Hello!
We have 3 dbs. With relp factor 3. Ver 4.5.3.2
So we delivered record. At the same moment, one server goes down. So we saving this record. And we have sucess status for writing. (2/3 are online) Now from another server we want to read this record (with 3 second timeout). But… we have message that this record dont found. Why so? What to do?
Thanks
This can happen if you are using the write-commit-level-override master feature or the per transaction policy from the client, for example for java see the commitLevel on the WritePolicy.
It can also happen if the cluster wasn’t fully formed when the write took place. The master for this record could have been in a single node sub-cluster when the write succeeded. In which case it would have only succeeded to a single node which may have been the node that failed.
If these types of consistency issues are a major concern I recommend looking into our Strong Consistency feature.
This topic was automatically closed 6 days after the last reply. New replies are no longer allowed.