Vladimir Ozerov created IGNITE-7971:
---------------------------------------
Summary: SQL: make sure CREATE INDEX doesn't corrupt the cluster
Key: IGNITE-7971
URL: https://issues.apache.org/jira/browse/IGNITE-7971
Project: Ignite
Issue Type: Task
Components: sql
Affects Versions: 2.4
Reporter: Vladimir Ozerov
Fix For: 2.5
Consider the following scenario:
1) Define query entity with field {{X}} of type {{int}}
2) Start two nodes
3) Put an object with field {{X}} of type {{int}} to node 1
4) Put an object with field {{X}} of type {{*String*}} to node 2
5) Invoke {{CREATE INDEX}} on that field
We need to investigate what would happen. In the worst case index creation will
fail on node 2. But it seems that we do not perform any rollback on server
nodes in that case, only client node is informed about the problem. If it is
so, we need to answer the following questions:
1) What would happen to subsequent SQL queries initiated from different nodes.
Will they work?
2) What would happen to data in case of restarts with persistence?
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
