I have a question about the RabbitMQ module, specially about the queue metricset.
My metricbeat send a request every 10s but my rabbitMQ take more than 10s to respond. So, I think that metricbeat continue sending requests to http://*:15672/api/queues which caused a problem. It blocked the memory of RabbitMQ and the access to this url.
Thank you for your response. So, I have been using the default settings : period 10s timeout 10s connect_timeout 10s.
I have another question. For example, Let's take a case of a failure request.
If I set the timeout to 30s and I start a request at 11:00:00. The next request will start at "11:00:30" or "11:00:40" or "I will have 2 parallel requests at 11:00:30 and 11:00:10"?
Could we apply this timeout only on the queue metricset and keep the default settings for the other metricsets?
With a timeout of 30s and everything else default you should have parallel requests every 10s for 30s until they start hitting the timeout. For example:
11:00:00: 1 connection
11:00:10: 2 connections
11:00:20: 3 connections
11:00:30: 3 connections (the connection from 11:00:00 would timeout right about then)
11:00:40: 3 connections (the connection from 11:00:10 would timeout right about then)
Full disclosure, I haven't used the RabbitMQ module before but this is how I've seen other timeout settings behave. Depending upon your use case you could increase your period to 30s so you wouldn't have multiple connections in flight at once. Test it out for your unique situation though.
Try creating multiple rabbitmq modules in rabbitmq.yml. For example, if you want node to have custom settings and the rest be default you could do something like:
Apache, Apache Lucene, Apache Hadoop, Hadoop, HDFS and the yellow elephant
logo are trademarks of the
Apache Software Foundation
in the United States and/or other countries.