3 ms·
> were restoring to a new node Zfs send / recv or replication. > I used to run a bunch of Postgres nodes at a similar scale. The most painful parts (by far) w
by Eikon 2y ago
> were restoring to a new node
Zfs send / recv or replication.
> I used to run a bunch of Postgres nodes at a similar scale. The most painful parts (by far) were restoring to a new node and major version upgrades.
Any tricks you used for those parts?
Replication makes this pretty painless :)