[chef] Re: Re: MySQL Cookbook Ignores Root Password Setting On First Converge

[Sean OMeara < >]

I tried, but can't seem to reproduce this.

The latest commit on the mysql cookbook sets a nondefault password on the test cookbooks, and adds a wrapper cookbook.

https://github.com/opscode-cookbooks/mysql/commit/a5f470a542f4ad92ccaef831eb92d537877c1b3a

Serverspec is passing with flying colors:

https://gist.github.com/someara/5410caf221bf1af25b1a

Maybe it has something to do with chef-vault?

PS: It's not going to have anything to do with node.set vs node.default. The password is being passed into the mysql_service resource as a parameter.

-s

On Sun, Jun 22, 2014 at 4:56 PM, DV < " target="_blank"> > wrote:

I suggest you replace the following in recipes/server.rb:

root_account = ChefVault::Item.load("mysql_accounts", "root")
node.set['mysql']['server_root_password'] = root_account['password']

With the following in attributes/default.rb:

root_account = ChefVault::Item.load("mysql_accounts", "root")
default['mysql']['server_root_password'] = root_account['password']

It's very likely that your node.set call happens after mysql cookbook tries to determine root password, which is why you're not seeing the behavior you want.

On Sat, Jun 21, 2014 at 2:51 AM, Stuart Ellis < " target="_blank"> > wrote:

Hello all,

I’m currently building a Chef setup and have hit a consistent issue with the MySQL cookbook. Given the popularity of this cookbook, I’m happy to believe that I’m just doing something wrong.

The problem is it’s ignoring the passwords in my wrapper cookbook the first time that the cookbook is applied, and instead, uses the default values. The second time that the cookbook runs it does try to apply the correct passwords, but fails, because the MySQL installation requires the root password that was set on the first run.

The test nodes are fresh Ubuntu 14.04 LTS VMs - I’m building and destroying the nodes each time to ensure that they are clean. I’ve tried using the block format for configuring the MySQL resource, and with hard-coded values instead of Chef Vault, but get the same behaviour in both cases. There’s no open issue in JIRA or GitHub to suggest that this is a known issue.

My wrapper cookbook is here:

https://bitbucket.org/stuartellis/se-mysql

Thanks in advance,

Stuart

---
Stuart Ellis
" target="_blank">

--
Best regards, Dmitriy V.