@chrisronline xpack.monitoring.collection.enabled is not in ESv5
My bad! But xpack.monitoring.enabled definitely is. Try enabling that
@chrisronline looks like you can have xpack.monitoring.enabled: true with xpack.monitoring.exporters.my_remote setting I get an error-
Caused by: java.lang.IllegalStateException: settings object contains values for [host=["x.xx.xx.xx:9200"]] and [host.0=x.xx.xx.xx:9200]
Can you please confirm?
Is there anything in the log that indicates that monitoring exporters are set and their post status?
Is this correct setting -
xpack.monitoring.exporters.my_remote.type: http
xpack.monitoring.exporters.my_remote.host: ["xx.xx.xx.xx:9200"]
The setting on the cluster nodes being monitored as this is correct-
xpack.monitoring.exporters.my_remote.type: http
xpack.monitoring.exporters.my_remote.host: ["xx.xx.xx.xx:9200"]
@chrisronline the kibana(v6.8.1) dashboard is displaying that my license will expire in x days. If this has the Gold version features of multi-cluster monitoring and alerts then I understand but I do not see those features, so why is this license expiring?
Secondly, how can I have a trial license to evaluate Gold features? How do I configure alerts. If you have a video explaining this can you send a link please?
Hi @shradhatx
See here for starting a trial license: https://www.elastic.co/guide/en/elasticsearch/reference/6.8/start-trial.html
See here for alerts information: https://www.elastic.co/guide/en/elastic-stack-overview/6.8/how-watcher-works.html
@chrisronline - Is there a way to turn on "spy" on monitoring dashboard? I want to utilize underlying queries to set up watcher.
Unfortunately, not currently.
The data shown in the charts will always be sourced from Elasticseach, but some graphs do some post and pre-processing of the data that you'd need to mimic in order for the watch to operate on the same exact dataset as seen in the graphs.
It's fairly advanced and I'd recommend driving into the codebase and tracing through how a particular graph sources its data and that'll help get you started
Let's look at an example.
Consider the Search Rate graph on the Elasticsearch overview page.
To start, we find the appropriate route that renders the data for this page. It's discoverable using the browser network developer tools. For this example, it's this route.
This particular line indices what set of metrics will be returned from this route, which are defined here.
The first one looks right: cluster_search_request_rate.
Then, we find the metric definition for that particular metric. Part of that definition includes the field from which the data is sourced, which in this case is: indices_stats._all.total.search.query_total.
But that's part of the battle.
Some of the collected metrics operate like counters, where they are ever increasing once the process starts. In these cases, we need to perform a little more work to calculate the data to show in the chart, specifically a derivative aggregation.
The class definition for our metric indicates that it does indeed need a derivative applied.
Tracing even further into the code, we can see how we use the result of the derivative aggregation instead of the raw value.
Another thing to consider is any custom pre or post processing happening with this metric.
Look up the class definition for the metric and note any overridden calculation functionality. In this case, there is none, but some will and those calculations will need to be applied in any custom watch created.
Hope this helps!
This topic was automatically closed 28 days after the last reply. New replies are no longer allowed.