Skip to content
This repository has been archived by the owner on Nov 6, 2018. It is now read-only.

message-gateway memory issue #516

Open
iscra opened this issue Dec 9, 2016 · 2 comments
Open

message-gateway memory issue #516

iscra opened this issue Dec 9, 2016 · 2 comments
Assignees

Comments

@iscra
Copy link

iscra commented Dec 9, 2016

The message-gateway seems to be leaking memory.

Doing some load testing, just 1 queue, after sending 50k message, the memory usage goes up to 1GB. With longer tests up to 8GB, seems to be linear to message count.
There are active consumers, all the messages are acknowledged.

This happens only when using gateway, not when using message-broker directly.

Also after fixing broker detection in the gateway, it shows (jmx):

  • gateway has local queue with all the messages, all acknowledged
  • gateway still leaks memory as before
  • broker has local queue with all the message but none are acknowledged
  • broker uses much less memory, about 400M for 50k messages
  • only the gateway queue has consumers

Is there some documentation / concept how the gateway and broker should behave in detail?
Describing exactly simple use case with 1 message and queue would help a lot.

Is it correct that the gateway also has a copy of the queue?
All consumers in gateway and none in broker - is that correct?
No ack messages in broker?

@iscra
Copy link
Author

iscra commented Dec 9, 2016

Broker detection fixed in #517

@jmjava
Copy link

jmjava commented Feb 16, 2018

We are considering using Fabric8io messaging. Is it ready for production use? Does this represent a critical issue?

Sign up for free to subscribe to this conversation on GitHub. Already have an account? Sign in.
Labels
None yet
Projects
None yet
Development

No branches or pull requests

3 participants