Red Hat Bugzilla – Bug 1271532
BPM cluster fails to setup due to NPE in uberfire
Last modified: 2015-11-04 07:50:16 EST
Created attachment 1082735 [details]
server log node one
Description of problem:
Having a BPM cluster with 2 EAP nodes in a domain, the cluster fails to initialize due to NPE in org.uberfire.io.impl.cluster.helix.ClusterServiceHelix.getNodeStatus(). The first node starts up just fine, but the second node fails to deploy Business Central.
The issue occurs when cluster starts with a clean environment - including DB, .niogit and .index for both nodes, etc. After stopping the EAP domain and running it again (without erasing environment mentioned above), both cluster nodes start.
Version-Release number of selected component (if applicable):
always with a clean environment, see above
Steps to Reproduce:
1. setup a BPM cluster with 2 nodes using EAP domain mode; use clean environment
2. start EAP via bin/domain.sh
3. watch for errors in server logs, check Business Central availability on both nodes
Please note that currently there is a blocker in EAP 6.4.3, applying a patch (or using EAP 6.4.4 when it comes) is necessary in order to get to this issue. See bug 1265723 for more details.
Created attachment 1082738 [details]
server log node two
Created attachment 1082742 [details]
Created attachment 1082743 [details]
Created attachment 1082840 [details]
helix and zookeeper configuration script
Verified with BPMS-6.2.0.ER5