Niels Ejrnæs created IGNITE-12058: ------------------------------------- Summary: AtomicReference issue with different userVersions Key: IGNITE-12058 URL: https://issues.apache.org/jira/browse/IGNITE-12058 Project: Ignite Issue Type: Bug Affects Versions: 2.7.5 Reporter: Niels Ejrnæs Attachments: ignite-playground.tar.gz
When running DeploymentMode CONTINUOUS or SHARED and connecting a client node with a different userVersion from the server nodes - IgniteAtomicLongs and IgniteAtomicReferences cannot be fetched directly from Ignite. {code:java} IgniteAtomicLong atomicLong = ignite.atomicLong("long", 0, false); IgniteAtomicReference<String> atomicReference = ignite.atomicReference("reference", "", false); {code} It throws an Exception {noformat} SEVERE: Error deserializing job response: GridJobExecuteResponse [nodeId=b52607ba-b6d1-47bc-983c-380af8a679bd, sesId=f13d9c48c61-0fabe984-ecf1-4acf-84ac-3bdd478405f7, jobId=023d9c48c61-0fabe984-ecf1-4acf-84ac-3bdd478405f7, gridEx=null, isCancelled=false, retry=null] class org.apache.ignite.IgniteCheckedException: Failed to unmarshal object with optimized marshaller at org.apache.ignite.internal.util.IgniteUtils.unmarshal(IgniteUtils.java:10147) ... Caused by: class org.apache.ignite.internal.IgniteDeploymentCheckedException: Failed to obtain deployment for class: org.apache.ignite.configuration.CacheConfiguration$IgniteAllNodesPredicate ... {noformat} It works if you use a compute job to fetch it. {code:java} long remoteLong = ignite.compute().call(() -> { return ignite.atomicLong("long", 0, false).get(); }); ignite.compute().call(() -> { return ignite.atomicLong("long", 0, false).incrementAndGet(); }); String remoteReference = ignite.compute().call(() -> { return ignite.atomicReference("reference", "", false).get(); }); {code} I've attached a small reproducer project. -- This message was sent by Atlassian JIRA (v7.6.14#76016)