CPU over 100% when queues are not empty

[matthew-gill]

I think the title says it all.

using docker stats when my queues are empty, CPU reports around 3%

If I use the following command ONCE, the CPU jumps to 130%

aws --endpoint-url http://localhost:9324 sqs send-message --queue-url http://localhost:9324/queue/my_queue --message-body "Hello, queue"

Is this a known issue? Any ideas on how to fix it?

[Pixel-Jack]

Hey @matthew-gill , I have the same issue !

And moreover this container is constantly getting some information as input (Here Input are at 17.9GB after like 5 min of running) docker stats:

CONTAINER ID        NAME                CPU %               MEM USAGE / LIMIT     MEM %               NET I/O             BLOCK I/O           PIDS
ea3332804018        dev_sqs             21.41%              49.23MiB / 50MiB      98.45%              1.25MB / 21.4kB     17.9GB / 16.4kB     31

To avoid this container to overload my system, I've launch it with a cpu restriction of 0.2 In docker compose:

cpus: 0.2
mem_limit: 200M

If someone has any new idea, please share

[matthew-gill]

@Pixel-Jack I ended up swapping to localstack instead of this as it supports SQS... I also needed KMS as part of my stack and it came with that, too.

Perhaps a bit heavy handed if you only need SQS, however, it seems a lot more polite with CPU usage, so maybe give it a go?

It also seems a lot more maintained than this image, too.

[matthew-gill]

If it helps, here's a snippet of my docker-compose file which uses localstack to set up SQS and create two queues:

services:
    localstack:
        image: localstack/localstack
        ports:
            - "8000:8080"
        environment:
            - SERVICES=sqs
            - HOSTNAME=localstack
            - DEFAULT_REGION=eu-west-1
        volumes:
            - ./docker/config/localstack:/docker-entrypoint-initaws.d

Then create the following file:

./docker/config/localstack/01-queues.sh:

#!/usr/bin/env bash
awslocal sqs create-queue --queue-name YOUR_FIRST_QUEUE_NAME
awslocal sqs create-queue --queue-name YOUR_SECOND_QUEUE_NAME

[Pixel-Jack]

@matthew-gill Thank you, for the tip and the conf files ! I'll try this, this whole stack should be useful later so it won't be a waste :)

[rciorba]

I suspect it's the insights node app that is doing something strange here. I attached to the container and stopped the web app and now cpu usage is back to normal.

[xdxiaodong]

+1

[brendandebeasi]

@rciorba Any update on this?

[ucarbehlul]

I agree that the node app seems to be acting weird. Is there a way to make container start without the node app and only the SQS part (I can view the queue contents with aws-cli if needed)?

[rciorba]

FYI, there's an MR which also fixes this issue: https://github.com/roribio/alpine-sqs/pull/37 As the MR isn't merged yet, you can build your own image or use one I built: https://hub.docker.com/r/rciorba/alpine-sqs

[ronster37]

I also have been having over 100% CPU usage but clearing my queues from time to time seems to resolve my CPU usage and I'm actually able to hear other things besides just my laptop fans lol