proxmox ve Deploying a two-node HA cluster and glusterfs distributed file system

root@pve1:~# cat /etc/hosts
192.168.1.50  pve1
192.168.1.60  pve2
192.168.1.50 gluster1
192.168.1.60 gluster2

In the twopve(used form a nominal expression)/etc/In hostname, add the following
root@pve1.~# cat /etc/hostname
pve1
 
root@pve2:~# cat /etc/hostname
pve2

wget -O - https:///pub/gluster/glusterfs/9/ | apt-key add - 
DEBID=$(grep 'VERSION_ID=' /etc/os-release | cut -d '=' -f 2 | tr -d '"') 
DEBVER=$(grep 'VERSION=' /etc/os-release | grep -Eo '[a-z]+') 
DEBARCH=$(dpkg --print-architecture) 
echo deb https:///pub/gluster/glusterfs/LATEST/Debian/${DEBID}/${DEBARCH}/apt ${DEBVER} main > /etc/apt// 
apt update [note this step can be left out when the source has been determined to use a specific source]
aptinstall -y glusterfs-server

option -address gluster1 
option -address gluster1 
option -address gluster1

option -address gluster2 
option -address gluster2 
option -address gluster2

systemctl enable  
systemctl start 

gluster peer probe gluster1
Display: peer probe: success and it's OK

command analysis
* gluster volume create: This is the GlusterFS command to create a new volume.
* VMS: This is the name you assign to the new volume. In this example, the name of the volume is VMS.
* replica 2: This specifies the type of volume and the replication factor。replica Indicates that this is a replicated volume，2 Indicates that the data will be replicated on two nodes. This means that you have two copies of the data, one on the master node and the other on the replica node.
* gluster1:/data/s: This is the first storage path。gluster1 be GlusterFS The name of a node in the cluster or IP address，/data/s is the directory on this node used to store GlusterFS volume data.
* gluster2:/data/s: This is the second storage path。Similar to the first path，but specifies a second node and storage directory。In this example，The data will be copied to the gluster1 cap (a poem) gluster2 These two nodes on the /data/s directory.
When this command is executed, GlusterFS creates a replicated volume called VMS on both nodes, gluster1 and gluster2, and stores the data in the two nodes'/data/s directory for replication. Doing so improves the reliability and availability of the data because if one of the nodes fails, the copy of the data on the other node is still available.
However, as mentioned in the error message you encountered, creating replicated volumes (especially when there are only two copies) has brain cracks (split-brain）risks。Brain splitting is when two or more nodes both consider themselves to be master nodes and both are receiving write operations，Data may become inconsistent。To avoid this，Arbitration nodes can be used（arbiter）or increase the replication factor to 3 or more. But in many cases, a simple two-node replicated volume is sufficient for most applications.
-------------------------------
Analyzing the gluster vol start VMS Command
1. Start Volume Services: This command starts services for the volume named VMS in the GlusterFS cluster, enabling clients to begin accessing data on that volume.
2. Ensuring Data Availability: When a volume is started, GlusterFS ensures that the data is available between nodes in the cluster and will manage and replicate the data based on the type of volume (e.g., distributed replicated volume).
3. Checking Node Status: Before starting a volume, GlusterFS checks the status of all nodes in the cluster that are participating in the volume to make sure they are all available and in the correct configuration.
4. Handling Client Requests: once a volume is successfully started, clients can access the data stored on it by mounting the volume. glusterFS handles read and write requests from clients and ensures that the data is consistent across the cluster.
5. Load Balancing: For distributed volumes and distributed replicated volumes, GlusterFS automatically performs load balancing at startup to ensure that data is evenly distributed across nodes, thus improving overall performance and reliability.
6. Monitoring and Logging: After a volume is started, GlusterFS continuously monitors the status and performance of the volume and records relevant log information. This information is useful for subsequent troubleshooting and performance tuning.
In summary, the gluster volume start VMS command is used to start a volume named VMS in a GlusterFS cluster to ensure data availability, consistency, and performance, and to handle read and write requests from clients. Before executing this command, you need to ensure that all nodes in the GlusterFS cluster are properly configured and can communicate with each other.

The two nodes of theglusteroccurssplit-Brain problem, that is, two nodes have the same number of votes, and no one listens to anyone, the solution is as follows.
gluster vol set VMS-timeout 5
gluster volume heal VMS enable
gluster vol set VMS -reads false
gluster vol set VMS -count 1
gluster vol set VMS network.ping-timeout 2
gluster volume set VMS -child-policy mtime
gluster volume heal VMS granular-entry-heal enable
gluster volume set VMS -self-heal-algorithm full

modifications /etc/pve/
Add it to quorum and it looks like this.
quorum {
    provider: corosync_votequorum
    expected_votes.1 
    two_node: 1 
}