Open Gama11 opened 7 hours ago
Possibly relevant, our rabbitmq.config
:
[
{
rabbit,
[
{ loopback_users, [] }
]
},
{
rabbitmq_management,
[
{ load_definitions, "/etc/rabbitmq/definitions.json" }
]
}
].
definitions.json
:
{
"users": [
{
"name": "test",
"password_hash": "<redacted>",
"hashing_algorithm": "rabbit_password_hashing_sha256",
"tags": "administrator,monitoring"
}
],
"vhosts": [
{
"name": "/"
}
],
"permissions": [
{
"user": "test",
"vhost": "/",
"configure": ".*",
"write": ".*",
"read": ".*"
}
],
"topic_permissions": [],
"parameters": [],
"global_parameters": [],
"policies": [],
"queues": [],
"exchanges": [],
"bindings": []
}
Thank you. This looks like a blank node's config with a pre-configured user for remote access.
I was able to reproduce with a blank node from main
:
# HELP rabbitmq_queue_exchange_messages_published_total Total number of messages published into a queue through an exchange
rabbitmq_queue_exchange_messages_published_total{vhost="/",queue="qq.1",vhost="/",exchange=""} 100000
I think this is related to https://github.com/rabbitmq/rabbitmq-server/pull/11559 the first vhost belongs to the queue, the second to the exchange but obviously they are the same so need to special handle this new case to eliminate the redundancy
Describe the bug
Prometheus seems to choke on duplicate labels:
Note how
vhost
appears twice here:Versions used:
This seems to be a regression compared to 3.13.7 which we upgraded from.
Reproduction steps
rabbitmq_prometheus
pluginrabbitmq_queue_exchange_messages_published_total
metric is included/metrics/per-object
endpointExpected behavior
There should not be duplicate labels.
Additional context
No response