Can't get metrics despite following instructions

[khachou]

I followed all the steps but can't get the metrics.Any help @danihodovic

[danihodovic]

:shrug:

[hammady]

No seriously, a troubleshooting guide will be helpful. I'm using the latest helm chart. For me, I made sure the celery workers send task events:

celery -A <app> control enable_events
Prod Environment
Debug: False
Using prod DB.
->  celery@app-8a0f19073c2f58c1-74b98d5d65-tzhk5: OK
        task events already enabled
->  celery@app-8a0f19073c2f58c1-74b98d5d65-lqbmw: OK
        task events already enabled
->  celery@app-8a0f19073c2f58c1-74b98d5d65-97kw7: OK
        task events already enabled
->  celery@app-8a0f19073c2f58c1-74b98d5d65-9hj9d: OK
        task events already enabled

The broker is redis. The exporter can connect to the broker. Tasks are run. When running the below inside the exporter container:

curl localhost:9808/metrics

All metrics are empty (not even 0) except the queue size returning 0:

# HELP celery_task_sent_total Sent when a task message is published.
# TYPE celery_task_sent_total counter
# HELP celery_task_received_total Sent when the worker receives a task.
# TYPE celery_task_received_total counter
# HELP celery_task_started_total Sent just before the worker executes the task.
# TYPE celery_task_started_total counter
# HELP celery_task_succeeded_total Sent if the task executed successfully.
# TYPE celery_task_succeeded_total counter
# HELP celery_task_failed_total Sent if the execution of the task failed.
# TYPE celery_task_failed_total counter
# HELP celery_task_rejected_total The task was rejected by the worker, possibly to be re-queued or moved to a dead letter queue.
# TYPE celery_task_rejected_total counter
# HELP celery_task_revoked_total Sent if the task has been revoked.
# TYPE celery_task_revoked_total counter
# HELP celery_task_retried_total Sent if the task failed, but will be retried in the future.
# TYPE celery_task_retried_total counter
# HELP celery_worker_up Indicates if a worker has recently sent a heartbeat.
# TYPE celery_worker_up gauge
# HELP celery_worker_tasks_active The number of tasks the worker is currently processing
# TYPE celery_worker_tasks_active gauge
# HELP celery_task_runtime Histogram of task runtime measurements.
# TYPE celery_task_runtime histogram
# HELP celery_queue_length The number of message in broker queue.
# TYPE celery_queue_length gauge
celery_queue_length{queue_name="celery"} 0.0
# HELP celery_active_consumer_count The number of active consumer in broker queue.
# TYPE celery_active_consumer_count gauge

Container logs:

2023-03-09 04:27:49.285 | INFO     | src.exporter:run:208 - Setting celery accept_content ['json', 'pickle']                             │
2023-03-09 04:27:49.489 | INFO     | src.http_server:start_http_server:66 - Started celery-exporter at port='9808'

It is unclear to me which settings should be put in the exporter other than the below:

│   - env:                                                                                                                                 │
│     - name: CE_BROKER_URL                                                                                                                │
│       value: redis://*****:6379/0                                                       │
│     - name: CE_ACCEPT_CONTENT                                                                                                            │
│       value: json,pickle                                                                                                                 

[hammady]

@danihodovic It was caused by the serializer for events (tasks & results) being pickle. Once changed to json, it started reporting right away. Is this a bug/limitation of the exporter or a just missing documentation?

[abiwill]

@hammady can you please share the changes. I am facing similar issue with the latest release of exporter.