Keywords: CouchDB - Amazon Web Services - Technical issue - Other
We recently upgraded our aws t1.micro instance to a t2.medium instance both with Bitnami CouchDB because the load on the micro was too high. After a few weeks of running, the server became unresponsive and AWS support diagnosed the problem to be:
However looking at the EBS volume metrics I found that the instance has exhausted all its burst balance  and this seems to be the cause of your issue. In fact I can see that starting from 15:03 UTC of the 28/02 the disk started reading a lot of data , producing a huge number of IOPS that your disk was not able to handle as it has been provisioned with the minimum baseline of IOPS. This event caused the termination of your burst balance and so other services will have a high latency in responses due to the fact that the disk is still busy and continues to receive requests. My guess is that CouchDB is performing some operation that is wasting all your disk resources.
Does this make sense? I would expect the Bitname Certified server to be auto-configured with the proper settings for CouchDB.
What is the best (EBS) setup for CouchDB?
AMI: bitnami-couchdb-2.1.1-1-r02-linux-ubuntu-16.04-x86_64-hvm-ebs-mp-73a0394f-331c-40ca-a96d-ef4f36fc46ec-ami-c7d1e1bd.4 (ami-114f2e68)