An attempt to recreate a load balancer in Provider's Control Panel leads to the LB in Unknown state.
The following error is found in
/var/log/pa/vps.log on IM server:
2017-04-19 03:03:08,737 (42f2a550-e96d-4438-9b91-1284d04d0209) ERROR ClusterImpl [Grizzly-worker(4)] - Failed to update pipeline #18 [(RECREATE_LB for: VeId[1000001.LB-test-2 (lb)]; @HN: 12), created at node [im1], step 1 ([VE re-creation start]), mode: EXEC, state:SHARED_RUN, reqIs: null] com.hazelcast.nio.serialization.HazelcastSerializationException: com.esotericsoftware.kryo.KryoException: java.lang.NullPointerException Serialization trace: backendVEs (com.parallels.c2u.im.tasks.LbConfigurationTask) tasks (com.parallels.c2u.im.tasks.Pipeline) at com.hazelcast.nio.serialization.SerializationServiceImpl.handleException(SerializationServiceImpl.java:354) ~[hazelcast-all-3.3.3.jar:3.3.3]
A PACI-im service restart hangs with the following messages repeated in the log:
2017-04-19 03:18:11,645 () INFO ClusterImpl [main] - There are 1 (or less) active pipelines in the cluster-wide queue
Any new Load Balancer created afterwards appear in Unknown too with the following message in logs:
2017-04-15 09:49:24,278 (7d013902-ff6c-40a5-89c6-fa0f65c420bb) DEBUG CheckImageUsePrecondition [Grizzly-worker(2)] - Checking image [1.LoadBalancer] use precondition 2017-04-15 09:49:24,278 (7d013902-ff6c-40a5-89c6-fa0f65c420bb) DEBUG CheckImageUsePrecondition [Grizzly-worker(2)] - Image [1.LoadBalancer] is busy
The issue is recognized as CCU-14873.
Contact Odin Technical Support to resolve the situation.