Closed HowellTan closed 4 years ago
Hi @HowellTan, thanks for your input. Yes, that's how Prometheus scraping is implemented currently (and it's likely to stay like that because otherwise, Verne nodes would have to somehow forward the scrape request. Or have all the metrics replicated & ready).
@ioolkos I have the same issue, how do you suggest monitoring a vernemq cluster with multiple replicas?
For instance if I have a swarm cluster with three nodes, then prometheus scraps metrics from only one node at a time.
@tartieret @HowellTan by scraping every node and then separate metrics by node name in Grafana. You could do an aggregated view (where cluster nodes are shown in a single view), or some form of drill-down view for each node from a general pane.
I know the Grafana thing in this repo is only very basic. I've seen people building elaborated views, but no project has given back what they built so far (maybe for the reason that it seemed too specific for a use case).
@tartieret Netdata recently also added great integration for VerneMQ. You can add multiple servers in the config file too: https://learn.netdata.cloud/docs/agent/collectors/go.d.plugin/modules/vernemq
I guess this is answered. Feel free to comment or reopen.
@ioolkos thanks for the info!
Environment
Expected behavior
Configure VerneMQ such that the metrics includes all nodes (i.e. show data for the entire cluster).
Actual behaviour
I have read the docs for Prometheus VerneMQ. For context, I have a VerneMQ cluster with 3 nodes behind a load balancer. I want to use Prometheus to scrape VerneMQ metrics, but each request to :8888/metrics only has data for one specific node at a time.