[🐛 BUG]: AMQP - declare with a readonly user

[denis-korolev]

No duplicates 🥲.

• I have searched for a similar issue in our bug tracker and didn't find any solutions.

What happened?

I try run RR with JOBS on AMQP Rabbit. But my user have readonly permittion.
So when RR starting, it try to create exhange/quee.
But RR is down when trying to create an exchange and quee.

In my situation I can only ACK.
I dont have permittion to create new quee.

So i want just read messages and save them in my inbox and run jobs later by several method.

Can you fix it tell how to avoid it.

Version (rr --version)

rr version 2025.1.2 (build time: 2025-06-14T22:23:27+0000, go1.24.4), OS: linux, arch: amd64

How to reproduce the issue?

rr version 2025.1.2 (build time: 2025-06-14T22:23:27+0000, go1.24.4), OS: linux, arch: amd64

ampq user with read only permittions

Relevant log output

[coderabbitai]

📝 CodeRabbit Plan Mode

Generate an implementation plan and prompts that you can use with your favorite coding agent.

• Create Plan

Examples

• Example 1
• Example 2

---

🔗 Similar Issues

Related Issues

• [🐛 BUG]: Redialer does not resume pipeline after delivery channel was closed by RabbitMQ #2091
• [🐛 BUG]: Panic is triggered when NoFreeWorkers #2125
• [🐛 BUG]: panic: runtime error: invalid memory address or nil pointer dereference in 2025.1.0 #2172

👤 Suggested Assignees

• algirdasci
• cv65kr
• Orrison

---

🧪 Issue enrichment is currently in open beta.

You can configure auto-planning by selecting labels in the issue_enrichment configuration.

To disable automatic issue enrichment, add the following to your .coderabbit.yaml:

issue_enrichment:
  auto_enrich:
    enabled: false

💬 Have feedback or questions? Drop into our discord or schedule a call!

[rustatian]

Hey @denis-korolev 👋🏻
Please, add an example of how to reproduce this problem. For example, a configured Dockerfile with a user w/o permissions.

[denis-korolev]

I have some errors from Rabbit admin.

Here RR trying to get acces to exchange have no permittion

handle_serve_command: Function call error:
    serve error from the plugin *jobs.Plugin stopping execution, error: jobs_plugin_serve: new_amqp_consumer:
    jobs_plugin_amqp_init: Exception (403) Reason: "ACCESS_REFUSED - configure access to exchange 'TEST_EXCHANGE' in vhost 'TEST' refused for user 'USER'"

Here RR trying to change queue params.

2025-12-18 16:05:32.061000+07:00 \[info\] <0.168786.0> connection 10.13.53.1:42962 -> 10.49.37.30:5675: user '_USER' authenticated and granted access to vhost 'TEST'
2025-12-18 16:05:32.068000+07:00 \[error\] <0.168795.0> Channel error on connection <0.168786.0> (10.13.53.1:42962 -> 10.49.37.30:5675, vhost: 'TEST', user: 'USER'), channel 1:
2025-12-18 16:05:32.068000+07:00 \[error\] <0.168795.0> operation exchange.declare caused a channel exception precondition\_failed: inequivalent arg 'type' for exchange 'OUT' in vhost 'TEST': received 'direct' but current is 'fanout'
2025-12-18 16:05:32.078000+07:00 \[warning\] <0.168786.0> closing AMQP connection <0.168786.0> (10.13.53.1:42962 -> 10.49.37.30:5675, vhost: 'TEST', user: 'CARGO\_USER', duration: '36ms'):
2025-12-18 16:05:32.078000+07:00 \[warning\] <0.168786.0> client unexpectedly closed TCP connection

version: '3'
rpc:
  listen: 'tcp://127.0.0.1:6001'
server:
  command: 'php entry/index.php'
  relay: pipes
kv:
  test:
    driver: memory
    config:
      interval: 10
http:
  address: '0.0.0.0:8088'
  middleware:
    - gzip
    - static
  static:
    dir: ../public
    forbid:
      - .php
      - .htaccess
  pool:
    debug: ${HTTP_DEBUG_MODE}
    num_workers: ${NUM_HTTP_WORKERS}
    supervisor:
      ttl: 600s
      max_worker_memory: 100
      exec_ttl: 60s
#rabbit
amqp:
  addr: ${RABBIT}

jobs:
  consume: [ "amqp_pipline_name"]
  pipelines:
    amqp_pipline_name:
      prefetch: 1
      driver: amqp
      config:
        queue: somename
        durable: true
        exchange: somename
        exchange_type: fanout
        exchange_durable: true
  pool:
    num_workers: ${NUM_JOBS_WORKERS}
    max_worker_memory: 100

metrics:
  address: '127.0.0.1:2112'

[rustatian]

@denis-korolev

2025-12-18 16:05:32.068000+07:00 \[error\] <0.168795.0> operation exchange.declare caused a channel exception precondition\_failed: inequivalent arg 'type' for exchange 'OUT' in vhost 'TEST': received 'direct' but current is 'fanout'

RR is not trying to change something; this is a check that your configuration is in sync with what was already declared in your RabbitMQ broker. I'm also not sure if the provided configuration is in sync with that log output. Since you have a named exchange - somename, but in the log I see - OUT. You need to sync the configuration and the already existing exchange.

And could you please attach a Dockerfile with a configured RabbitMQ or a script with a rabbitmqctl with the user you created and permissions for that user.

[denis-korolev]

Sorry, i send you new correct yaml file.
At first time it was that (exchange_type:direct)

version: '3'
rpc:
  listen: 'tcp://127.0.0.1:6001'
server:
  command: 'php entry/index.php'
  relay: pipes
kv:
  test:
    driver: memory
    config:
      interval: 10
http:
  address: '0.0.0.0:8088'
  middleware:
    - gzip
    - static
  static:
    dir: ../public
    forbid:
      - .php
      - .htaccess
  pool:
    debug: ${HTTP_DEBUG_MODE}
    num_workers: ${NUM_HTTP_WORKERS}
    supervisor:
      ttl: 600s
      max_worker_memory: 100
      exec_ttl: 60s
#rabbit
amqp:
  addr: ${RABBIT}

jobs:
  consume: [ "amqp_pipline_name"]
  pipelines:
    amqp_pipline_name:
      prefetch: 1
      driver: amqp
      config:
        queue: somename
        durable: true
        exchange: somename
        exchange_type: direct
        exchange_durable: true
  pool:
    num_workers: ${NUM_JOBS_WORKERS}
    max_worker_memory: 100

metrics:
  address: '127.0.0.1:2112'

And here
for exchange 'OUT' must be for exchange 'somename'

I ask admins to send me configurations and more logs.

[denis-korolev]

!!! UPDATE FROM ADMIN!!!

Correct config

#rabbit
amqp:
  addr: amqp://CARGO_USER:........@20.20.20.20:5675/TEST

jobs:
  consume: [ "amqp_pipline_name"]
  pipelines:
    amqp_pipline_name:
      prefetch: 1
      driver: amqp
      config:
        queue: CARGO_ECN_EVENTS
        durable: true
        exchange: CARGO_OUT
        exchange_type: fanout
        exchange_durable: true
  pool:
    num_workers: ${NUM_JOBS_WORKERS}
    max_worker_memory: 100

metrics:
  address: '127.0.0.1:2112'

Correct settings in Rabbit

Setting when We Have ERRORS

AND LOGS

2025-12-18 16:55:26.882000+07:00 [info] <0.172052.0> connection 10.13.53.1:40158 -> 10.49.37.30:5675: user 'CARGO_USER' authenticated and granted access to vhost 'TEST'
2025-12-18 16:55:26.882000+07:00 [error] <0.172061.0> Channel error on connection <0.172052.0> (10.13.53.1:40158 -> 10.49.37.30:5675, vhost: 'TEST', user: 'CARGO_USER'), channel 1:
2025-12-18 16:55:26.882000+07:00 [error] <0.172061.0> operation exchange.declare caused a channel exception access_refused: configure access to exchange 'CARGO_OUT' in vhost 'TEST' refused for user 'CARGO_USER'
2025-12-18 16:55:26.908000+07:00 [warning] <0.172052.0> closing AMQP connection <0.172052.0> (10.13.53.1:40158 -> 10.49.37.30:5675, vhost: 'TEST', user: 'CARGO_USER', duration: '26ms'):
2025-12-18 16:55:26.908000+07:00 [warning] <0.172052.0> client unexpectedly closed TCP connection

[rauanmayemir]

Correct way here would be adding an optional config.skip_declare: true, so that amqpjobs skips declare call altogether.

[rustatian]

I don't think so. skip_declare is a confusing configuration option, since it does not relate to an entity. Like skip declaration of exchange? Or queue?
I'm thinking about combining options for the exchange/queue/probably listener into a separate yaml objects. For example, amqp.config.exchange.declare = false, or amqp.config.queue.qurable = true