On 11/25/22 07:29, Katja Zinchenko wrote:
Afterwards I have backup in the file system of a random server. So, I
have to allocate enough memory on each node, which will probably stay
empty. Are there any other solutions to define which server will be used
for the backups?
Maybe one requirement escaped your notice: The Collections API backup
functionality requires that every server have the backup location
mounted to the same storage on the same path. If you look at the
requests shown here:
https://solr.apache.org/guide/solr/latest/deployment-guide/collection-management.html#backup
The first thing it says there is "Backs up Solr collections and
associated configurations to a shared filesystem - for example a Network
File System."
Some kind of shared storage (usually SMB, NFS, or similar) is a
requirement for that backup/restore functionality. I have never seen
anything that will specify which machine actually performs the backup.
I *think* it's probably the leader replica for each shard, and the
replica that is leader can change.
Thanks,
Shawn
