On a fresh instance-store EC2 instance with ami-74199744 in us-west-2, I configured one EBS volume to be mounted. The recipe blew up at line 4's Dir.glob('/dev/xvd?').sort.last[-1,1].succ because last returned nil (because no existing device matched the glob). I patched it with this:

diff --git a/cookbooks/ebs/recipes/volumes.rb b/cookbooks/ebs/recipes/volumes.rb
index e6a7503..edbfbb2 100644
--- a/cookbooks/ebs/recipes/volumes.rb
+++ b/cookbooks/ebs/recipes/volumes.rb
@@ -1,7 +1,8 @@
 node[:ebs][:volumes].each do |mount_point, options|
   if !options[:device] && options[:size]
     credentials = Chef::EncryptedDataBagItem.load('credentials', 'aws')
-    devid = Dir.glob('/dev/xvd?').sort.last[-1,1].succ
+    existing_xvdevices = Dir.glob('/dev/xvd?')
+    devid = existing_xvdevices.empty? ? 'f' : existing_xvdevices.sort.last[-1,1].succ
     device = "/dev/sd#{devid}"

     vol = aws_ebs_volume device do

If you like I can put that into a pull request. Let me know. If you think there's a better way, please let me know.

Hi, I'm attempting to add a new raid ebs volume, and /dev/xvdf never becomes ready. I also don't see new ebs volumes being created in the in the aws console. I'm using the latest master branch of chef-ebs and latest chefdk:

default[:ebs][:raids]['/dev/md0'][:num_disks] = 4
default[:ebs][:raids]['/dev/md0'][:disk_size] = 10

default[:ebs][:raids]['/dev/md0'][:piops] = 30

default[:ebs][:raids]['/dev/md0'][:raid_level] = 0
default[:ebs][:raids]['/dev/md0'][:fstype] = 'xfs'
default[:ebs][:raids]['/dev/md0'][:mount_point] = '/mnt/mysql'
default[:ebs][:raids]['/dev/md0'][:uselvm] = true

ec2-54-82-103-248.compute-1.amazonaws.com * ruby_block[wait for devices] action run[2016-04-04T15:53:23+00:00] INFO: Processing ruby_block[wait for devices] action run (ebs::raids line 58)
DEBUG: received packet nr 1044 type 94 len 156
INFO: channel_data: 0 131b
ec2-54-82-103-248.compute-1.amazonaws.com [2016-04-04T15:53:23+00:00] DEBUG: Providers for generic ruby_block resource enabled on node include: [Chef::Provider::RubyBlock]
DEBUG: received packet nr 1045 type 94 len 156
INFO: channel_data: 0 130b
ec2-54-82-103-248.compute-1.amazonaws.com [2016-04-04T15:53:23+00:00] DEBUG: Provider for action run on resource ruby_block[wait for devices] is Chef::Provider::RubyBlock
DEBUG: received packet nr 1046 type 94 len 108
INFO: channel_data: 0 83b
ec2-54-82-103-248.compute-1.amazonaws.com [2016-04-04T15:53:23+00:00] INFO: Waiting for individual disks of RAID /mnt/mysql
DEBUG: received packet nr 1047 type 94 len 92
INFO: channel_data: 0 72b
ec2-54-82-103-248.compute-1.amazonaws.com [2016-04-04T15:53:23+00:00] INFO: device /dev/xvdf not ready - waiting
DEBUG: read 116 bytes
DEBUG: received packet nr 1048 type 94 len 92
INFO: channel_data: 0 72b
ec2-54-82-103-248.compute-1.amazonaws.com [2016-04-04T15:53:33+00:00] INFO: device /dev/xvdf not ready - waiting
DEBUG: read 116 bytes
DEBUG: received packet nr 1049 type 94 len 92

I've been using this cookbook for over a year now and have yet to run into any issues. This is my first attempt at a RAID array created with this cookbook.

It seems that it's just hanging waiting for /dev/xvdc to be mounted but it's never creating the ebs volumes in AWS.

I've tested that I can create single ebs volumes without issue on the same system with the same cookbook. I've let this run for about 2 hours and it's never gotten any further.

Below is the output from an strace on the chef worker

root@box1:/home/ubuntu# strace -p 10177
Process 10177 attached - interrupt to quit
restart_syscall(<... resuming interrupted call ...>) = -1 ETIMEDOUT (Connection timed out)
futex(0x1a57618, FUTEX_WAKE_PRIVATE, 1) = 0
clock_gettime(CLOCK_MONOTONIC, {976, 573213306}) = 0
stat("/dev/xvdc", 0x7fff2333b4b0) = -1 ENOENT (No such file or directory)
clock_gettime(CLOCK_MONOTONIC, {976, 573604612}) = 0
clock_gettime(CLOCK_MONOTONIC, {976, 573730133}) = 0
futex(0x1a575b4, FUTEX_WAIT_BITSET_PRIVATE, 25, {986, 573730133}, ffffffff) = -1 ETIMEDOUT (Connection timed out)
futex(0x1a57618, FUTEX_WAKE_PRIVATE, 1) = 0
clock_gettime(CLOCK_MONOTONIC, {986, 574298508}) = 0
stat("/dev/xvdc", 0x7fff2333b4b0) = -1 ENOENT (No such file or directory)
clock_gettime(CLOCK_MONOTONIC, {986, 574664859}) = 0
clock_gettime(CLOCK_MONOTONIC, {986, 574789903}) = 0
futex(0x1a575b4, FUTEX_WAIT_BITSET_PRIVATE, 27, {996, 574789903}, ffffffff) = -1 ETIMEDOUT (Connection timed out)
futex(0x1a57618, FUTEX_WAKE_PRIVATE, 1) = 0
clock_gettime(CLOCK_MONOTONIC, {996, 575247390}) = 0
stat("/dev/xvdc", 0x7fff2333b4b0) = -1 ENOENT (No such file or directory)

Any info/advice would be greatly appreciated.

Thanks.

The latest (pre-fork) code for this cookbook is at https://github.com/aws/opsworks-cookbooks/ which would be great to have a reference to in the documentation.

In raids.rb line 53, bails out on chef-client runs on 11.04, ubuntu 12.04 lts. It's to do with the new strict property enforcement, but I can't figure how quite why your merge is throwing it, sorry.

Need to pull from cstewart87@0893799 to fix volume mounting errors.

This cookbook supports piops and standard volumes, but EC2 now supports general purpose SSD volumes with a type of gp2.

Thoughts: add volume_type to node[:ebs][:volumes] options. If not set, use the current logic. If set, accept the volume_type.