cammarata]

Hi Jaq,

Those two files are the temporary files pushed to the system (the module code and the archived file). If the archived file is very large, it could be an issue of the output from the module not being correctly read in, so the connection is hanging waiting for I/O. We have fixed at least one bug in this area in the devel branch (1.8), so you might want to give that a test on a non-production system to see if it resolves your issue.

Beyond that, any other information you can provide like an strace of the execution would be helpful in determining what the process may be hung upon (you can attach the strace on the server side, if possible too).

Thanks!