Kibana/Metricbeat Visualization Errors (Too Many Records?)

I am noticing a lot of performance issues in Kibana with our Metricbeat metrics. We have 150 servers sending metric data and there are currently 375 million records in our Metricbeat index.

I reduced the frequency of the metrics from 10s to 30s in the Metricbeat config file, but I am not sure how to handle the 400 million records already in the index. None of the Metricbeat visualizations seem to load and I get "Discover: Gateway timeout" and "Metrics: error" errors. How should I proceed to fix the performance issues?

Can you share a little more info about your index setup? A lot of this depends on how the data is sharded.

It's common to use a new index per day with metricbeat:

output.elasticsearch.index: metricbeat-%{[beat.version]}-%{+yyyy.MM.dd}

Elasticsearch knows to only search indices that have timestamps within the time range. Daily indices make it handy for archiving too. If you're already doing this, is there a time range you're comfortable with limiting search results to?

This topic was automatically closed 28 days after the last reply. New replies are no longer allowed.