[CASSANDRA-14129] BatchStatement (LOGGED BATCH) fails: Not enough replica available for query at consistency ONE (1 required but only 0 alive)

• Type: Bug
• Status: Open
• Priority: Normal
• Resolution: Unresolved
• Component/s: Legacy/Coordination
• Severity: Normal
• Assignee: Unassigned
• Reporter: Erik Scheuter
• Created: 18/Dec/17 09:53
• Updated: 16/Apr/19 09:29

Description

In our test-cluster I get the error "Not enough replica available for query at consistency ONE (1 required but only 0 alive)" when 1 node is down and a BatchStatement is used. By default it uses LOGGED BATCHES and DCAwareRoundRobinPolicy.
We have 2 DC's, 4 nodes and a RF 2:2. Write Consistency of QUORUM (not LOCAL QUORUM).

I noticed as soon as the driver marked the LocalDC containing the one with the node down, this error occurs. If the other DC is marked as the LocalDC, everything works as expected.

The issue https://issues.apache.org/jira/browse/CASSANDRA-7870 points to the comment from Olivier Michallat at StackOverflow:
This batch log must be replicated to at least one replica other than the coordinator, otherwise that would defeat the above mechanism.

I think it needs an other replica in the same DC.

These constraints can lead to downtime, so I expect them to find in the documentation (and a more detailed error message).

Used:
Cassandra 2.0.17
Datastax-driver 3.3.0