Consul lock prefix does not match cluster_formation.consul.lock_prefix

[xixka]

Dockerfile

FROM rabbitmq:3.8.9-management
COPY erlang.cookie /var/lib/rabbitmq/.erlang.cookie
COPY rabbitmq.conf /etc/rabbitmq/rabbitmq.conf
RUN chmod 777 /etc/rabbitmq/rabbitmq.conf
ENV RABBITMQ_VM_MEMORY_HIGH_WATERMARK=0.9
RUN rabbitmq-plugins --offline enable rabbitmq_peer_discovery_consul

rabbitmq.conf

cluster_formation.peer_discovery_backend = consul
cluster_formation.consul.host = consul-1
cluster_formation.consul.svc_addr_auto = true
cluster_formation.consul.deregister_after = 90
cluster_formation.consul.lock_prefix = ir

Startup parameters

    environment:
      - RABBITMQ_DEFAULT_USER=ir
      - RABBITMQ_DEFAULT_PASS=ir
      - RABBITMQ_NODENAME=rabbitmq

Then the following picture bug appears
lock_prefix name is irloopback_users.guest = false

[michaelklishin]

There is no evidence of a bug in RabbitMQ. The Consul plugin configuration schema generation is reasonably thoroughly tested so I don't expect any surprises there.

I remember seeing something like this before and it was a Docker image configuration "generation" side effect IIRC. The image concatenates its loopback_users setting without correctly adding a new line, so final rabbitmq.conf ends up begin

# note the missing new line
cluster_formation.consul.lock_prefix = irloopback_users.guest = false

or something like this. You can inspect effective node configuration to further confirm this. You can also try adding extra new lines at the end.

[michaelklishin]

Here is where the Docker image messes with the configuration file (for historical reasons).

If you unset RABBITMQ_DEFAULT_USER and RABBITMQ_DEFAULT_PASS , I suspect it won't mess with rabbitmq.conf.
Our team strongly recommends against using environment variables where they can be avoided, with Docker or not.
Sensitive configuration values such as default user credentials can be encrypted in advanced.config.

[michaelklishin]

Filed docker-library/rabbitmq#456 as I cannot reproduce this with just a rabbitmq.conf and a Consul node:

cluster_formation.peer_discovery_backend = rabbitmq_peer_discovery_consul

cluster_formation.consul.host = localhost

cluster_formation.consul.svc = rabbitmq

## do compute service address
# cluster_formation.consul.svc_addr_auto = true
## compute service address using node name
# cluster_formation.consul.svc_addr_use_nodename = true
## use long RabbitMQ node names?
# cluster_formation.consul.use_longname = true

cluster_formation.consul.svc_addr_auto = true
# cluster_formation.consul.svc_addr = warp10
cluster_formation.consul.use_longname = true

# cluster_formation.proxy.http_proxy = proxy.example.local
# cluster_formation.proxy.proxy_exclusions.1 = localhost
# cluster_formation.proxy.proxy_exclusions.2 = 127.0.0.1

cluster_formation.consul.svc_tags.1 = qa
cluster_formation.consul.svc_tags.2 = 3.8

cluster_formation.consul.svc_meta.owner = team-xyz
cluster_formation.consul.svc_meta.service = service-one
cluster_formation.consul.svc_meta.stats_url = https://service-one.eng.megacorp.local/stats/

# cluster_formation.consul.svc_ttl = 60

log.console.level = debug
log.file.level = debug

cluster_formation.consul.lock_prefix = rabbitmq-prefix

The prefix is what I'd expect: