[BUG] 2.11.0 makes chrome die on virtualised workstations

[m4rkw]

Describe the bug

2.11.0 causes dramatic high CPU in chrome on virtualised workstations. In our heavily security-focused environment these are accessed over Citrix which may also be a contributing factor. Appreciate this is unlikely to be a configuration that's ever tested or likely to be ever tested, however 2.10.0 does not do this and 2.11.0 is completely unusable for us because of this issue. The VM workstations in this case are Windows-based, have 8 cores and 16GB RAM and run on VMware.

The setup of our environment is likely immaterial, the more fundamental issue is that 2.11.0 puts much greater load on the browser than 2.10.0. With 2.11.0 if I open kibana and just click around cpu usage of chrome spikes to 60% and the entire VM lags severely for several seconds to the point of becoming almost unusable.

To Reproduce

I think it's unrealistic for these steps to be followed but will list them anyway.

1. Install dashboards 2.11.0 on a VM that is accessed over citrix
2. Seed some data into it
3. Set time range to 7 days
4. Click around in discover mode, clicking on the view for individual records is very slow to load, as is scrolling around

Expected behavior
Interface is quick and snappy

OpenSearch Version
2.11.1

Dashboards Version
2.11.0

Plugins

Security Dashboards

Screenshots

Not able to do this due to corporate security policies.

Host/Environment (please complete the following information):
Dashboards is running on CentOS 7.9 with 4 cores / 8GB ram
Google Chrome 110.0.5481.104 (64bit)

Additional context

Chrome cpu usage spikes dramatically when the dashboards interface is loaded. I suspect the javascript changes between 2.10.0 and 2.11.0 are the cause here.

Also we backported the fix for this issue: #5484

into our build of dashboards 2.11.0 by patching data.plugin.js to remove the async property.

[joshuarrrr]

@m4rkw This is an interesting finding, but one of the first things we'd want to look into is which OpenSearch Dashboards plugins you have configured and running, to narrow down the scope. Can you update this issue following the bug report template, so that we have a better idea of which plugins and configurations are in use?

[m4rkw]

@joshuarrrr sure, done, let me know if you need any more info

[wbeckler]

I think next step is for someone to do a Lighthouse audit on each of the versions to see if there's a measurable decline in performance in 2.11.