Proxmox / GlusterFS - Volume configuration

[KevinA]

Hello everybody,
First of all, a big thank you to the Proxmox developer community. I have been using Proxmox for several years now with GlusterFS and it works perfectly.Exactly, I'm using a Proxmox cluster of 3 Proxmox servers with a GlusterFS cluster for identical storage on each Proxmox server. GlusterFS contains the qcow2 disks of my VMs. This allows, in case of failure of one of the proxmox, to restart the VMs it hosts automatically on the other proxmox because the storage is replicated. Lately, I encounter more and more errors when restarting a VM on another proxmox.The errors are of the type:

• HS file system
• Database corruption
• General file corruption (.php, .html, etc.)

I use the following settings on my GlusterFS volume:

gluster volume set <glusterfs_volume_name> network.ping-timeout 30
gluster volume set <glusterfs_volume_name> cluster.data-self-heal-algorithm full
gluster volume set <glusterfs_volume_name> features.shard-block-size 64MB
gluster volume set <glusterfs_volume_name> features.shard on
gluster volume set <glusterfs_volume_name> performance.io-cache off
gluster volume set <glusterfs_volume_name> performance.read-ahead off
gluster volume set <glusterfs_volume_name> performance.quick-read off
gluster volume set <glusterfs_volume_name> cluster.eager-lock enable
gluster volume set <glusterfs_volume_name> network.remote-dio enable
gluster volume set <glusterfs_volume_name> cluster.server-quorum-type server
gluster volume set <glusterfs_volume_name> cluster.quorum-type auto
gluster volume set <glusterfs_volume_name> performance.readdir-ahead on

Are these settings still valid on Proxmox 7.x and GlusterFS 9/10? Asking otherwise, what configuration do you recommend for glusterFS 9/10 volumes with Proxmox 7.x volume configuration?

Thanks,

 

[RolandK]

https://forum.proxmox.com/threads/proxmox-with-glusterfs-storage-backend-vms-get-corrupted.128288/


here is another report on corruption happening with glusterfs