check-rabbitmq-node-health picks only first nodes in the cluster

[rgedagit]

When we have a list of nodes in cluster it always pick first nodes in list and show health is ok.

call is made for nodes ("#{url_prefix}://#{host}:#{port}/api/nodes/) -> returns a list of all nodes in cluster
function to pick node nodeinfo = JSON.parse(resource.get)[0] -> picked first nodes in the list.

ex: if the list is rabbitmq1, rabbitmq2, rabbitmq3, it checks the health on rabbitmq1 only always as it is first is list.

[arjuntalla]

it is low hanging fruit and here is the proposal for the fix if everyone agrees i will move forward with the fix

change the call from "#{url_prefix}://#{host}:#{port}/api/nodes”, -> "#{url_prefix}://#{host}:#{port}/api/nodes/rabbit@#{host}”

change the nodeinfo from nodeinfo = JSON.parse(resource.get)[0] -> nodeinfo = JSON.parse(resource.get)

make the host address not optional and make the hostname as real hostname

[rgedagit]

Not a bad idea, but why can't we use rabbitmqctrl instead of the admin API and also makes less complex

[majormoses]

@arjuntalla can you link me to the proposal?

[majormoses]

Apologies misread thought you meant there was an existing proposal.

[majormoses]

that makes sense to me, we should maybe have a default of the first node (or even better try to find the right node based on something we can know like its ipaddress?). This makes it a backwards compatible change.

[rocdove]

"#{url_prefix}://#{host}:#{port}/api/nodes/rabbit@#{host}”， it‘s a good diea. but the ’host ‘ must a hostname, IP address will don't work.

[majormoses]

makes sense from what I saw of your commit I think yours works, do you want to pr it for a bit closer review?