Persist measures on the monitored host in case of network faillure

[anon24174512]

Hello,

Is there a way from the metricbeat sink to persist the measures on the monitored node in case of network failure so as to send them back to elasticsearch once the network is back and not loosing them ?

Or do I have to user a logstash instance (or kafka or any other means) to act as an intermediary persisted queue on the monitored node ?

Thanks,

[exekias (Carlos Pérez Aradros)]

Hi @anon24174512,

You can configure a number of max retries when sending Metricbeat data to Elasticsearch: https://www.elastic.co/guide/en/beats/metricbeat/current/elasticsearch-output.html#_literal_max_retries_literal.

Of course, this is not enough, as messages are kept in memory. As of today, we don't have spooling to disk implemented for this situations (but we want it), so you guessed correctly, you can use Logstash for this, as it implements this mechanism, kafka would be another option.

Best regards