Creation of EBS volume fails with "volumeID is required"


I’m trying to create a new EBS volume from docker-compose but it fails with the following error:

> Activating (Expected state running but got error: /VolumeDriver.Attach 'ebstest_test-volume_6894b' (driver 'rancher-ebs') returned status 500: 500 Internal Server Error)

I’m using this docker-compose as a test:

version: '2'

        image: nginx:latest
            io.rancher.container.pull_image: 'always'
            - test-volume:/var/lib/storage
        driver: rancher-ebs
            name: test-volume
            size: 10
            volumeType: gp2

I checked the EBS driver log for the machine in question and it yields the following error:

3/30/2017 8:02:33 PMtime="2017-03-30T18:02:33Z" level=error msg="Failed to attach, opts=={\"name\":\"ebstest_test-volume_6894b\",\"rancher\":\"true\",\"size\":\"10\",\"volumeType\":\"gp2\"}: volumeID is required"
3/30/2017 8:02:33 PMtime="2017-03-30T18:02:33Z" level=error msg=attach.response error="volumeID is required"

The volume is show under Infrastructure -> Storage but is never created in AWS.

If I create the same volume from Infrastructure -> Storage the volume is created in AWS and I’m able to use it from my compose file.

I’m running Rancher 1.5.3 and the machines in this environment are all RancherOS 0.9.0.

Any idea on what could be wrong?


ehm, this beats me. It suddenly started working. I created a few new volumes via infrastructure -> storage, and suddenly I’m able to create them via docker-compose.yml as well…

I first thought this was related to one of my machines, so I deleted the machine in question and added another. I now get the same error on the new machine. Some of the machines seems to be working fine, but other not. The new machines is provisioned via Rancher and is running RancherOS 0.9.1.
The error message is the same as before and the volume is not created when trying to do so via docker-compose.yml.

Can I provide more logs to help debug this issue?