This is a basic configuration of a multiple-node MooseFS cluster based on the Debian Buster Docker image. It consists of a Master Server, a CGI Monitoring Interface Server, 4 Chunkservers, and one Client. After a successful installation, you will have a fully working MooseFS cluster to play with its amazing features.
- All MooseFS processes are now correctly handling signals.
- Metadata and data are now persistent and mounted as volumes.
- TEST and PROD moosefs master metadata behavior defined by
MFS_ENVvariable. - Specify storage size per chunk server (env:
SIZE) default not defined. Depends on your local storage free space. - Specify label per chunk server (env:
LABEL, default: empty). - Switched to debian:buster as a base image.
- Example with 4 chunk servers (labels: M, MB, MB, B).
File docker-compose.yml
- Master Server:
172.20.0.2 - CGI Monitoring Interface: http://localhost:9425, on Linux also http://172.20.0.3:9425
- Metalogger:
172.20.0.4 - Chunkserver 1:
172.20.0.11, labels:M - Chunkserver 2:
172.20.0.12, labels:M, B - Chunkserver 3:
172.20.0.13, labels:M, B - Chunkserver 4:
172.20.0.14, labels:B - Client:
172.168.20.0.100
Install Docker with Docker Composer from https://docs.docker.com/compose/install
Clone MooseFS docker config files:
git clone https://github.com/moosefs/moosefs-docker-cluster
cd moosefs-docker-cluster
Build and run:
docker-compose build
docker-compose up
On Linux OS run docker-compose as root:
sudo -E docker-compose build
sudo -E docker-compose up
You can also run docker-compose in detached mode. All running Docker nodes will run in the background, so Docker console output will be invisible.
sudo -E docker-compose build
sudo -E docker-compose up -d
You can check if instances are running:
docker ps
You should have 1 Master Server, 1 Metalogger, 4 Chunkservers and 1 Client running (first configuration). The expected output should be similar to the following:
CONTAINER ID IMAGE COMMAND CREATED STATUS PORTS NAMES
abf5910b53bc moosefsdockercluster_mfsclient "mfsmount -f /mnt/mo…" 7 minutes ago Up 7 minutes mfsclient
7a1152cc31f3 moosefsdockercluster_mfschunkserver3 "chunkserver.sh" 7 minutes ago Up 7 minutes 9422/tcp mfschunkserver3
b8c2cd770187 moosefsdockercluster_mfschunkserver2 "chunkserver.sh" 7 minutes ago Up 7 minutes 9422/tcp mfschunkserver2
100f20683b3a moosefsdockercluster_mfschunkserver1 "chunkserver.sh" 7 minutes ago Up 7 minutes 9422/tcp mfschunkserver1
68ffb70ab361 moosefsdockercluster_mfschunkserver4 "chunkserver.sh" 7 minutes ago Up 7 minutes 9422/tcp mfschunkserver4
82a2c3bd831d moosefsdockercluster_mfsmetalogger "metalogger.sh" 7 minutes ago Up 7 minutes mfsmetalogger
05736e4bdd3c moosefsdockercluster_mfscgi "mfscgiserv -f" 7 minutes ago Up 7 minutes 0.0.0.0:9425->9425/tcp mfscgi
e83a1fb062a1 moosefsdockercluster_mfsmaster "master.sh" 7 minutes ago Up 7 minutes 9419-9421/tcp mfsmaster
For example, if you like to attach to the client node execute this command:
docker exec -it mfsclient bash
To detach from container, just press Ctrl + D keys combination.
MooseFS filesystem is mounted at /mnt/moosefs. If everything is ok you should see this ASCII art:
cat /mnt/moosefs/.mooseart
\_\ /_/
\_\_ _/_/
\--/
/OO\_--____
(__) )
``\ __ |
||-' `||
|| ||
"" ""
The MooseFS CGI Monitoring Interface is available here: http://localhost:9425.
Also on Linux, CGI Server container is available at the IP address: http://172.20.0.3:9425 (be aware of a local 172.20.0.x network).
Your MooseFS Docker cluster is persistent. It means all files you created in the /mnt/moosefs folder will remain there even after turning containers off.
All data and metadata files are stored in the host ./data directory.
There might be situations where you would want to setup a config file on the container start. For that scenario you can pass the config file as a base64 encoded text. For example lets say you want to setup your chunk servers to connect to master in k8s cluster where IP's are dynamically assigned to pods. You have you master yaml definition set up as:
apiVersion: apps/v1
kind: Deployment
metadata:
name: moosefs-master
namespace: storage
spec:
replicas: 1
selector:
matchLabels:
app: moosefs-master
template:
metadata:
labels:
app: moosefs-master
spec:
nodeSelector:
"beta.kubernetes.io/os": linux
containers:
- name: moosefs-master
image: moosefs/master:latest
resources:
requests:
cpu: 100m
memory: 128Mi
limits:
cpu: 250m
memory: 256Mi
ports:
- containerPort: 9419
- containerPort: 9420
- containerPort: 9421
volumeMounts:
- name: moosefs-master-mfs
mountPath: /var/lib/mfs
volumes:
- name: moosefs-master-mfs
azureDisk:
kind: Managed
diskName: MooseMasterMfs
diskURI: /subscriptions/<subscriptionID>/resourceGroups/<resourceGroup>/providers/Microsoft.Compute/disks/MooseMasterMfs
---
apiVersion: v1
kind: Service
metadata:
name: moosefs-master
namespace: storage
spec:
type: NodePort
ports:
- port: 9419
targetPort: 9419
name: listen-metalogger
- port: 9420
targetPort: 9420
name: listen
- port: 9421
targetPort: 9421
name: listen-client
selector:
app: moosefs-master
This will reserve an IP in the cluster where the ports will be reached.
In order for your chunkservers to automatically connect to this IP you would need to have your mfschunkserver.cfg defined as:
MASTER_HOST = $MOOSEFS_MASTER_SERVICE_HOST
CSSERV_LISTEN_PORT = $MOOSEFS_CHUNKSERVER_SERVICE_PORT
DATA_PATH = /mnt/hdd0/mfs
MOOSEFS_MASTER_SERVICE_HOST variable is set by k8s cluster and contains IP where the master service is accessible by. MOOSEFS_CHUNKSERVER_SERVICE_PORT this is the port on which we will expose our chunk server.
Base64 encoded config data is:
TUFTVEVSX0hPU1QgPSAkTU9PU0VGU19NQVNURVJfU0VSVklDRV9IT1NUCkNTU0VSVl9MSVNURU5fUE9SVCA9ICRNT09TRUZTX0NIVU5LU0VSVkVSX1NFUlZJQ0VfUE9SVApEQVRBX1BBVEggPSAvbW50L2hkZDAvbWZzCg==
Now we can spin up chunkserver(s) with:
apiVersion: apps/v1
kind: Deployment
metadata:
name: moosefs-chunkserver-1
namespace: storage
spec:
replicas: 1
selector:
matchLabels:
app: moosefs-chunkserver-1
template:
metadata:
labels:
app: moosefs-chunkserver-1
spec:
nodeSelector:
"beta.kubernetes.io/os": linux
containers:
- name: moosefs-chunkserver-1
env:
- name: MFS_CHUNKSERVER_CONFIG
value: TUFTVEVSX0hPU1QgPSAkTU9PU0VGU19NQVNURVJfU0VSVklDRV9IT1NUCkNTU0VSVl9MSVNURU5fUE9SVCA9ICRNT09TRUZTX0NIVU5LU0VSVkVSX1NFUlZJQ0VfUE9SVApEQVRBX1BBVEggPSAvbW50L2hkZDAvbWZzCg==
- name: SIZE
value: 16
image: moosefs/chunkserver:latest
ports:
- containerPort: 9422
volumeMounts:
- name: moosefs-chunkserver-data-1
mountPath: /mnt/hdd0
volumes:
- name: moosefs-chunkserver-data-1
azureDisk:
kind: Managed
diskName: MfsHangfireData1
diskURI: /subscriptions/<subscriptionID>/resourceGroups/<resourceGroup>/providers/Microsoft.Compute/disks/MfsHangfireData1
---
apiVersion: v1
kind: Service
metadata:
name: moosefs-chunkserver-1
namespace: storage
spec:
type: NodePort
ports:
- port: 9422
targetPort: 9422
selector:
app: moosefs-chunkserver-1
Repeat this for other chunk servers modifying your base64 string accordingly. If you leave all the chunkservers on default port 9422 you can use same base64 encoded string TUFTVEVSX0hPU1QgPSAkTU9PU0VGU19NQVNURVJfU0VSVklDRV9IT1NUCkRBVEFfUEFUSCA9IC9tbnQvaGRkMC9tZnMK which will only set correct MASTER_HOST and DATA_PATH
| Image name | Image size | Pulls | Stars | Build |
|---|---|---|---|---|
| moosefs/master | ||||
| moosefs/chunkserver | ||||
| moosefs/client | ||||
| moosefs/metalogger | ||||
| moosefs/cgi |