Open jesusmah opened 5 years ago
What do we do with this? Seems to be a problem with RHEL 7.4 (devicemapper)...
I also have problem with RHEL 7.6. I found out that the install-docker.sh is one issue.
I can do a fork of this and solve the issue for you and then marge it.
Hi @Issen007, Im glad to hear you know how to fix it. Can you please check out our contribution guideline https://github.com/ibm-cloud-architecture/terraform-icp-vmware/blob/master/CONTRIBUTING.md for doing so? More precisely look at the code contribution section and the github flow in particular. Anyway, thanks in advance. Looking forward to seeing your contribution.
After some investigation the main problem is in the code that comes from
https://github.com/ibm-cloud-architecture/terraform-module-icp-deploy
So I'm working on this to make it working with RHEL and CentOS
As soon I have solve that part, we can update the icp-deploy.tf
file so you download the correct package.
The terraform requires docker to be installed in direct-lvm mode. If the template has not already installed docker then the terraform uses the install-docker.sh script to install docker, add the docker disk, and bring docker up in direct-lvm mode so it can use the disk as a native block device. Using RHEL this fails for the nodes that are not master nodes. The sequence on the nodes is that docker installs (in loopback mode), the script manipulates the disk at
/dev/sdb
, updates thedaemon.json
,stops docker
, thenstarts docker
. In the actual node the docker start actually fails due todevicemapper
errors. However, docker restarts on a time delay after thesystemd
failure and comes up clean in direct-lvm mode with the docker disk configured. But technically the script has failed in that the last command, docker start, failed. This is not a problem on the master node since the next step creates thecfc disk
so that script exits cleanly. But on the other nodes the last step is the docker install script and the terraform sees the script as a failure and kills the deployment. I added asleep 2
andsudo systemctl restart docker
after thesudo systemctl start docker
command to force the final command to complete cleanly and the terraform got past this problem