awslabs / awslabs/amazon-sqs-java-messaging-lib

Spring DMLC - SQSSessionCallbackScheduler too many zombie threads

Open
#47 1 comment 0 reactions 0 assignees View on GitHub
Dominant language
Java
Stars
183
Forks
150
PR merge metrics
No merged PRs in 30d

Description

We've been facing high CPU utilization after putting in SQS JMS based implementation on production.

The root cause turned out to SQS connectivity issues and the use of CacheLevel: CACHE_CONSUMER and a Fixed BackOff. Turns out CACHE_CONSUMER is too aggressive and even with an ExponentialBackOff, it will try to retry quite aggressively. After changing the CacheLevel to CACHE_CONNECTION, the CPU utilization is stable in our test environments.

We are facing another issue though. To simulate SQS connectivity issues, we kill the network connection and monitor using VisualVM. The no. of threads created for SQSSessionCallbackScheduler are quite high and they never exit.

See screenshot below:
screenshot 2017-11-26 11 12 54
screenshot 2017-11-26 11 21 18

After resuming network connectivity, the threads don't seem to go down and new threads are created after this point.

We are using Spring DMLC with the below configuration:

ConcurrentConsumers: **10**
MaxConcurrentConsumers: **25**
CacheLevel: **CACHE_CONNECTION (1)**
AcknowledgeMode: **CLIENT**
BackOff: **Exponential ( InitialInterval: 5000, Multiplier: 1.5)**
IdleConsumerLimit: **1**
IdleTaskExecutionLimit: **5**
ReceiveTimeout: **1000**

Any help?

Contributor guide

Open the contributing guide

Assessment

This issue has not been assessed yet.

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.