Huawei Cloud DLI Resource Management Skill

Overview

This Skill provides a full-lifecycle operations toolkit for Huawei Cloud DLI (Data Lake Insight) using the dli service of the hcloud CLI, covering ten capability groups:

1. Queue management — ListQueues, ShowQueue, CreateQueue, DeleteQueue, RunQueueAction (scale), queue plans, CIDR
2. Batch jobs (SQL/Spark) — CreateSqlJob, ListSqlJobs, ShowSqlJobStatus/Detail/Progress, CancelSqlJob, ExportSqlJobResult, PreviewSqlJobResult, Spark equivalents + ShowSparkJobLog
3. Flink stream jobs — CreateFlinkSqlJob, CreateFlinkJarJob, ListFlinkJobs, ShowFlinkJob, UpdateFlink*, BatchRunFlinkJobs, BatchStopFlinkJobs, BatchDeleteFlinkJobs, ShowFlinkMetric
4. Database management — ListDatabases, CreateDatabase, DeleteDatabase, UpdateDatabaseOwner, ListDatabaseUsers
5. Table management — ListTables, ShowTable, CreateTable, DeleteTable, UpdateTableOwner, ListPartitions, PreviewTable, table users/privileges
6. Enhanced connections — ListEnhancedConnections, Create/Update/DeleteEnhancedConnection, queue binding, CreateConnectivityTask/ShowConnectivityTask
7. Resource packages — ListJobResources, ShowJobResource, UploadJar/PythonFile/FileJobResources, DeleteJobResource, UpdateJobResourceOwner
8. Global variables — ListGlobalVariables, CreateGlobalVariable, UpdateGlobalVariable, DeleteGlobalVariable; Job templates — SQL/Flink SQL/Spark templates: list, create, update, delete
9. Privileges & patrol — authorization actions, queue/database/table users, elastic resource pools (CU), quota, agency, catalogs

Important Constraints

Prerequisites

1. hcloud CLI installed and authenticated with a valid AK/SK profile (installation guide and verification: see references/cli-installation-guide.md).
2. IAM permissions: DLI read permissions for query operations; DLI admin/write permissions for create/update/delete. See references/iam-policies.md.
3. Region: DLI resources are regional; pass --cli-region={region} on every command.
4. OBS paths (resource upload / result export): a pre-existing OBS bucket with readable paths, passed as OBS object URLs for --paths.[N].
5. Queue for SQL/Spark jobs: a running SQL or general queue is required to submit jobs; check ListQueues first.

Workflow

Scenario 1: Daily Resource Patrol (巡检)

ListQueues --with-charge-info -> queue inventory/CU; ListElasticResourcePools + ScaleRecords -> pool CU
ListSqlJobs / ListFlinkJobs -> running & failed jobs; ShowQuota -> CU used/remaining (>80% flag)
ShowDliAgency -> agency state; Output: structured patrol report (inventory + risks + cleanup proposals)

Scenario 2: Batch Job Troubleshooting (批作业故障排查)

ListSqlJobs --job-status=failed / ListSparkJobs --state=failed -> find failures
ShowSqlJobStatus/ShowSqlJobDetail -> reason & logs; (Spark) ShowSparkJobLog -> log tail
ShowSqlJobProgress -> running progress; Output: root-cause report (SQL/parameter/quota/connection)

Scenario 3: Flink Stream Job Operations (流作业运维)

ListFlinkJobs --job_type --status -> jobs; ShowFlinkJob -> detail/restart/checkpoint info
BatchRunFlinkJobs / BatchStopFlinkJobs (--trigger_savepoint) -> start/stop after double confirmation
ShowFlinkMetric -> CU metrics; Output: change snapshot of start/stop actions

Scenario 4: Metadata Governance (库表元数据治理)

ListDatabases / ListTables --with-detail -> inventory; ShowTable -> schema/location/owner
ListPartitions -> partition layout; UpdateTableOwner / UpdateDatabaseOwner -> ownership alignment
DeleteDatabase/DeleteTable -> ONLY with user double confirmation (cascade risk!); Output: change snapshot

Scenario 5: Cross-Source Connection (跨源连接)

ListEnhancedConnections -> inventory + status; CreateConnectivityTask + ShowConnectivityTask -> test result
AssociateQueueToEnhancedConnection / DisassociateQueueFromEnhancedConnection -> queue binding (if test fails)
UpdateEnhancedConnection --hosts.[N].ip/.name -> host whitelist (full overwrite); Output: connectivity risk report

Scenario 6: Idle Resource Cleanup (闲置资源清理)

ListQueues/ListQueuePlans/ListJobResources/ListGlobalVariables/ListSqlJobTemplates -> stale inventory
Report cleanup proposals to user -> wait for confirmation
After confirmation: DeleteQueue, DeleteJobResource, DeleteGlobalVariable, BatchDeleteSqlJobTemplates; Output: snapshot

Safety Gate (all scenarios)

Before any high-risk command (see Constraints), present the exact command, its scope, and irreversibility warning to the user and wait for an explicit confirmation keyword. Never batch-execute destructive commands without per-command confirmation. Mask AK/SK in outputs and logs.

Core Commands

All commands use hcloud dli <Operation>. --cli-region and --project_id are auto-filled by KooCLI (from profile) and omitted in parameter tables below. The full operation catalog with per-command parameter tables lives in references/dli-operation-catalog.md.

1. Queue Management

List Queues

hcloud dli ListQueues --cli-region={region} --queue_type={sql|general|all} [--tags={key=value}]

Show Queue

hcloud dli ShowQueue --cli-region={region} --queue_name={queue_name}

Create Queue

hcloud dli CreateQueue --cli-region={region} --queue_name={queue_name} --cu_count={cu_count} [--queue_type={sql|general}] [--engine={spark|flink|hetero}] [--charging_mode={prePaid|postPaid}]

Scale Queue (risk: double confirm for scale_in)

hcloud dli RunQueueAction --cli-region={region} --queue_name={queue_name} --action={scale_out|scale_in} [--cu_count={cu_count}] [--force=true]

Delete Queue (risk: double confirm)

hcloud dli DeleteQueue --cli-region={region} --queue_name={queue_name}

Queue Plans (定时扩缩容)

hcloud dli ListQueuePlans --cli-region={region} --queue_name={queue_name}
hcloud dli CreateQueuePlan --cli-region={region} --queue_name={queue_name} --plan_name={plan_name} \
  --start_hour={hour} --start_minute={minute} --target_cu={cu_count} [--activate={true|false}]
hcloud dli UpdateQueuePlan --cli-region={region} --queue_name={queue_name} --plan_id={plan_id} \
  --plan_name={plan_name} --start_hour={hour} --start_minute={minute} --target_cu={cu_count}
hcloud dli DeleteQueuePlan --cli-region={region} --queue_name={queue_name} --plan_id={plan_id}

2. Batch Jobs (SQL / Spark)

Submit SQL Job (cloud-side only; see Constraints)

hcloud dli CreateSqlJob --cli-region={region} --sql={sql_statement} \
  [--queue_name={queue_name}] [--currentdb={database}] [--sql_type={ddl|dcl|import|query|insert}] [--conf.1={key=value}]

List SQL Jobs

hcloud dli ListSqlJobs --cli-region={region} [--job-status={pending|running|success|failed|cancelled}] [--job-type={DDL|DCL|IMPORT|QUERY|INSERT}] [--db_name={database}] [--queue_name={queue_name}] [--page-size={n}] [--current-page={n}]

Show SQL Job Status / Detail / Progress

hcloud dli ShowSqlJobStatus --cli-region={region} --job_id={job_id}
hcloud dli ShowSqlJobDetail --cli-region={region} --job_id={job_id}
hcloud dli ShowSqlJobProgress --cli-region={region} --job_id={job_id}

Cancel SQL Job (risk: double confirm for running jobs)

hcloud dli CancelSqlJob --cli-region={region} --job_id={job_id}

Export / Preview SQL Result

hcloud dli ExportSqlJobResult --cli-region={region} --job_id={job_id} --data_path={obs_path} --data_type={csv|json|parquet} [--compress={gzip|none}]
hcloud dli PreviewSqlJobResult --cli-region={region} --job_id={job_id} [--queue-name={queue_name}]

Spark Jobs

hcloud dli CreateSparkJob --cli-region={region} --file={jar_resource} --className={main_class} \
  [--queue={queue_name}] [--name={job_name}] [--args.1={arg}] [--conf.1={key=value}] [--driver_memory={mem}] [--executor_memory={mem}] [--executor_cores={n}] [--num_executors={n}]
hcloud dli ListSparkJobs --cli-region={region} [--state={starting|running|success|failed|cancelled}] [--job_name={name}] [--queue_name={queue}] [--size={n}] [--from={offset}]
hcloud dli ShowSparkJob --cli-region={region} --batch_id={batch_id}
hcloud dli ShowSparkJobStatus --cli-region={region} --batch_id={batch_id}
hcloud dli ShowSparkJobLog --cli-region={region} --batch_id={batch_id} [--size={n}] [--type={all|driver|executor}]
hcloud dli CancelSparkJob --cli-region={region} --batch_id={batch_id}   # risk: double confirm

3. Flink Stream Jobs

hcloud dli CreateFlinkSqlJob --cli-region={region} --name={job_name} [--queue_name={queue}] [--cu_number={cu}] [--sql={flink_sql}] [--parallel_number={n}] [--smn_topic={topic}]
hcloud dli CreateFlinkJarJob --cli-region={region} --name={job_name} [--queue_name={queue}] [--entrypoint={jar_resource}] [--cu_number={cu}]
hcloud dli ListFlinkJobs --cli-region={region} [--job_type={flink_sql_job|flink_jar_job}] [--status={job_running|job_cancelled|job_failed|...}] [--name={name}] [--queue_name={queue}] [--limit={n}] [--offset={n}]
hcloud dli ShowFlinkJob --cli-region={region} --job_id={job_id}
hcloud dli UpdateFlinkSqlJob --cli-region={region} --job_id={job_id} [--cu_number={cu}] [--sql={flink_sql}] [--parallel_number={n}]
hcloud dli UpdateFlinkJarJob --cli-region={region} --job_id={job_id} [--cu_number={cu}] [--entrypoint={jar_resource}]
hcloud dli BatchRunFlinkJobs --cli-region={region} --job_ids.1={job_id} [--resume_savepoint={true|false}]      # risk: double confirm
hcloud dli BatchStopFlinkJobs --cli-region={region} --job_ids.1={job_id} [--trigger_savepoint={true|false}]     # risk: double confirm
hcloud dli BatchDeleteFlinkJobs --cli-region={region} --job_ids.1={job_id}                                      # risk: double confirm
hcloud dli ShowFlinkMetric --cli-region={region} --job_ids.1={job_id} [--job_ids.2={job_id_2}]

4. Database Management

hcloud dli ListDatabases --cli-region={region} [--keyword={kw}] [--limit={n}] [--offset={n}] [--tags={k=v}]
hcloud dli CreateDatabase --cli-region={region} --database_name={db_name} [--description={desc}]
hcloud dli DeleteDatabase --cli-region={region} --database_name={db_name} [--cascade={true|false}] [--async={true|false}]   # risk: double confirm
hcloud dli UpdateDatabaseOwner --cli-region={region} --database_name={db_name} --new_owner={user_name}
hcloud dli ListDatabaseUsers --cli-region={region} --database_name={db_name}

5. Table Management

hcloud dli ListTables --cli-region={region} --database_name={db_name} [--table-type={MANAGED|EXTERNAL|VIEW}] [--keyword={kw}] [--with-detail={true|false}] [--page-size={n}] [--current-page={n}]
hcloud dli ShowTable --cli-region={region} --database_name={db_name} --table_name={table_name}
hcloud dli CreateTable --cli-region={region} --database_name={db_name} --table_name={table_name} --data_location={DLI|OBS|VIEW} \
  --columns.1.column_name={col} --columns.1.type={TYPE} [--columns.2.column_name={col2}] [--columns.2.type={TYPE2}] \
  [--data_path={s3a://...}] [--data_type={Parquet|ORC|CSV|JSON|Carbon|Avro}] [--delimiter={c}]
hcloud dli DeleteTable --cli-region={region} --database_name={db_name} --table_name={table_name} [--async={true|false}]   # risk: double confirm
hcloud dli UpdateTableOwner --cli-region={region} --database_name={db_name} --table_name={table_name} --new_owner={user_name}
hcloud dli ListPartitions --cli-region={region} --database_name={db_name} --table_name={table_name} [--limit={n}] [--offset={n}]
hcloud dli PreviewTable --cli-region={region} --database_name={db_name} --table_name={table_name} [--mode={preview|sample}]
hcloud dli ListTableUsers --cli-region={region} --database_name={db_name} --table_name={table_name}
hcloud dli ListTablePrivileges --cli-region={region} --database_name={db_name} --table_name={table_name} --user_name={user}

6. Enhanced / Cross-Source Connections

hcloud dli ListEnhancedConnections --cli-region={region} [--name={name}] [--status={ACTIVE|FAILED}] [--limit={n}] [--offset={n}]
hcloud dli ShowEnhancedConnection --cli-region={region} --connection_id={connection_id}
hcloud dli CreateEnhancedConnection --cli-region={region} --name={conn_name} --dest_vpc_id={vpc_id} --dest_network_id={subnet_id} [--routetable_id={rt_id}]
hcloud dli UpdateEnhancedConnection --cli-region={region} --connection_id={connection_id} [--hosts.1.ip={ip}] [--hosts.1.name={host}]   # full overwrite of host list
hcloud dli DeleteEnhancedConnection --cli-region={region} --connection_id={connection_id}                                            # risk: double confirm
hcloud dli AssociateQueueToEnhancedConnection --cli-region={region} --connection_id={connection_id} --queues.1={queue_name}
hcloud dli DisassociateQueueFromEnhancedConnection --cli-region={region} --connection_id={connection_id}
hcloud dli CreateConnectivityTask --cli-region={region} --queue_name={queue_name} --address={ip_or_domain:port}
hcloud dli ShowConnectivityTask --cli-region={region} --queue_name={queue_name} --task_id={task_id}

7. Resource Packages

hcloud dli ListJobResources --cli-region={region} [--kind={jar|pyFile|file}] [--tags={k=v}]
hcloud dli ShowJobResource --cli-region={region} --resource_name={resource_name} [--group={group_name}]
hcloud dli UploadJarJobResources --cli-region={region} --group={group_name} --paths.1={obs://bucket/xx.jar}
hcloud dli UploadPythonFileJobResources --cli-region={region} --group={group_name} --paths.1={obs://bucket/xx.py}
hcloud dli UploadFileJobResources --cli-region={region} --group={group_name} --paths.1={obs://bucket/xx.conf}
hcloud dli DeleteJobResource --cli-region={region} --resource_name={resource_name} [--group={group_name}]                         # risk: double confirm
hcloud dli UpdateJobResourceOwner --cli-region={region} --group_name={group_name} --new_owner={user_name} [--resource_name={name}]

8. Global Variables

hcloud dli ListGlobalVariables --cli-region={region} [--limit={n}] [--offset={n}]
hcloud dli CreateGlobalVariable --cli-region={region} --var_name={var_name} --var_value={var_value} [--is_sensitive={true|false}]
hcloud dli UpdateGlobalVariable --cli-region={region} --var_name={var_name} --var_value={var_value}
hcloud dli DeleteGlobalVariable --cli-region={region} --var_name={var_name}   # risk: double confirm

9. Job Templates

hcloud dli ListSqlJobTemplates --cli-region={region} [--keyword={kw}]
hcloud dli CreateSqlJobTemplate --cli-region={region} --sql_name={template_name} --sql={sql_body} [--description={desc}] [--group={group}]
hcloud dli UpdateSqlJobTemplate --cli-region={region} --sql_id={template_id} [--sql_name={name}] [--sql={sql_body}] [--description={desc}]
hcloud dli BatchDeleteSqlJobTemplates --cli-region={region} --sql_ids.1={template_id}                                            # risk: double confirm
hcloud dli ListFlinkSqlJobTemplates --cli-region={region} [--name={name}] [--limit={n}] [--offset={n}]
hcloud dli CreateFlinkSqlJobTemplate --cli-region={region} --name={template_name} [--sql_body={flink_sql}] [--job_type={flink_sql_job|flink_jar_job}] [--desc={desc}]
hcloud dli UpdateFlinkSqlJobTemplate --cli-region={region} --template_id={template_id} [--name={name}] [--sql_body={flink_sql}] [--desc={desc}]
hcloud dli DeleteFlinkSqlJobTemplate --cli-region={region} --template_id={template_id}                                         # risk: double confirm
hcloud dli ListSparkJobTemplates --cli-region={region} [--type={1|2}] [--page-size={n}] [--current-page={n}]
hcloud dli CreateSparkJobTemplate --cli-region={region} --name={template_name} --type={1|2} [--description={desc}] [--group={group}] [--language={en|zh}]
hcloud dli UpdateSparkJobTemplate --cli-region={region} --template_id={template_id} --name={template_name} [--description={desc}] [--group={group}]

10. Privileges & Patrol

hcloud dli ListAuthorizationPrivileges --cli-region={region} --object={object}
hcloud dli RunAuthorizationAction --cli-region={region} --action={grant|revoke} [--user_name={user}] [--privileges.1.object={object}] [--privileges.1.privileges.1={privilege}]
hcloud dli RunDataAuthorizationAction --cli-region={region} --action={grant|revoke} --user_name={user}          # data-level auth
hcloud dli RegisterAuthorizedQueue --cli-region={region} --queue_name={queue_name} --user_name={user} --action={grant|revoke}
hcloud dli ShowEnhancedConnectionPrivilege --cli-region={region} --connection_id={connection_id}
hcloud dli ListElasticResourcePools --cli-region={region} [--name={name}] [--status={AVAILABLE|SCALING|CREATING|FAILED}] [--limit={n}]
hcloud dli UpdateElasticResourcePool --cli-region={region} --elastic_resource_pool_name={pool_name} [--min_cu={n}] [--max_cu={n}] [--description={desc}]   # risk: double confirm for shrinking
hcloud dli ListElasticResourcePoolScaleRecords --cli-region={region} --elastic_resource_pool_name={pool_name} [--status={SUCCESS|FAIL}] [--limit={n}]
hcloud dli ShowQuota --cli-region={region}
hcloud dli ShowDliAgency --cli-region={region}
hcloud dli ShowFlinkMetric --cli-region={region} --job_ids.1={job_id} [--job_ids.2={job_id_2}]
hcloud dli ListCatalogs --cli-region={region} [--limit={n}] [--offset={n}]
hcloud dli ShowCatalog --cli-region={region} --catalog_name={catalog_name}

Parameter Confirmation

Common conventions:

• Indexed/array parameters use .[N] suffixes (--paths.1=..., --job_ids.1=...); booleans accept true/false; time params use RFC3339.
• OBS paths use obs://bucket/key, except CreateTable --data_path which requires s3a://.

KooCLI Command Format Standard

Generic command template (do not execute as-is — replace <service>, <Operation> and parameters):

hcloud <service> <Operation> --cli-region=<region> [--key=value ...]

Reference Documents

• references/iam-policies.md — Least-privilege IAM policies for DLI operations
• references/cli-installation-guide.md — hcloud CLI installation & AK/SK configuration
• references/dli-operation-catalog.md — Full DLI operation catalog with parameter tables
• references/common-workflows.md — Scenario-based workflow recipes
• references/dataflow-diagram.md — Mermaid data flow diagram
• references/verification-method.md — Verification method details
• references/acceptance-criteria.md — Acceptance criteria

Best Practices

• Confirm the queue exists (ListQueues) before submitting jobs; never submit to an unknown queue.
• Keep output small: paginate with --limit/--offset or --page-size/--current-page; use status filters (--job-status, --status) for targeted troubleshooting.
• Merge multiple List* outputs into one structured markdown report; mask AK/SK and sensitive variable values (--is_sensitive variables render masked).
• Always present high-risk commands to the user with expected impact and wait for explicit confirmation (see Constraints).
• When a connectivity check fails, verify queue binding (AssociateQueueToEnhancedConnection), security group, and VPC route before recreating the connection.
• CU patrol: compare ShowQuota used vs quota; flag > 80%; use ListElasticResourcePoolScaleRecords before suggesting scaling changes.

Verification Method

hcloud dli ListQueues --cli-region={region} --queue_type=all        # read smoke test
hcloud dli ListDatabases --cli-region={region} --limit=5            # metadata read smoke test
hcloud dli ShowQuota --cli-region={region}                          # quota read smoke test

See references/verification-method.md for the detailed verification matrix.