I have a cluster of
- One dgraphzero with --replicas 3
- 6 alphas split over two groups
I took two alphas of group 1 down, expecting writes to fail, but reads to succeed
The client connects to an alpha that is up. The query is issued, but no response comes back until the request eventually times out.
In the documentation I can only find quorum requirements for write transactions, but never explicitly for read.
Is it expected that reads block, if 2/3 of a group are down?
Using dgraph 21.03.02