failed to transfer file; Read from socket failed

Not sure if my error is related to the control path (see link below) or not, but I got a error about not being able to transfer a file in my logs.

http://docs.ansible.com/intro_configuration.html#control-path

2014-12-04 10:02:47,137 p=1602 u=jmcmahon | TASK: [census.deploy | Copy .ruby-gemset to releases] *************************
2014-12-04 10:07:21,740 p=1602 u=jmcmahon | fatal: [mqa-cbeserver1.flatworldknowledge.com] => failed to transfer file to /tmp/ansible-tmp-1417705367.85-253946577954926/copy:

Read from socket failed: Connection reset by peer^M
Connection closed^M

2014-12-04 10:07:21,741 p=1602 u=jmcmahon | FATAL: all hosts have already failed – aborting
2014-12-04 10:07:21,741 p=1602 u=jmcmahon | PLAY RECAP ********************************************************************
2014-12-04 10:07:21,743 p=1602 u=jmcmahon | to retry, use: --limit @/Users/jmcmahon/deploy-census.retry

2014-12-04 10:07:21,744 p=1602 u=jmcmahon | mqa-cbeserver1.flatworldknowledge.com : ok=11 changed=7 unreachable=1 failed=0

Any ideas on what happened? I was able to manually run the script again and all worked fine.

Hi Jim,

Does this happen frequently or just the one time? If the task re-ran without issues, it’s probably not a control path issue. It could be a transient issue, for example if you just happened to catch the ControlPersist process when it was shutting down due to a timeout, or it could be a ulimit issue (too many open files, if a large number of forks are being used).

It’s not a frequent thing. It happened today and perhaps a month ago. That’s why I thought it might not be the control path issue (I am deploying to EC2 from a Mac laptop, but changed the host names on the EC2 box to be shorter and more readable).

I’ve been running the cron job without verbose logging. I’ve change the job to include -v logging in case it happens again. It’s not a big issue, but just want to be sure it’s not a configuration issue I should be aware of.

Thanks,

Jim McMahon