[chef] Re: chef-client (still) randomly failing

[Daniel DeLeo < >]

On Tuesday, November 19, 2013 at 10:13 AM, Phil Cryer wrote:

I'm still failing to get chef-client to run on it's own without failing due to timeouts. I have 4 CentOS 6.4 nodes in QA, overnight usually 2 will fail and show as 'red' the next day - it runs hourly, so they'll be down for 10+ hours. I login to the node, stop chef-client, run it manually, and it works. 

1) Why is it timing out on NTP, and even failing as it retries 4 times in a row? 

2) Why doesn't it fix itself on the next run? It seems to fail, and then never recover.

Full logs here:

https://gist.github.com/philcryer/7549743

Can you check lsof and ps to see if there is a child process of chef that does not get killed and keeps the yum lock?

You may be hitting this issue: https://tickets.opscode.com/browse/MIXLIB-16

That’s been fixed in master and should get released with the next Chef 11.x release. You could try building the gem manually and installing it on one of your systems to see if it fixes the problem which would be greatly appreciated.

-- 

Daniel DeLeo