Details
-
Bug
-
Status: Open
-
Major
-
Resolution: Unresolved
-
None
-
None
-
None
-
Hosts: CentOS6.7_x86_64
NiFi: 1.1.2
-
Important
Description
Since the beginning of 2017, we have setup 4 NiFi 1.1.2 cluster, one of them is our BTS environment, one is our Prod environment, totally we have met the problem more than 20 times, so we organize a team for researching after same issue occurred on Prod.
After debugging and analysis, we found the cause is that every time cluster node connect to cluster, cluster will get cluster data flow then compare with local data flow, if it's different, cluster node can't connect to cluster. So we start to research why the data flow is different next, and we found a case which will cause this problem.
We open two browser tab, we called them A and B; tab A connect to node-1, at first, we disconnect node-1 in tab B, next, we will add a processor on tab A, finally we refresh tab A we will find node-1 can't connect to cluster due to difference between local data flow and cluster data flow.