[chef] Re: aws_ebs_raid fails on ec2 node, but only when it's rebooted

[Daniel DeLeo < >]

On Wednesday, June 26, 2013 at 4:06 AM, Jorge Bianquetti wrote:

Hi!

Something strange it's happening with one mongoDB node that needs and

EBS raid with provisioned IOPS, using Chef 10.26.0

First, node creates EBS raid type 10 with two disks without problems,

and then use it to hosts mongoDB databases.

Then, I reboot or stops the ec2 node.

When the node finished the boot process, EBS raid devices are

attached to the machine, but when I run chef-client, it fails with

message:

FATAL: ArgumentError: ruby_block[wait_mongodb_data_raid]

(/var/chef/cache/cookbooks/aws/providers/ebs_raid.rb line 182) had an

error: ArgumentError: You tried to set a nested key, where the parent

is not a hash-like object:

aws/raid//var/lib/mongodb/device_map/sdi1/sdi2/sdi1

Node's info seems right to me:

"aws": {

"raid": {

"/var/lib/mongodb": {

"raid_dev": "md0",

"device_map": {

"sdi1": "vol-6bba1647",

"sdi2": "vol-bbbb1797"

}

}

},

"ebs_volume": {

"sdi1": {

"volume_id": "vol-6bba1647"

},

"sdi2": {

"volume_id": "vol-bbbb1797"

}

}

}

So, I tried with different aws cookbook versions: 0.101.3 0.101.2

0.100. All fails in the same steps.

Any suggestions?

What does the ruby_block look like? Also, if you can try it with Chef 11, the attribute system is reworked so you should not see such inscrutable errors.

-- 

Daniel DeLeo