Open coin8086 opened 5 years ago
This should be due to Linux node manager listening port conflict. Need to be fixed by modifying config in ACM Linux nodemanager, ACM Node Agent and HpcNodeAgent.
ACM Linux nodemanager config when installing:
- "ListeningUri":"http://0.0.0.0:40000"
+ "ListeningUri":"http://0.0.0.0:40010"
ACM Node Agent code change in file hpc-acm.git\src\Services\NodeAgent\NodeCommunicatorOptions.cs:
- public string NodeManagerUriBase { get; set; } = "http://localhost:40000/api";
+ public string NodeManagerUriBase { get; set; } = "http://localhost:40010/api";
HpcNodeAgent code change in file One\Azure\hpc\src\Scheduler\NodeManager\RemotingCommunicator\AgentConfig.cs:
- public string ListeningUri { get => ConfigurationManager.AppSettings["ListeningUri"] ?? "http://*:40000"; }
+ public string ListeningUri { get => ConfigurationManager.AppSettings["ListeningUri"] ?? "http://*:40010"; }
Cool! Please make a Pull Request for code change and associate it with the issue. @chezhang
acmnodemanager on Linux nodes may not start due to conflict with hpcnodemanager after adding HPC Pack cluster to ACM cluster