glusterfs + sharding + afr question

[RolandK]

is anybody using proxmox with glusterfs ?

if i use afr / replication together with sharding (for example with dual node and 3rd node with arbiter volume), now if one node goes down and comes back after a short time, and there where only tiny changes to the virtual disk (hosted on glusterfs), is self-healing intelligent enough to only heal the shards where changes occured ?

in other words do we have "intelligent/fast/incremental network raid1 resync" with that ?

i the glusterfs docs/forums a lot for this, but did not find any information about that...

 

[RolandK]

https://www.linux-kvm.org/images/9/...entImprovementsInGlusterForVMImageStorage.pdf

Reducing heal time - sharding
● File is divided into shards with pre-defined shard size
● Default shard size is 4MB
● Heals only shards that need healing <- !

 

[RolandK]

nobody using glusterfs with proxmox ?