Why is Citrix so bad at error-handling seemingly simple issues?

[Peter Fibæk]

We've just been hit by yet another instance of bad error handling from Citrix in general.

One of our 6 Citrix client sesssion servers (Virtual APP) had run out of space on some disks.
This caused all connected users to either become disconnected and unable to connect or with temporary profiles connect to other Citrix servers; becoming handicapped users (not access to all systems).
Also prevents some new users from connecting, because they are being distributed to that same faulting server.

I don't understand how one server going "bad" can create so many problems and the system itself doesn't just quickly find alternatives or isolate the problem creating server.

1. Don't assign new users to the server; since it has low space and needs human attention. Because if it runs OUT of space, all hell breaks loose!

2. If a user reconnects and has a session on the faulting server; ask them to relocate to another server (give them the option to log out of the faulting server and be assigned another).

  - They can choose to reconnect, if they have important stuff going on in the session or relocate, loosing any unsaved work; but that will be their choice and it will alleviate a problematic situation by removing a user from the equation.

 

I can think of many more automatic responses to any problem situation, that I think the Citrix system should employ. Far too many times some odd little problem (low space) seems to create much larger systemic problems for us, which could be handled or mitigated quite quickly and easily by automatic mitigation, before users even notice there is a problem.

But noo... a server behaving badly? Lets just keep assigning users!
A server seems to mishandle all users assigned (lots of user connection issues), so it doesn't have many users? Well, here, have some more!
 - IT SHOULD STOP BEING ASSIGNED USERS IF IT STARTS HAVING ANY KIND OF PROBLEMS!

Citrix developers!
Work on your DDC's so they gain some bloody intelligence in assigning users and monitoring the system for problems...

Light up the alarm board instead and stop assigning users to faulty servers!

[Michael Adam]

Hello @Peter Fibæk

 

i hope you are aware about configure the Load Management via policy. If you disk is in heavy usage or full the Disc queue lenght is growing. With this you can control new user sessions distribution.

 

Cause i can`t add a screenshot today here you find the explanation from the Citrix Policy "Disk Usage" unter "Load Management".

 

Description

Defines the disk queue length value at which the server reports 75% load.

Zero load is automatically determined to be one third of the 75% load disk queue length value. It is a common rule of thumb that when disk queue length exceeds twice the number of spindles, a disk bottleneck is likely developing. This should be considered in conjunction with the disk topology of the server when setting the '75% disk queue length' value.

 

Related settings

Maximum number of sessions, Concurrent logons tolerance, CPU usage, Memory usage, Memory usage base load

 

 

Regards,

Michael 

[Michael Adam]

PS: via iPhone the picture upload was now possible - with notebook not. Am I really the only one who has had problems answering questions, searching and following here in the community for weeks? I receive so often a HTTP 403 CloudFront error  anyone else? 

 

Studio Policies for Load Management:

Disk usage Policy: