Petrosian in Docker - strange Service UI behaviour and significantly slower response

[Fryguy]

Discussed in https://github.com/orgs/ManageIQ/discussions/22992

^{Originally posted by eggoide April 15, 2024}
Hi all, I would like to kindly ask you for a help or advice regarding strange behaviour we are experiencing with Petrosian on Docker.

We have our environment (najdorf-1.3.20220715135859_39a22e6) consisting of 2 servers. As we need to upgrade and move to dockerized solution, I have cloned the current DB and created 2 docker containers with similar CPU/RAM specs as current Najdord environment.

There are certain weird issues in Service UI:

1. In Service Catalogs, there is a multiple occurence of a service catalog in html code in drawer-pf hide element. The catalogs in this element are not displayed but the element is significantly increasing the html size. There is no such thing in Najdorf.
   Petrosian:

Najdorf:

2. A random form also contains multiple occurence of a text field with ID, that should be unique and that has been unique in Najdorf. In Petrosian, we have them again in drawer-pf hide element. Now, you cannot use automated tools for testing as you cannot select the text field as its ID is no longer unique within the html code.
   Petrosian:

Najdorf:

3, Overall speed of the UI that can be seen i.e. in initial Dashboard. It takes more than 34 seconds to load in Petrosian, while the same Dashboard takes 11 seconds in Najdorf. Even though the initial screen is loaded, it continues to load more elements for many minutes (which I guess is a feature and not an issue). However, overall experience is way slower than in the previous version.
Petrosian:

Najdorf:

Anyone has been dealing with similar problems? Any advice would be higly appreciated! Thank you for your time!

[eggoide]

Ok, probably the common cause why Service UI (it is not happening in regular ManageIQ UI) is so slow with multiple html elements is actually that the whole page is being mirrored multiple times in the notification bar. Here are the examples:
Services:

Service Catalog:

Random form:

That means that all the services, VMs and form logic is being initiated multiple times which slows down whole UI.
The behaviour is same when accessing ManageIQ via Nginx domain or directly using ManageIQ's hostname. So the webserver does not play any role.

Any idea why is it happening?

[eggoide]

Another related question is - I tried to turn off the notifications temporarily by disabling Service UI notification in the role of the user but it had no effect. Is it possible to turn it off somehow?

[eggoide]

[eggoide]

@Fryguy @jeffibm Hi, I have found out that the problem is probably related to the https://github.com/ManageIQ/manageiq-ui-service/pull/1865
As quinteros is not stable and this bug is not fixed in petrosian , is it possible to build it for petrosian?

[eggoide]

I have additional information - we have replace /ui/service in Petrosian with Quinteros version and it seems that it fixed the problem with Notification bar.