Cisco ASA clientless VPN breaks Atlassian Jira with Zephyr Scale

[Philip P]

Hello,

There's an Atlassian Jira instance I occasionally look after.  Users can access it using a a Cisco ASA clientless webvpn, which is a firewall that presents an SSL-secured website which users can access to browse the private side of the firewall.  To do this, it intercepts and rewrites the various web code passing through it to maintain its position between the user and the protected assets.

The clientless VPN worked fine until we installed a software test app called Zephyr Scale.  When it's enabled, large chunks of the Jira interface disappear when browsing via the clientless VPN, making it useless.

The authors of Zephyr Scale advise that Jira combines multiple resources into a single batch.js file which isn't being translated correctly by the clientless VPN.  To test this theory I set Jira to run in its 'developer mode' which disables batching.  The result is that Jira works perfectly over clientless VPN, but very slowly since developer mode enables extensive debugging.

Would anyone be able to offer any further suggestions please?

Is there a way for me to disable batching without having to use developer mode?

Thanks,

Philip

[Nic Brough -Adaptavist-]

Welcome to the Atlassian Community!

The short answer is "no". 

The batching is done for speed and efficiency and can't be turned off.  Smartbear (Zephyr) are taking advantage of the efficiencies by bundling their code into it (I actually wish more app vendors would do it more, because it's almost always faster than doing it all separately)

My instinct is that this is something that needs to go to Cisco, their clientless VPN is failing, not the services you are trying to use.  For a start, you mention "rewrite" - that is a terrible thing to do, especially for Cloud based services, and you've run into a case where they should stop doing it.

[Philip P]

Think of it more as a translation than rewriting and you may find it less horrifying.  I agree it's messy, however for the most part it provides a useful solution to a problem when it's needed.

It's disappointing that Adaptavist / SmartBear won't provide an option to turn off the super-efficient batching that unfortunately breaks my use case.

[Nic Brough -Adaptavist-]

It's not the batching that is the problem, it's what your VPN is doing with the results.

The batching is fine with most other systems, and it's done to standards.  If your VPN doesn't support standards, you can't expect someone to have to code for non-standard things.

Edit - sorry, that was a short and terse response, I did not mean it to be rude.

I'm just trying to point out that the problem is not with Atlassian, it's the VPN software that is breaking it.

You yourself have tested what happens when batching is turned off - the software becomes painfully slow.  I would report exactly that to Cisco and ask them to fix their software!