I have a super noob question,

replicationPerPartition

if I do 1 does that mean no replication or like one more copy of the data if I do 2 is original data + 2 more?

It is the absolute value 1 -> 1 copy, 2 -> 2 copies.

oh cool, so i have a question, i have this topic with 8 partitions, when i’m consuming in the realtime table, i see 8 segments being generated, i have 2 servers, then i see that that’s split between the 2 servers to 4 segments on one and 4 segments on another one, if i do

replicationPerPartition

1 should i see double in both servers? like 8, 8

replicationPerPartition = 1

-> You will see 4 + 4

so is it safe to assume that

replicationPerPartition = 1

if i lose one server i still have the rest of the data in the other one?

If you lose 1 server, then 4 of the partitions will no longer be available. The other 4 on the other server will be available: https://apache-pinot.slack.com/archives/C011C9JHN7R/p1645131340156869?thread_ts=1645130053.977889&cid=C011C9JHN7R

which means that data from those 4 partitions won’t be available on the other server right? If I do

replicationPerPartition = 2

even if i lose one server all the data will be in the one that is still alive

Replication =1, you loose one server ->The 4 partitions on that lost server are no longer available. If replication = 2, Bother servers will have all 8 partitions. Now if you lose 1 server, you still got 1 full copy of the data from the other server.

perfecto, thank you, i have to do

replication 2

then, same would go for the offline table right? when I have it setup with the pinot offline managed flows i should also do

replication 2

probably

Yes, please ensure you have set replication > 1 for offline table config.