User session are colapsing server enonic app

Enonic version: 6.4.3
OS: Ubuntu Server 18.04.2

Hello there Enonic community.

Today I’m writing to you asking for some help. This is the case: Currently, we have a web application that which purpose is to handle more than 2000 connections but for some reason, the sessions do not expire consuming the resources of the server.

My question regarding this situation is, do you recommend some special server configuration o settings management for an Enonic application that was created to handle that amount of incoming connections. Keep in mind that the server specifications are capable to manage that traffic, the problem is with 100 sessions the memory is totally consumed. Also, our server is hosted by Azure. If is there another information that you may need please don’t hesitate and ask for it :slight_smile:

In addition, do you know how to set up the blob storage configuration in order to implement an auto-scaling model in our server to prevent these scenarios?

This is a screen capture ofthe sessions that colapse the server memory.

I’m having troubles right now, I appreciated any comment that you may have to solve this.


First questions:

  • Why are you using a 3 years old version of XP?
  • How big is your Heap size?

Also, you might find these release notes interesting:

From the other improvements section: “Sessions are now lazy created on first access”

Simply meaning, sessions are not created unless needed…
BTW: you can also control default session timeout if required.

Recommend that you upgrade to newest version of XP 6.x as a start…

Hello there Tsi, thank you in advance for your replies.

In the first place we are concern about the importance of an upgrade thats why is there an 6.15 instance runing in a stateging environment but is not ready yet for a production realease. This cuestion is more related with which would be the recommended server configuration for a enonic application for that purspose (session management).

About your first question this the de heap memory.

In addition do you know how to manage the blob storage to implement auto-scaling model do you have any comments about it?


Easiest way to get autoscaling is to use dedicated front-end nodes, these nodes should have data=false flag set in properties. As these nodes are stateless you can quickly start and stop new once. The file system used for the blobstore will need to be on a shared disk. I.e. NFS, Amazon Elastic Filesystem etc.