After a bit more investigation it appears this can be solved by setting peerClassLoadingEnabled=false. Probably not a huge problem but we currently have a grid set up with peerClassLoadingEnabled=true which has a couple of different separate uses, so we would either need to change our deployment process for the other worker nodes to ensure all required classes are deployed, or partition our grid. However I'd prefer to leave peer class loading on if this is possible.
Just for info I'm currently using the TcpDiscoveryJdbcIpFinder for grid discovery and partition my worker nodes using attributes e.g. clusterName=xxx. Is it possible to use the jdbc finder on the same database to specify separate grids, or would I need to set up a separate database for each grid? I prefer to use the jdbc method rather than hard coding the servers using TcpDiscoveryVmIpFinder. Thanks for any info -- View this message in context: http://apache-ignite-users.70518.x6.nabble.com/Loading-a-native-library-on-a-grid-node-tp11227p11230.html Sent from the Apache Ignite Users mailing list archive at Nabble.com.
