Closed michaelcoburn closed 9 years ago
Well, while my fix advances the provisioning process, it does break puppet. Specifically, variables defined in Vagrantfile such as $datadir_dev
get ignored by puppet as if they don't exist:
==> master: Notice: /Stage[main]/Mysql::Datadir/Exec[datadir_dev_noop]/returns: sh: /sys/block//queue/scheduler: No such file or directory
==> master: Error: echo 'noop' > /sys/block//queue/scheduler returned 1 instead of one of [0]
==> master: Error: /Stage[main]/Mysql::Datadir/Exec[datadir_dev_noop]/returns: change from notrun to 0 failed: echo 'noop' > /sys/block//queue/scheduler returned 1 instead of one of [0]
Someone with more vagrant-percona / ruby-fu than myself is needed to debug the correct setting to use instead of id: manifest_file
.
Next to a new vagrant, the Centos version was updated to 7, which includes new ways of dealing with settings and new puppet versions as well.
For me everything is kind of broken atm. as well.
I have not yet had any time to fix this.
Fixed in https://github.com/jayjanssen/vagrant-percona/commit/c81d22a140225e66935db410eed3b06e6a7bdfd0
I had this patch already, just hadn't committed/pushed it. This will likely break vagrant < 1.7
vagrant up
on 1.7.1 is failing with* The following settings shouldn't exist: id
:I am able to clear the error by making the following edit to
lib/vagrant-common.rb
however this is probably a valuable directive and it seems to break puppet a little bit later in the execution, but at least it doesn't stop vagrant from running.Now it runs properly (truncated):
Here are my versions: