Autoscaling fails with Premature end of chunk coded message body

Products

• VMware Tanzu Application Service for VMs

Issue/Introduction

You noticed issues with autoscaling performance which includes the following symptoms:

• applications not being scaled up to the desired instance
• scaling applications results application stuck in starting state

When checking for the autoscaling app logs

cf target -o system -s autoscaling
cf logs autoscale --recent

OR

cf logs autoscale

The logs contain the following error:

Non HTTP response code: org.apache.http.ConnectionClosedException/Non HTTP response message: Premature end of chunk coded message body: closing chunk expected

Environment

TAS autoscaler

Cause

The error:

Non HTTP response code: org.apache.http.ConnectionClosedException/Non HTTP response message: Premature end of chunk coded message body: closing chunk expected

means that connections are being closed prematurely. One common cause is due to heavy load on autoscaler.

Resolution

 

 As your autoscale requests increases you need to tune other components of TAS which include the following:

• make sure quotas are not near reaching its limit
• review scaling values and check if log cache is underscaled. This includes tuning log cache's number of instances, memory and settings for Maximum number of envelopes stored in Log Cache per source
• Scale or increase your autoscale apps instances to accommodate more requests