Get Total number of queues, channels,connections,consumers running within a specific host

[Chavya Khanna]

Hello Everyone,

Was hoping if one of you could assist - so basically we have multiple vhosts on RMQ cluster and we have several queues under every vhost.

Is there a way we can get total number of :

1. Queues
2. Channels
3. Connections
4. Consumers

On the basis of filter vhost. Unfortunately vhost is not supported as a parameter of "rabbitmq_queues" and so on.

Was hoping if someone could assist on it.

Regards,

Chavya Khanna

[Michal Kuratczyk]

Hi,

What are you referring to when you say "rabbitmq_queues"?

There is the Management API that can return this information: https://rawcdn.githack.com/rabbitmq/rabbitmq-server/v3.10.1/deps/rabbitmq_management/priv/www/api/index.html

There are per-vhost endpoints such as /api/queues/vhost, /api/vhosts/vhost/connections, /api/vhosts/vhost/channels and /api/consumers/vhost

There is also a Prometheus endpoint from which you can get that list, although perhaps not directly. For example in the /metrics/per-object endpoint, you'll find:

rabbitmq_queue_consumers{vhost="/",queue="myqq"} 1

Since Prometheus metrics only have numerical values, the list of queues will be the list of "queue" label values for a given "vhost" label value.

Best,

--
You received this message because you are subscribed to the Google Groups "rabbitmq-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to rabbitmq-user...@googlegroups.com.
To view this discussion on the web, visit https://groups.google.com/d/msgid/rabbitmq-users/300d54f4-0356-4872-b771-0abc004638afn%40googlegroups.com.

--

Michał

[Chavya Khanna]

Thanks Michael,

So in Prometheus we do have  rabbitmq_queue_consumers{vhost="/",queue="myqq"} , However the metric listed below does not accept vhost as a parameter.

1. rabbitmq_channel_consumers
2. rabbitmq_queues
3. rabbitmq_identity_info
4. rabbitmq_channels
5. rabbitmq_connections

As per my use case, i need Total queues, Total channels, etc that fall under a specific vhost.

Let me know if this clarified my ask?

Regards

[Michal Kuratczyk]

As you noticed, these aggregated metrics are not split into per-vhost values (because they are aggregated :) ). Some metrics returned by /metrics/per-object, do return the vhost, so for example, you can count

the number of values returned by "rabbitmq_queue_messages_ready{vhost="/"}" to get a number of queues in vhost /. Similarly, you can count the number of values returned by "rabbitmq_channel_get_empty_total{vhost="/"}",

to get a number of channels in the / vhost.

The point is - the fact that the value you are looking for is not explicitly there, doesn't mean you can't calculate it - you can use the count() function, as in:

count(rabbitmq_queue_messages_ready{vhost="/"}")

But the Management API could be a simpler solution.

To view this discussion on the web, visit https://groups.google.com/d/msgid/rabbitmq-users/fe1954c6-ec2b-4779-a0f8-c1eec35aa970n%40googlegroups.com.

--

Michał

RabbitMQ team

[Chavya Khanna]

thanks alot Michal,

Apologies to bother you but it seems the metrics to get Channels does not show accurate results when we compare with RMQ Console

1. Total Channels
   count(rabbitmq_channel_get_total{vhost="$rabbitmq_vhost"} * on(instance) group_left(rabbitmq_cluster) rabbitmq_identity_info{rabbitmq_cluster="$rabbitmq_cluster"})Output: 322 - Expected: 2046 
   tried using rabbitmq_channel_get_empty_total - still not accurate
2. Could not find any way to get total connections, consumers, publishers ,  "queue declared, queue created, queue deleted" at Host level  - there's no metric anywhere close to it that is fetching this info - for consumers i could find one but it seems to reflect data at individual queue level and not just the vhost level - for vhost it just shows "N/A" when there are consumers 
3. Would you happen to know where we can get parameter information for "label_values" which is usually used in Prometheus at the time of defining variables, just curious if we can filter queues in dropdown based on vhost

Thanks alot for your assistance on this matter, hope to hear from you on this.

[Michal Kuratczyk]

You need to play around and see where the problem is. The metrics are collected here https://github.com/rabbitmq/rabbitmq-server/blob/master/deps/rabbitmq_prometheus/src/collectors/prometheus_rabbitmq_core_metrics_collector.erl but it's one of the most cryptic piece of code we have as it exposes metrics mostly designed pre-Prometheus into a Prometheus data model.

If you have specific requests to add a label to a given metric - please raise an issue/PR.

To view this discussion on the web, visit https://groups.google.com/d/msgid/rabbitmq-users/eb2bff6e-48fc-4d49-8642-6873cfb1d108n%40googlegroups.com.

--

Michał

RabbitMQ team