Table of Contents
- Global Requirements
- Installation Steps
- CentOS Installation
- Files Inventory
EMC's Elastic Cloud Storage (ECS) 3.0 Software Docker multiple node deployment is intended to be used by developers and has multiple options for them. The most universal method for deploying ECS software is through Docker applied across whichever means are at your disposal (IaaS/PaaS/Hypervisor). In addition to this, you can leverage Vagrant for local VirtualBox instances.
In terms of cloud deployments, there are a range of options. The most compatible methods of deployment across any provider are the CentOS and CoreOS options to run the Docker instances.
While these instructions will cover Docker installations, you can find documentation on alternate deployment methods such as Puppet and Google Compute Engine (as well as previous releases) from the project repository on GitHub.
An ECS cluster deployment requires a minimum of four data nodes to provide the full set of features. Each one of the instances should have the following minimum requirements:
- Operative system: CentOS 7
- CPU/Cores: 4 Cores
- Memory: Mininum of 16 GB RAM
- Disks: An unpartitioned/raw disk with at least 100 GB of storage per disk. Multiple disks can be attached on each ECS node to increase capacity and performance. Each disk need to be unpartitioned before running the installation scripts.
Installation also requires internet connectivity to receive the requisite utility packages and Docker images.
Supported Host Operative Systems and Docker Version
The ECS Docker installation has been tested and verified against the following platforms:
|OS Name||Version||Docker Version|
The installation script is comprised of two main steps:
|1||Host Configuration||This step controls the flow and contains the configuration changes required for the host OS that will run the ECS Software Docker container.||3-15 min|
|2||ECS Object Configuration||This step performs the ECS configuration so it can start serving objects.||10-30 min|
CentOS is a well-known Linux distribution with the ability to deploy containers with Docker. Common public cloud platforms have CentOS templates ready to be used, so getting ECS 3.0 Software on a Docker container up is simple.
These steps are to be performed prior running the installation scripts on each of the ECS nodes:
- Attach Data Disk(s): ECS requires one or more disks to be attached to each host. The disk(s) will hold the object data store. The disk(s) will be formatted as XFS by the installation script.
The data disk(s) attached to each host must be unpartitioned or RAW. For example: We have a new host where we execute the command
fdisk -l:In the picture, we can see two disks: sda and sdb. A
mount -llooks like this: Now we attach a new disk to the host VM. The new disk /dev/sdc looks like this after executing
fdisk -lagain:Note: Depending on the environment or the cloud provider you may be using, the attached disk name(s) will be different. In this example, the attached disk came as /dev/sdc. The attached disk will be formatted and mounted during step 1, so do not mount the ECS data disk before executing step 1:
- Open Ports: ECS requires the following ports to be open:
Note: These are the most commonly-sued ports by ECS; please refer to List of Ports to be Open on the troubleshooting page for a full list. In addition, please refer to the ECS Security Configuration Guide and our the troubleshooting page if you encounter any issues.
Port Number Port Description 22 SSH, needed if using remote access 443 Port used for accessing the ECS management Website 3218 Port used by the CAS service 4443 ECS Management API Port 9020 Port used for the S3 API 9021 Port used for the S3 API on HTTPS 9022 Port used for Atmos API 9023 Port used for Atmos API on HTTPS 9024 Port used for SWIFT API 9025 Port used for SWIFT API on HTTPS 9100 Port used for DT Query service 9101 Port used for Diagnostic Service Index
- Network configuration: Define your network configuration. ECS Data nodes should be on the same subnet and be able to talk to each other. This is an example:
‼️ WARNING: This is a destructive operation. Existing data on selected storage devices will be overwritten. Existing Docker installations AND images will be removed.
The following section needs to be performed on each one of the ECS nodes:
- Perform Updates: Perform a Yum update using
sudo yum updateand download packages required for installation using
sudo yum install git tar wget
on each host.
- Git Clone/Pull the repository on each host: https://github.com/EMCECS/ECS-CommunityEdition
- Navigate to the /ecs-multi-node folder.
- Gather the IP addresses, desired hostnames, ethernet adapter name (which can be obtained by executing ifconfig on the host), and designated data disk(s). For example:
Hostname IP Address Disk Name Ethernet Adapter ecstestnode1 10.0.1.10 sdc sdd eth0 ecstestnode2 10.0.1.11 sdc sdd eth0 ecstestnode3 10.0.1.12 sdc sdd eth0 ecstestnode4 10.0.1.13 sdc sdd eth0
- Use gathered values for each ECS node (IP addresses, hostnames, ethernet adapter name, disk names) to build the
step1_ecs_multinode_install.pyscript, which will be the same across all nodes. Be advised that the hostname can not be localhost for any node. For our example values, the command should look like this:
sudo python step1_ecs_multinode_install.py --ips 10.0.1.10 10.0.1.11 10.0.1.12 10.0.1.13 --hostnames ecstestnode1 ecstestnode2 ecstestnode3 ecstestnode4 --disks sdc sdd --ethadapter eth0
The execution of this script is will take about 3-15 minutes depending on how many packages need to be installed or updated and the speed of certain services on the host. For a list of all arguments with their full descriptions and including more detailed options, use the --help flag, e.g. python step1_ecs_singlenode_install.py --help
- At the conclusion of step1, you may have to wait a few minutes until the administrative web UI becomes available. ECS' administrative portal can be accessed from the data node on port 443 ( https://<ecs-node-ip-address> ). Once you see the screen below, you are ready to execute step 2.
The next step is the ECS Object configuration. This can be accomplished in two ways:
- ECS' Administration UI: Please follow these Instructions.
- Automated script: Follow the instructions in the section below.
Both methods provide the same results; the first walks you through ECS' administrative web interface and the second uses ECS' Management API (exposed on port 4443 and port 9010)
ECS Object Configuration via an automated script
- Navigate to the /ecs-multi-node folder
- Verify that the
step2_object_provisioning.pyscript for the environment you are in can access the 4443 and 9011 ports of the host machine, such as through the output of nmap -sT -O localhost
- Before executing
step2_object_provisioning.py, decide upon values for the following variables:
Variable Name Variable Description Example Value ECSNodes IP Addressess of the ECS Nodes (comma-delimited list). 10.0.1.10,10.0.1.11,10.0.1.12,10.0.1.13 NameSpace The objects' Namespace ns1 ObjectVArray The objects' Virtual Array ova1 ObjectVPool The objects' Virtual Pool ov1 DataStoreName The name of the Data Store. ds1 VDCName The name of the Virtual Data Center. vdc1 MethodName The name of the specific step to be executed. Leave blank to complete all provisioning steps. [empty]
Once the variables are defined, they should be placed in the script command. Using the example values, the command becomes:
sudo python step2_object_provisioning.py --ECSNodes=10.0.1.10,10.0.1.11,10.0.1.12,10.0.1.13 --Namespace=ns1 --ObjectVArray=ova1 --ObjectVPool=ovp1 --UserName=emccode --DataStoreName=ds1 --VDCName=vdc1 --MethodName=
For more granular way of executing the Object Configuration, you can follow the instructions on this document showing how to run the process step by step.
The execution of this script may take 10 to 30 minutes to complete.
After the successful execution of the ECS Object Configuration, the system is ready to begin serving objects. Object users can read and write objects using free tools like S3 browser
InIn addition, access to the ECS's administrative panel is available on any node via https://<ecs-node-ip-address>. The default login and password for the portal is
root/ChangeMe (which you will be prompted to change when first accessing the portal)
To remove an existing installation of ECS Community Edition and clean up its associated resources, you can run the step1 script with the --cleanup option enabled, e.g.
sudo python step1_ecs_multinode_install.py --ips 10.0.1.10 10.0.1.11 10.0.1.12 10.0.1.13 --hostnames ecstestnode1 ecstestnode2 ecstestnode3 ecstestnode4 --disks sdc sdd --ethadapter eth0 --cleanup
This will remove the ECS Community Edition docker container, remove local directories linked to resources within the container, and de-partition, unmount, and de-register any associated data disks.
If you have any issues with the installation, you can review this page for troubleshooting tips and/or go to the support section bellow.
Please file bugs and issues at the Github issues page. For more general discussions you can contact the EMC Code team at Google Groups or tagged with EMC on stackoverflow.com. The code and documentation are released with no warranties or SLAs and are intended to be supported through a community-driven process.