Quorum write failing partially and no rollback

(2 posts) (2 voices)

• Started 1 year ago by
• Latest reply from
• This topic is resolved

Tags:

No tags yet.

1. 

   Anonymous
   Unregistered

   Hello,

   Something in the doc is puzzling me ( http://www.datastax.com/docs/1.0/dml/about_writes ) :

   "For example, if using a write consistency level of QUORUM with a replication factor of 3, Cassandra will send the write to 2 replicas. If the write fails on one of the replicas but succeeds on the other, Cassandra will report a write failure to the client. However, the write is not automatically rolled back on the other replica."

   If it is not rolled back, does it means that upon a read repair or a node repair this data will be copied to other replicas ?

   Thanks in advance
   Nicolas.

   Posted 1 year ago #
2. 

   Anonymous
   Unregistered

   That is correct. In addition, hinted handoff (http://www.datastax.com/dev/blog/understanding-hinted-handoff) can also propagate the mutation to the node that failed to receive the write in the first place.

   Cassandra has no distributed transactions, and hence no distributed rollback. The error returned to the client only indicates that the contract/requirement (consistency level) that the client specified could not be fulfilled. That error doesn't make any claims about the actual state of the data from that write in the cluster.

   Best practices in that situation would be to do any of
   1) Have your client retry the write
   2) Have your client retry the write with a lower consistency level
   3) Have your client send a delete operation to overwrite that previous mutation (assuming that the columns you are writing are new columns and can be safely deleted).

   -Tupshin

   Posted 1 year ago #

RSS feed for this topic

Reply

You must log in to post.