The second part of the vSphere Integrated Containers series will be focusing on the prerequisites and the deployment. I will list the requirements for the Docker client and VCH in vSphere as well as go through the process of deploying the VCH. We will also run a couple tests at the end to make sure that VIC is working correctly. If you haven’t had a chance to go through vSphere Integrated Containers Part 1 – Getting started, I suggest taking a look to understand the components of VIC; otherwise, let’s get started!
- Windows 7 or 10
- Mac OSX 10.11 (El Capitan) Note: I’m running on Sierra without any issues.
- Ubuntu 16.04 LTS
- vSphere 6.0 or higher
One of the following configurations
- DRS enabled cluster
- Standalone hosts managed by vCenter
- Standalone hosts unmanaged by vCenter
- Enterprise plus licensing
- Allow outbound TCP port 2377
- Allow inbound HTTPS/TCP on port 443
Download and Install Docker Toolbox for Mac or Windows
Click here to download Docker Toolbox and install it on your windows machine or mac to use as a client.
It’s possible to deploy a VCH to a standalone ESXi host, but I don’t see many use cases for that so I will deploy it to a vCenter Server cluster.
First, download the VIC engine bundle, the link is on Part 1 – Getting started with VIC at the bottom of the page. Download both the engine and the registry. When you download the engine, the following binaries are zipped (Credit)
|appliance.iso||The Photon based boot image for the virtual container host (VCH) endpoint VM|
|bootstrap.iso||The Photon based boot image for the container VMs.|
|ui/||A folder that contains the files and scripts for the deployment of the vSphere Web Client Plug-in for vSphere Integrated Containers Engine.|
|vic-machine-darwin||The OSX command line utility for the installation and management of VCHs.|
|vic-machine-linux||The Linux command line utility for the installation and management of VCHs.|
|vic-machine-windows.exe||The Windows command line utility for the installation and management of VCHs.|
|vic-ui-darwin||The OSX executable for the deployment of the vSphere Web Client Plug-in for vSphere Integrated Containers Engine.
NOTE: Do not run this executable directly.(1)
|vic-ui-linux||The Linux executable for the deployment of the vSphere Web Client Plug-in for vSphere Integrated Containers Engine.
NOTE: Do not run this executable directly.(1)
|vic-ui-windows.exe||The Windows executable for the deployment of the vSphere Web Client Plug-in for vSphere Integrated Containers Engine.
NOTE: Do not run this executable directly.(1)
|README||Contains a link to the vSphere Integrated Containers Engine repository on GitHub.|
|LICENSE||The license file for vSphere Integrated Containers Engine|
Step 1. Create a VDS port group with the name “vic-bridge” Optionally, create port groups for the client, management, and public networks.
Step 2. Once you have unzipped the engine bundle, navigate to the directory with a terminal. Run the following command. For windows or linux, change to vic-machine-(windows|linx) create. For a full list of create commands click here. I have specified static IPs, hence I have a lot more input in my command. This command wasn’t the most fun to input and it took a lot of playing around, using DHCP would cut this to about one line as you would just need to specify the bridge network port group.
./vic-machine-darwin create -t vcsa1.corp.local -u "email@example.com" -p VMware1! -n harbor -r mgmt-edge-compute -i drobo1 -vs "drobo1/ContainerVolumeStore:default" -b vic-bridge --bnr 10.10.0.0/12 -cln vic-all --dns-server 172.16.10.2 -pn vic-all -mn vic-all -cn vic-all --public-network-ip 172.16.10.48 --public-network-gateway 172.16.10.1/24 --registry-ca ./HarborCert/ca.crt --no-tlsverify -f
Note: I ended up hitting several errors below. I decided to put the resolutions of each of those here in case anyone else hits the same issues.
ERRO[2017-03-31T11:57:33-06:00] Unable to load certificates: cname option doesn't match existing server certificate in certificate path virtual-container-host ERRO[2017-03-31T11:57:33-06:00] -------------------- ERRO[2017-03-31T11:57:33-06:00] vic-machine-darwin failed: cname option doesn't match existing server certificate in certificate path virtual-container-host
Resolution: After a couple of failed attempts, I needed new certificates because of naming changes. I navigated to the folder “virtual-container-host” and moved both certificates to folder called “Old Certs” so that it would generate new ones.
ERRO[2017-03-31T12:37:48-06:00] -------------------- ERRO[2017-03-31T12:37:48-06:00] Static IP on network sharing port group with public network - Configuration ONLY allowed through public network options ERRO[2017-03-31T12:37:48-06:00] Failed to configure static IP for additional networks using port group "vic-public-and-mgmt" ERRO[2017-03-31T12:37:48-06:00] client network gateway specified without at least one routing destination ERRO[2017-03-31T12:37:48-06:00] Firewall must permit dst 2377/tcp outbound to the VCH management interface ERRO[2017-03-31T12:37:48-06:00] Create cannot continue: configuration validation failed ERRO[2017-03-31T12:37:48-06:00] -------------------- ERRO[2017-03-31T12:37:48-06:00] vic-machine-darwin failed: validation of configuration failed
Resolution: Edit Firewall on ESXi Host
ERRO[2017-03-31T13:22:44-06:00] client network gateway specified without at least one routing destination ERRO[2017-03-31T13:22:44-06:00] Create cannot continue: configuration validation failed ERRO[2017-03-31T13:22:44-06:00] -------------------- ERRO[2017-03-31T13:22:44-06:00] vic-machine-darwin failed: validation of configuration failed
Resolution: Specify a routing destination for client gateway
--client-network-gateway routing_destination_1/subnet_mask, routing_destination_2/subnet_mask: gateway_address/subnet_mask
ERRO[2017-03-31T13:33:08-06:00] Failed to collect 1bbaa091-57e3-4987-b86e-3d38262eb2c2 vpxd.log: Post https://vcsa1.corp.local/sdk: net/http: request canceled while waiting for connection WARN[2017-03-31T13:33:08-06:00] No log data for 1bbaa091-57e3-4987-b86e-3d38262eb2c2 vpxd.log ERRO[2017-03-31T13:33:08-06:00] -------------------- ERRO[2017-03-31T13:33:08-06:00] vic-machine-darwin failed: Create timed out: if slow connection, increase timeout with --timeout
Resolution: Delete the VCH instance and specify bridge network in CIDR with a minimum of a /12. I am not sure why it needs this many networks, but I’ll save everyone the pain that tries to specify /16 or higher. I also needed to specify the bridge network since the default is 172.16.0.0/12 and that overlapped with my public network which was causing the same error.
Phew… got through all the errors with some log review and patience. Once complete you will see the following.
INFO[2017-03-31T15:47:50-06:00] Initialization of appliance successful INFO[2017-03-31T15:47:50-06:00] INFO[2017-03-31T15:47:50-06:00] VCH Admin Portal: INFO[2017-03-31T15:47:50-06:00] https://172.16.10.45:2378 INFO[2017-03-31T15:47:50-06:00] INFO[2017-03-31T15:47:50-06:00] Published ports can be reached at: INFO[2017-03-31T15:47:50-06:00] 172.16.10.45 INFO[2017-03-31T15:47:50-06:00] INFO[2017-03-31T15:47:50-06:00] Docker environment variables: INFO[2017-03-31T15:47:50-06:00] DOCKER_HOST=172.16.10.45:2376 INFO[2017-03-31T15:47:50-06:00] INFO[2017-03-31T15:47:50-06:00] Environment saved in virtual-container-host/virtual-container-host.env INFO[2017-03-31T15:47:50-06:00] INFO[2017-03-31T15:47:50-06:00] Connect to docker: INFO[2017-03-31T15:47:50-06:00] docker -H 172.16.10.45:2376 --tls info INFO[2017-03-31T15:47:50-06:00] Installer completed successfully
Verify VCH Connectivity with Docker Client
To verify connectivity, run the following command.
docker -H 192.168.0.246:2376 --tls info
If you get the error message below, you need to run the command “export DOCKER_API_VERSION=1.23” as shown below. For windows the command is “SET DOCKER_API_VERSION=1.23”
swhitney-m01:vic swhitney$ docker -H 172.16.10.45:2376 --tls info Error response from daemon: client is newer than server (client API version: 1.24, server API version: 1.23) swhitney-m01:vic swhitney$ export DOCKER_API_VERSION=1.23 swhitney-m01:vic swhitney$ docker -H 172.16.10.45:2376 --tls info Containers: 0 Running: 0 Paused: 0 Stopped: 0 Images: 0 Server Version: v0.8.0-7315-c8ac999 Storage Driver: vSphere Integrated Containers v0.8.0-7315-c8ac999 Backend Engine VolumeStores: vSphere Integrated Containers v0.8.0-7315-c8ac999 Backend Engine: RUNNING VCH mhz limit: 28352 Mhz VCH memory limit: 113.5 GiB VMware Product: VMware vCenter Server VMware OS: linux-x64 VMware OS version: 6.0.0 Plugins: Volume: Network: bridge vic-container Swarm: NodeID: Is Manager: false Node Address: Operating System: linux-x64 OSType: linux-x64 Architecture: x86_64 CPUs: 28352 Total Memory: 113.5 GiB Name: virtual-container-host ID: vSphere Integrated Containers Docker Root Dir: Debug Mode (client): false Debug Mode (server): false Registry: registry-1.docker.io Experimental: false Live Restore Enabled: false
We can also see the VCH deployed in a vApp in vCenter Server!
Let’s pull down a container and deploy it to ensure everything is functioning correctly. Run the following command to pull busybox from the VCH. Mine seemed to take a few minutes.
swhitney-m01:vic swhitney$ docker -H 172.16.10.45:2376 --tls pull busybox Using default tag: latest Pulling from library/busybox 7520415ce762: Pull complete a3ed95caeb02: Pull complete Digest: sha256:8d7fe3e157e56648ab790794970fbdfe82c84af79e807443b98df92c822a9b9b Status: Downloaded newer image for library/busybox:latest
Then, you can run the container!
docker -H 172.16.10.45:2376 --tls run -it --name containersftw busybox
We can see the VM in the vSphere Web Client as shown below.
There are many commands to play around with from the Docker client. Not all of the commands will be supported and we are continually adding more. For now, the list of available supported commands should be sufficient for enterprise deployments and developers who are already using or looking into Docker. I will be writing additional posts when I get the chance to play around with the docker commands as well as a networking deep dive entailing connecting containers to external networks for testing.