Improvement/zenko 4941

.github/scripts/end2end/configs/zenko.yaml

@@ -108,6 +108,9 @@ spec:
         command-timeout: "60s"
         pending-job-poll-after-age: "10s"
         pending-job-poll-check-interval: "10s"
+  scuba:
+    logging:
+      logLevel: debug
   ingress:
     workloadPlaneClass: 'nginx'
     controlPlaneClass: 'nginx'

tests/ctst/common/common.ts

@@ -352,9 +352,9 @@ When('the user tries to perform the current S3 action on the bucket {int} times
                 this.addToSaved('copyObject', `objectrepeatcopy-${Utils.randomString()}`);
             }
             await runActionAgainstBucket(this, this.getSaved<ActionPermissionsType>('currentAction').action);
-            if (this.getResult().err) {
-                // stop at any error, the error will be evaluated in a separated step
-                return;
+            if (this.getResult().err && this.getResult().retryable?.throttling !== true) {
+                this.logger.debug('Error during repeated action', { error: this.getResult().err });
+                break;
             }
             await Utils.sleep(delay);
         }

tests/ctst/common/hooks.ts

@@ -54,7 +54,13 @@ After(async function (this: Zenko, results) {
     );
 });
 
-After({ tags: '@Quotas' }, async function () {
+After({ tags: '@Quotas' }, async function (this: Zenko, results) {
+    if (results.result?.status === 'FAILED') {
+        this.logger.warn('quota was not cleaned for test', {
+            bucket: this.getSaved<string>('bucketName'),
+        });
+        return;
+    }
     await teardownQuotaScenarios(this as Zenko);
 });
 

tests/ctst/features/quotas/CountItems.feature → tests/ctst/features/CountItems/CountItems.feature

@@ -5,6 +5,7 @@ Feature: CountItems measures the utilization metrics
 @PreMerge
 @CronJob
 @CountItems
+@Utilization
 Scenario Outline: Countitems runs without error and compute utilization metrics
     Given an existing bucket "" "without" versioning, "without" ObjectLock "without" retention mode
     And an object "" that "exists"

tests/ctst/features/quotas/Quotas.feature

@@ -68,6 +68,7 @@ Feature: Quota Management for APIs
     @CronJob
     @DataDeletion
     @NonVersioned
+    @Utilization
     Scenario Outline: Quotas are affected by deletion operations
         Given an action "DeleteObject"
         And a permission to perform the "PutObject" action
@@ -80,11 +81,11 @@ Feature: Quota Management for APIs
         And a <userType> type
         And an environment setup for the API
         And an "existing" IAM Policy that "applies" with "ALLOW" effect for the current API
-        When I wait 3 seconds
+        When I wait 6 seconds
         And I PUT an object with size <uploadSize>
         Then the API should "fail" with "QuotaExceeded"
         When i delete object "obj-1"
-        And I wait 3 seconds
+        And I wait 6 seconds
         And I PUT an object with size <uploadSize>
         Then the API should "succeed" with ""
 
@@ -96,3 +97,78 @@ Feature: Quota Management for APIs
             | 100        | 200         | 0            | IAM_USER |
             | 100        | 0           | 200          | IAM_USER |
             | 100        | 200         | 200          | IAM_USER |
+
+    @2.6.0
+    @PreMerge
+    @Quotas
+    @CronJob
+    @DataDeletion
+    @NonVersioned
+    @Utilization
+    Scenario Outline: Quotas are affected by deletion operations between count items runs
+        Given an action "DeleteObject"
+        And a permission to perform the "PutObject" action
+        And a STORAGE_MANAGER type
+        And a bucket quota set to 1000 B
+        And an account quota set to 1000 B
+        And an upload size of 1000 B for the object "obj-1"
+        And a bucket quota set to <bucketQuota> B
+        And an account quota set to <accountQuota> B
+        And a <userType> type
+        And an environment setup for the API
+        And an "existing" IAM Policy that "applies" with "ALLOW" effect for the current API
+        When I wait 3 seconds
+        And I PUT an object with size <uploadSize>
+        Then the API should "fail" with "QuotaExceeded"
+        When the "count-items" cronjobs completes without error
+        # Wait for inflights to be read by SCUBA
+        When I wait 3 seconds
+        # At this point if negative inflights are not supported, write should
+        # not be possible, as the previous inflights are now part of the current
+        # metrics.
+        And i delete object "obj-1"
+        # Wait for inflights to be read by SCUBA
+        And I wait 3 seconds
+        And I PUT an object with size <uploadSize>
+        Then the API should "succeed" with ""
+
+        Examples:
+            | uploadSize | bucketQuota | accountQuota | userType |
+            | 100        | 200         | 0            | ACCOUNT  |
+
+    @2.6.0
+    @PreMerge
+    @Quotas
+    @CronJob
+    @DataDeletion
+    @NonVersioned
+    Scenario Outline: Negative inflights do not allow to bypass the quota
+        Given an action "DeleteObject"
+        And a permission to perform the "PutObject" action
+        And a STORAGE_MANAGER type
+        And a bucket quota set to <bucketQuota> B
+        And an account quota set to <accountQuota> B
+        And an upload size of <uploadSize> B for the object "obj-1"
+        And a <userType> type
+        And an environment setup for the API
+        And an "existing" IAM Policy that "applies" with "ALLOW" effect for the current API
+        When I wait 3 seconds
+        And I PUT an object with size <uploadSize>
+        Then the API should "fail" with "QuotaExceeded"
+        # The inflights are now part of the current metrics
+        When the "count-items" cronjobs completes without error
+        # Wait for inflights to be read by SCUBA
+        When I wait 3 seconds
+        # At this point if negative inflights are not supported, write should
+        # not be possible, as the previous inflights are now part of the current
+        # metrics.
+        # Delete 200 B, inflights are now -200
+        And i delete object "obj-1"
+        # Wait for inflights to be read by SCUBA
+        And I wait 6 seconds
+        And I PUT an object with size 300
+        Then the API should "fail" with "QuotaExceeded"
+
+        Examples:
+            | uploadSize | bucketQuota | accountQuota | userType |
+            | 200        | 200         | 0            | ACCOUNT  |

tests/ctst/package.json

@@ -24,7 +24,7 @@
     "@aws-sdk/client-s3": "^3.583.0",
     "@aws-sdk/client-sts": "^3.583.0",
     "@eslint/compat": "^1.1.1",
-    "cli-testing": "github:scality/cli-testing.git#1.2.3",
+    "cli-testing": "github:scality/cli-testing.git#1.2.4",
     "eslint": "^9.9.1",
     "eslint-config-scality": "scality/Guidelines#8.3.0",
     "typescript-eslint": "^8.4.0"

tests/ctst/steps/quotas/quotas.ts

@@ -6,6 +6,7 @@ import { Scality, Command, Utils, AWSCredentials, Constants, Identity, IdentityE
 import { createJobAndWaitForCompletion } from '../utils/kubernetes';
 import { createBucketWithConfiguration, putObject } from '../utils/utils';
 import { hashStringAndKeepFirst20Characters } from 'common/utils';
+import assert from 'assert';
 
 export async function prepareQuotaScenarios(world: Zenko, scenarioConfiguration: ITestCaseHookParameter) {
     /**
@@ -136,6 +137,16 @@ Given('a bucket quota set to {int} B', async function (this: Zenko, quota: numbe
         result,
     });
 
+    // Ensure the quota is set
+    const resultGet: Command = await Scality.getBucketQuota(
+        this.parameters,
+        this.getCommandParameters());
+    this.logger.debug('GetBucketQuota result', {
+        resultGet,
+    });
+
+    assert(resultGet.stdout.includes(`${quota}`));
+
     if (result.err) {
         throw new Error(result.err);
     }
@@ -158,6 +169,9 @@ Given('an account quota set to {int} B', async function (this: Zenko, quota: num
         result,
     });
 
+    // Ensure the quota is set
+    assert(JSON.parse(result.stdout).quota === quota);
+
     if (result.err) {
         throw new Error(result.err);
     }

tests/ctst/steps/utils/kubernetes.ts

@@ -1,3 +1,5 @@
+import fs from 'fs';
+import * as path from 'path';
 import { KubernetesHelper, Utils } from 'cli-testing';
 import Zenko from 'world/Zenko';
 import {
@@ -71,12 +73,42 @@ export function createKubeCustomObjectClient(world: Zenko): CustomObjectsApi {
     return KubernetesHelper.customObject;
 }
 
-export async function createJobAndWaitForCompletion(world: Zenko, jobName: string, customMetadata?: string) {
+export async function createJobAndWaitForCompletion(
+    world: Zenko,
+    jobName: string,
+    customMetadata?: string
+) {
     const batchClient = createKubeBatchClient(world);
     const watchClient = createKubeWatchClient(world);
+
+    const lockFilePath = path.join('/tmp', `${jobName}.lock`);
+
+    let lockAquired = false;
+    let tries = 600;
+    while (!lockAquired && tries > 0) {
+        try {
+            fs.writeFileSync(lockFilePath, 'lock', {
+                flag: 'wx',
+            });
+            lockAquired = true;
+        } catch {
+            world.logger.debug(`Failed to acquire lock for job: ${jobName}`, {
+                tries,
+            });            
+        }
+        tries--;
+        if (!lockAquired) {
+            await Utils.sleep(1000);
+        }
+    }
+
     try {
+        world.logger.debug(`Acquired lock for job: ${jobName}`);
+
+        // Read the cron job and prepare the job spec
         const cronJob = await batchClient.readNamespacedCronJob(jobName, 'default');
         const cronJobSpec = cronJob.body.spec?.jobTemplate.spec;
+
         const job = new V1Job();
         const metadata = new V1ObjectMeta();
         job.apiVersion = 'batch/v1';
@@ -87,50 +119,53 @@ export async function createJobAndWaitForCompletion(world: Zenko, jobName: strin
             'cronjob.kubernetes.io/instantiate': 'ctst',
         };
         if (customMetadata) {
-            metadata.annotations = {
-                custom: customMetadata,
-            };
+            metadata.annotations.custom = customMetadata;
         }
         job.metadata = metadata;
 
+        // Create the job
         const response = await batchClient.createNamespacedJob('default', job);
-        world.logger.debug('job created', {
-            job: response.body.metadata,
-        });
+        world.logger.debug('Job created', { job: response.body.metadata });
 
         const expectedJobName = response.body.metadata?.name;
 
+        // Watch for job completion
         await new Promise<void>((resolve, reject) => {
             void watchClient.watch(
                 '/apis/batch/v1/namespaces/default/jobs',
                 {},
                 (type: string, apiObj, watchObj) => {
-                    if (job.metadata?.name && expectedJobName &&
-                        (watchObj.object?.metadata?.name as string)?.startsWith?.(expectedJobName)) {
+                    if (
+                        expectedJobName &&
+                        (watchObj.object?.metadata?.name as string)?.startsWith?.(expectedJobName)
+                    ) {
                         if (watchObj.object?.status?.succeeded) {
-                            world.logger.debug('job succeeded', {
-                                job: job.metadata,
-                            });
+                            world.logger.debug('Job succeeded', { job: job.metadata });
                             resolve();
                         } else if (watchObj.object?.status?.failed) {
-                            world.logger.debug('job failed', {
+                            world.logger.debug('Job failed', {
                                 job: job.metadata,
                                 object: watchObj.object,
                             });
-                            reject(new Error('job failed'));
+                            reject(new Error('Job failed'));
                         }
                     }
-                }, reject);
+                },
+                reject
+            );
         });
     } catch (err: unknown) {
-        world.logger.error('error creating job', {
+        world.logger.error('Error creating or waiting for job completion', {
             jobName,
             err,
         });
         throw err;
+    } finally {
+        fs.unlinkSync(lockFilePath);
     }
 }
 
+
 export async function waitForZenkoToStabilize(
     world: Zenko, needsReconciliation = false, timeout = 15 * 60 * 1000, namespace = 'default') {
     // ZKOP pulls the overlay configuration from Pensieve every 5 seconds