Skip to main content

Troubleshooting — RabbitMQ

The cluster stays "Creating" or switches to "Error"​

Cause: provisioning is in progress, or the platform could not apply the cluster's creation or its last change (for example due to a lack of available resources). In the second case, the status switches to Error; after a change, the cluster may keep running with its previous configuration.

Solution:

  1. Wait a few minutes: the detail page and the list refresh automatically.
  2. If the status stays Creating for an abnormally long time or switches to Error, contact support, giving the project name, the cluster name and its identifier (displayed under the cluster name, with a copy button).

A vhost or a user is refused​

Cause: its name does not follow the naming rules. Vhost names and usernames only accept lowercase letters, digits and hyphens, and must start with a letter and end with a letter or a digit: my_vhost, App.User or /prod are refused.

Solution: choose a compliant name, for example prod or app-user. See Concepts.

"Storage quota exceeded for this project" in the wizard​

Cause: the disk size multiplied by the number of replicas exceeds the remaining storage of the project quota.

Solution:

  1. Reduce the Disk size (GB) or the Number of replicas.
  2. If needed, free up storage in the project or have the project quota increased.

"A cluster with this name already exists"​

Cause: a RabbitMQ cluster in the project already has this name.

Solution: go back to the General step and choose another Cluster Name.

The External zone shows "Disabled" or "Waiting for allocation…"​

Cause: External Access is disabled, or the public address has not been assigned yet.

Solution:

  1. "Disabled": enable external access (see Configure external access). From the project, it is not needed: use the host in the Internal zone.
  2. If external access is enabled, wait and then reload the page.

AMQP connection refused (ACCESS_REFUSED)​

Cause: incorrect credentials, or the user has no right on the requested vhost.

Solution:

  1. In the Users table (VHosts column), check that the user has a right on the vhost used by the client.
  2. If needed, add the access with Manage Access.
  3. If the password has been lost or is in doubt, generate a new one with Change Password and update the client.
  4. Check that the client specifies the right vhost (exact name, case-sensitive).

Unable to connect (timeout, connection refused)​

Cause: external access disabled, wrong address or wrong port, or network filtering on the client side.

Solution:

  1. Check the address used in the Connection section: External zone from the Internet, Internal zone from the project.
  2. Use port 5672.
  3. Test whether the port is open from the client machine: 
    nc -zv <host> 5672
  4. Check that your local network or firewall allows outgoing connections to this port.

Publishing blocked (flow control, memory or disk alarm)​

Cause: RabbitMQ blocks publishing when it reaches its memory threshold (high watermark) or when disk space is insufficient, to protect the broker. Clients then receive a connection.blocked notification.

Solution:

  1. On the application side, check that consumers keep up with producers and purge the queues that accumulate unconsumed messages.
  2. Increase the Disk size (GB) from Edit if the alarm concerns the disk (see Change a cluster's configuration).
  3. The preset (memory) cannot be changed after creation: create a cluster with a larger preset, or contact support.

Unrouted messages​

Cause: the producer publishes to an exchange without a matching binding (wrong exchange type, incorrect routing key, missing binding). The message is then dropped.

Solution:

  1. Check the exchange name and the routing key in the producer code.
  2. Check that the consumer declares the binding between the queue and the exchange.
  3. Publish with the mandatory flag to be notified of unrouted messages, or declare an alternate exchange to capture them.

Deleting the cluster fails​

Cause: a conflict prevents deletion ("Cannot delete this cluster (conflict).") or the service is temporarily unavailable.

Solution: try again a few minutes later. If the error persists, contact support.