Pass instanceId to the mongodb metadata backend

instanceId will be used when generating versionIds.
This will ensure uniqueness of the versionID across
all Cloudserver pods in the cluster.

The instanceId is a combination of the short hash
contained in the name of the pods, which is unique
across all pods within the same deployment, and a
prefix for differenciating the different types of
deployments (internal vs external).

see. https://scality.atlassian.net/wiki/spaces/OS/pages/3225583692/VersionID+Collisions+in+Zenko

Issue: CLDSRV-665

Author: Kerkesni

lib/Config.js

@@ -910,6 +910,29 @@ class Config extends EventEmitter {
             this.replicationGroupId = 'RG001';
         }
 
+        if (process.env.HOSTNAME) {
+            // If running in Kubernetes, the pod name is set in the HOSTNAME env var
+            // and is of the form: cloudserver-7869f99955-nr984
+            // where:
+            // - cloudserver is the name of the pod
+            // - 7869f99955 is a long hash that guarantees uniqueness within the namespace
+            // - nr984 is a short hash that is unique only within the current ReplicaSet/Deployment
+            // We use the last part of the pod name as the instanceId, prefixed with 'i' for internal
+            // or 'e' for external cloudserver.
+            // This allows having a unique identifier that will be used when generating versionIds
+            const instanceType = process.env.HOSTNAME?.includes('internal') ? 'i' : 'e';
+            const parts = process.env.HOSTNAME.split('-');
+            if (parts.length > 2) {
+                const shortId = parts[parts.length - 1];
+                this.instanceId = instanceType + shortId;
+            } else {
+                // fallback: generate a random string if format is unexpected
+                this.instanceId = uuidv4().replace(/-/g, '').slice(0, 6);
+            }
+        } else {
+            this.instanceId = uuidv4().replace(/-/g, '').slice(0, 6);
+        }
+
         this.replicationEndpoints = [];
         if (config.replicationEndpoints) {
             const { replicationEndpoints } = config;

lib/metadata/wrapper.js

@@ -30,6 +30,7 @@ if (clientName === 'mem') {
     params = {
         mongodb: config.mongodb,
         replicationGroupId: config.replicationGroupId,
+        instanceId: config.instanceId,
         config,
     };
 } else if (clientName === 'cdmi') {

tests/unit/Config.js

@@ -810,4 +810,42 @@ describe('Config', () => {
             assert.deepEqual(config.internalListenOn, [{ ip: '127.0.0.1', port: 9000 }]);
         });
     });
+
+    describe('instanceId', () => {
+        let previousHostname;
+
+        beforeEach(() => {
+            previousHostname = process.env.HOSTNAME;
+        });
+
+        afterEach(() => {
+            if (previousHostname) {
+                process.env.HOSTNAME = previousHostname;
+            } else {
+                delete process.env.HOSTNAME;
+            }
+        });
+
+        it('should generate a new instanceId for external cloudserver', () => {
+            process.env.HOSTNAME = 'connector-cloudserver-69958b4697-bs5pn';
+            const config = new ConfigObject();
+            assert.strictEqual(config.instanceId, 'ebs5pn');
+        });
+
+        it('should generate a new instanceId for internal cloudserver', () => {
+            process.env.HOSTNAME = 'internal-cloudserver-54dc76b796-48m2x';
+            const config = new ConfigObject();
+            assert.strictEqual(config.instanceId, 'i48m2x');
+        });
+        it('should generate a random instanceId if HOSTNAME has an unexpected format', () => {
+            process.env.HOSTNAME = 'cldsrv1';
+            const config = new ConfigObject();
+            assert.strictEqual(config.instanceId.length, 6);
+        });
+        it('should generate a random instanceId if HOSTNAME is not set', () => {
+            delete process.env.HOSTNAME;
+            const config = new ConfigObject();
+            assert.strictEqual(config.instanceId.length, 6);
+        });
+    });
 });