fix(cluster): use instance name as node id

The IDs used to be generated on each restart, which leads to confusing
logs and wrong removal behavior. The instance name should be unique
anyway, so we might as well use it for the clustering.

src/cluster/cluster.service.ts

@@ -12,6 +12,7 @@ import * as os from 'os';
 import { NetworkInterfaceInfo } from 'os';
 import { ConfigService } from '../config/config.service';
 import { ClusterConfig } from './cluster.config';
+import { makeId } from '../util/id';
 
 let mdns;
 try {
@@ -33,6 +34,7 @@ export class ClusterService extends Democracy
 
   constructor(configService: ConfigService) {
     const config = configService.get('cluster');
+    const globalConfig = configService.get('global');
 
     const networkInterfaces = _.flatMap<NetworkInterfaceInfo>(
       config.networkInterface
@@ -43,6 +45,7 @@ export class ClusterService extends Democracy
       address => address.internal === false && address.family === 'IPv4'
     ).address;
     super({
+      id: makeId(globalConfig.instanceName),
       source: `${ip}:${config.port}`,
       peers: Array.from(config.peerAddresses),
       timeout: config.timeout * 1000,