Could you please clarify, which exceptions you are experiencing? Your POC looks alright, the only issue is the `ignite.clientConnector.port` property, which is the same in all configs, they must be different if you are running the nodes on the same machine. Also please make sure to add the `--add-opens` JVM parameters mentioned in the documentation.
On Thu, Oct 2, 2025 at 5:00 PM Ihor Parashynets <[email protected]> wrote: > Thanks for the hint - I got the point. Now I'm getting some weird behavior > when starting the cluster that way - exceptions, some nodes go down, etc. > > > May I ask you for additional verification on a dumb PoC, which I created > as a maven project? I'm attaching it here - it has 3 dedicated classes, > which start "node[1-3]" respectively and "node3" is responsible for > initializing the cluster (I start every node manually). > > > Again - very appreciated for your help! > > > > Best regards > > > On 2025/10/02 10:16:47 Aleksandr Polovtsev wrote: > > > Thank you for the provided information, I was able to reproduce the > issue. > > > > > > The problem is with the non-intuitive Embedded API, which is targeted at > > > running single-node clusters. To solve your problem, you need to call > > > `IgniteServer#waitForInitAsync` on other two nodes, for example: > > > > > > InitParameters initParameters = InitParameters.builder() > > > .metaStorageNodeNames("node1", "node2", "node3") > > > .clusterName("cluster") > > > .build(); > > > > > > CompletableFuture.allOf( > > > node1.initClusterAsync(initParameters), > > > node2.waitForInitAsync(), > > > node3.waitForInitAsync() > > > ).join(); > > > > > > -- > > > With regards, > > > Aleksandr Polovtsev > > > > -- With regards, Aleksandr Polovtsev
