-
-
Notifications
You must be signed in to change notification settings - Fork 2.2k
Failover Master Server
Some user will ask for no single point of failure. Although google runs its file system with a single master for years, no SPOF seems becoming a criteria for architects to pick solutions.
Luckily, it's not too difficult to enable Weed File System with failover master servers.
This section is a quick way to start 3 master servers and 3 volume servers. All done!
weed server -master.port=9333 -dir=./1 -volume.port=8080 \
-master.peers=localhost:9333,localhost:9334,localhost:9335
weed server -master.port=9334 -dir=./2 -volume.port=8081 \
-master.peers=localhost:9333,localhost:9334,localhost:9335
weed server -master.port=9335 -dir=./3 -volume.port=8082 \
-master.peers=localhost:9333,localhost:9334,localhost:9335
Or, you can use this "-peers" settings to add master servers one by one.
> weed server -master.port=9333 -dir=./1 -volume.port=8080
> weed server -master.port=9334 -dir=./2 -volume.port=8081 -master.peers=localhost:9333
> weed server -master.port=9335 -dir=./3 -volume.port=8082 -master.peers=localhost:9334
The master servers are coordinated by Raft protocol, to elect a leader. The leader took over all the work to manage volumes, assign file ids. All other master servers just simply forward requests to the leader.
If the leader dies, another leader will be elected. And all the volume servers will send their heartbeat together with their volumes information to the new leader. The new leader will take the full responsibility.
During the transition, there could be moments where the new leader has partial information about all volume servers. This just means those yet-to-heartbeat volume servers will not be writable temporarily.
Now let's start the master and volume servers separately, the usual way.
Usually you would start several (3 or 5) master servers, then start the volume servers:
weed master -port=9333 -mdir=./1
weed master -port=9334 -mdir=./2 -peers=localhost:9333
weed master -port=9335 -mdir=./3 -peers=localhost:9334
# now start the volume servers, specifying any one of the master server
weed volume -dir=./1 -port=8080
weed volume -dir=./2 -port=8081 -mserver=localhost:9334
weed volume -dir=./3 -port=8082 -mserver=localhost:9335
These 6 commands will actually functioning the same as the previous 3 commands from the cheatsheet.
Even though we only specified one peer in "-peers" option to bootstrap, the master server will get to know all the other master servers in the cluster, and store these information in the local directory.
If you need to restart the master servers, just run the master servers WITHOUT the "-peers" option.
weed master -port=9333 -mdir=./1
weed master -port=9334 -mdir=./2
weed master -port=9335 -mdir=./3
To understand why, remember that the cluster information is "sticky", meaning it is stored on disk. If you restart the server, the cluster information stay the same, so the "-peers" option is not needed again.
This "sticky" cluster information can cause some misunderstandings. For example, here is one:
https://code.google.com/p/weed-fs/issues/detail?id=70
The previously used value "localhost" would come up even not specified. This could cause your some time to figure out.
- Replication
- Store file with a Time To Live
- Failover Master Server
- Erasure coding for warm storage
- Server Startup Setup
- Environment Variables
- Filer Setup
- Directories and Files
- Data Structure for Large Files
- Filer Data Encryption
- Filer Commands and Operations
- Filer JWT Use
- Filer Cassandra Setup
- Filer Redis Setup
- Filer YugabyteDB Setup
- Super Large Directories
- Path-Specific Filer Store
- Choosing a Filer Store
- Customize Filer Store
- Migrate to Filer Store
- Add New Filer Store
- Filer Store Replication
- Filer Active Active cross cluster continuous synchronization
- Filer as a Key-Large-Value Store
- Path Specific Configuration
- Filer Change Data Capture
- Cloud Drive Benefits
- Cloud Drive Architecture
- Configure Remote Storage
- Mount Remote Storage
- Cache Remote Storage
- Cloud Drive Quick Setup
- Gateway to Remote Object Storage
- Amazon S3 API
- AWS CLI with SeaweedFS
- s3cmd with SeaweedFS
- rclone with SeaweedFS
- restic with SeaweedFS
- nodejs with Seaweed S3
- S3 API Benchmark
- S3 API FAQ
- S3 Bucket Quota
- S3 API Audit log
- S3 Nginx Proxy
- Hadoop Compatible File System
- run Spark on SeaweedFS
- run HBase on SeaweedFS
- run Presto on SeaweedFS
- Hadoop Benchmark
- HDFS via S3 connector
- Async Replication to another Filer [Deprecated]
- Async Backup
- Async Filer Metadata Backup
- Async Replication to Cloud [Deprecated]
- Kubernetes Backups and Recovery with K8up