4.5.0 - High Availability (2 Nodes) w/ Load Balancer

[mwatts-seg]

Description of Issue

Hitting F5 Load Balancer (https://psu.ourdomain.com/) between Two Servers (Server01 & Server02) with a DFS Share between the two servers for repository.

Server01 is showing the dashboard as Starting and Server02 is showing running.
I did open a ticket. Not sure why the inconsistency between the dashboards in HA.

I plan on opening a feature request to be able to monitor dashboard status between hosts for instances like ours that it is load balanced.

Version

4.5.0

Severity

High

Hosting Method

MSI (Windows Service)

Operating System

Windows

Database

SQL

Licensed

Yes

Features

ADFS SSO

Additional Environment data

Windows Server 2022

Screenshots/Animations

No response

[mwatts-seg]

@adamdriscoll - Not sure if this is related or should I open a new issue?
A script I created and scheduled is unable to execute:

Error executing job: Unknown script: Retail\Citrix\Admin\CitrixRebootApplicationServers.ps1

The script is on a DFS Share between the two servers and I am able to see it on both nodes, but it refuses to launch it via Schedule.

[adamdriscoll]

Is this a new issue with 4.5? I know you've been using DFS for a while and I'm wondering if it stopped working at some point.

[mwatts-seg]

Is this a new issue with 4.5? I know you've been using DFS for a while and I'm wondering if it stopped working at some point.

This does seem new. We've been running other jobs without any issues, but this particular one was created on 4.5 (or 4.4.1?). I logged into each node directly (not through the load balancer URL), and both nodes can see the scripts.

The strange part is that it feels like the nodes aren’t aware of each other. They show up in "Computers," but some users were getting a "Dashboard not running" message. I opened a ticket and started digging and I found one of the nodes stuck in Starting.

On another note, I’ve noticed in the past that if I made a change to the dashboard, some users wouldn’t see the update right away. I always assumed it was just a caching issue, but now I’m wondering if that could’ve been a symptom of this problem.

I also found a bunch of failures in Hangfire, but they’re from two years ago.

On a side note, we’re planning to move our dashboards and scripts to Azure DevOps in the future, and we’ve started testing 5.X with it.