I used the template: https://railway.com/deploy/rabbitmq to deploy rabbimq. After that, I went to the control panel and found that some settings were outdated for version 4.2.1-management. Monitoring was also disabled. Inside the template, the launch is performed using this command:

/bin/sh -c "CONFIG_PATH=/etc; SYSTEM_FILE=hosts; echo 127.0.0.1 rabbitmq >> ${CONFIG_PATH}/${SYSTEM_FILE} && echo management.tcp.ip = :: >> /etc/rabbitmq/conf.d/10-defaults.conf && docker-entrypoint.sh rabbitmq-server"

Is it possible to use newer parameters for launching and still enable monitoring?

I'm not too sure what you mean about settings being outdated, because this uses the default Docker configuration. Also not too sure what you mean by monitoring. Could you clarify?

Of course, I attached screenshots of what I mean. After launching and going to the control panel, I find this message about outdated settings.

Also, the monitoring I meant is in the second screenshot, “Message rates.”

I don't see outdated settings. Could that be with your usage?
Message rates appears to be from the monitoring plugin, and you'll have to push your own Docker image to enable that. Read the description here to see how: https://hub.docker.com/_/rabbitmq

Yep, I also tried to create my own copy, but it didn't really work out for me.

/bin/sh -c "CONFIG_PATH=/etc; SYSTEM_FILE=hosts; echo 127.0.0.1 ${RABBITMQ_HOSTNAME:-rabbitmq} >> ${CONFIG_PATH}/${SYSTEM_FILE} && rm -f /etc/rabbitmq/conf.d/management_agent.disable_metrics_collector.conf && echo management.listener.ip = :: > /etc/rabbitmq/conf.d/10-railway.conf && echo management_agent.disable_metrics_collector = false >> /etc/rabbitmq/conf.d/10-railway.conf && docker-entrypoint.sh rabbitmq-server"

Same problem, here.

I’m experiencing an issue with two newly deployed RabbitMQ nodes:

• Issue: Both nodes show “Overview: Management only mode” in the Management UI. Additionally, stats are disabled on these nodes.

• Context: I have one older node (7 months old) that is working normally. I did not redeploy it. The two problematic nodes were redeployed today.

• Impact: I cannot access queue/exchange statistics or monitor the nodes properly.

• Steps taken so far:

  • Checked the nodes’ logs — no obvious errors.

  • Verified the management plugin is enabled.

  • Compared configuration with the working node.

Questions:

1. Why are the newly redeployed nodes entering management-only mode?

2. How can I restore full stats and functionality similar to the working node?

Thank you for your help.

Additional info: The same configuration works perfectly on a Google VPS with EasyPanel, even without any special start commands.

The template uses the official RabbitMQ image. It's always been "management only mode" as that's the image we use - you may want to check the rabbitmq Docker image, as this may be a recent change by them. I'll check if there's any way to get the old behavior back.

@belkior @nekky can you both confirm what image you are using? It should be rabbitmq:management . It doesn't look like there was any recent changes regarding this.

Yes, that's right

Is there something wrong with the launch command itself?
/bin/sh -c "CONFIG_PATH=/etc; SYSTEM_FILE=hosts; echo 127.0.0.1 rabbitmq >> ${CONFIG_PATH}/${SYSTEM_FILE} && echo management.tcp.ip = :: >> /etc/rabbitmq/conf.d/10-defaults.conf && docker-entrypoint.sh rabbitmq-server"

FYI, the metrics are deprecated (and have been for 4 years):
https://www.rabbitmq.com/blog/2021/08/21/4.0-deprecation-announcements#disable-metrics-delivery-via-the-management-api--ui
I'm not sure why it doesn't work on Railway though - I'm able to run it locally.

This command helped me to start:

/bin/sh -c " CONFIG_PATH=/etc; SYSTEM_FILE=hosts; echo 127.0.0.1 rabbitmq >> ${CONFIG_PATH}/${SYSTEM_FILE}; echo management_agent.disable_metrics_collector = false \ > /etc/rabbitmq/conf.d/20-management_agent.disable_metrics_collector.conf; echo management.tcp.ip = :: \ > /etc/rabbitmq/conf.d/10-defaults.conf; exec docker-entrypoint.sh rabbitmq-server "

Btw, yes, it works locally for me too.

As I managed to find out:

In the Railway template for rabbitmq:management (version 4.2.2), the configuration file 20-management_agent.disable_metrics_collector.conf is connected by default.

This file sets the parameter management_agent.disable_metrics_collector = true. According to the RabbitMQ documentation (section “Management Plugin”), this flag completely disables internal statistics collection because it is intended for deployments that use external monitoring (Prometheus).

https://www.rabbitmq.com/docs/management#:~:text=In%20order%20to%20completely%20disable,even%20if%20collection%20is%20disabled

I think this is a bug with Railway's image pulling process, and they don't delete files correctly. I'll open an issue for this.

I can confirm that this is the case and I have created a bug report.

@six were you able to resolve?
I am looking to get queue metrics into my prometheus:grafana
Hoping I can monitor connections, queue size, etc

https://www.rabbitmq.com/docs/prometheus#quick-start

Sorry - Railway has not communicated anything about a fix:
https://station.railway.com/questions/image-layers-with-deleted-files-are-not-d7d808ee
Please comment on that thread so the issue gains more visibility and gets fixed.