Red Hat Bugzilla – Bug 1265723
BPM cluster does not respect uberfire configuration system properties
Last modified: 2015-11-04 07:49:56 EST
Created attachment 1076249 [details]
Running a BPM cluster on EAP 6.4.3 domain containing two nodes on a single machine, cluster fails to start due to address already in use /127.0.0.1:9418. Both nodes have been configured to bind to a different port using system properties org.uberfire.nio.git.daemon.port and org.uberfire.nio.git.daemon.hostport to values 9418 resp. 9419.
.niogit location has been specified by org.uberfire.nio.git.dir to point to different directories for each server node, but the .niogit directory has been created in its default location. This applies to other directories like .index as well, which resulted in some runs in org.apache.lucene.store.LockObtainFailedException: Lock obtain timed out (probably happened before starting git daemon)
Scanning for open ports revealed that only default ports (8001, 9418) were opened, in contrast with configured values. See the attached configuration files.
Created attachment 1076250 [details]
tested version: BPM Suite 6.2.0.ER3
One comment I have is that this problem does NOT seem to occur on Windows boxes. There are no port collisions, but there is an NPE relating to uberfire sometimes. I can post logs if that will help.
Verified with BPMS-6.2.0.ER5