Stable environment

[stufraser1]

We need to create a stable environment to run higher numbers of simulations and enable multiple concurrent users.
This may mean not using R SHiny but first step is to see if we can access more memory for processing and data, e.g. via AWS or Azure, which would provide scalable solution and could resolve issue without recoding the tool.

[stufraser1]

Shiny Proxy proposed as a solution to provide larger server to run multiple instances.

1. Tests on initial setup of Shiny Proxy with 4GB single core (?) were very responsive until c. 6 tabs running.

2. @KamalCharles Increased instance size to t3.2xlarge 8 (8x vpu, 32Gb RAM). I've also rebuilt the docker image/containers - With no errors present in the logs.

3. Tested by @jmcil @stufraser1, feedback:

• For both us, https://ui.oasislmf-ui-test.com/app/RiskExplorer loaded with incorrect formatting (everything loading up on one page instead of each tab). When we restarted the app, for me it loaded as we would expect to see with correct formatting. James still had some unformatted text related to shiny proxy at the top and bottom of page (which I didn’t have) – see attached. Both of us are on Firefox, and have been using Firefox on previous tests without these issues.
• Map loading was problematic – very slow when previously it loaded instantly, and jumped between zoom level erratically when zooming it
• It was the bits of code using Leaflet that were not very responsive and also plotly loss exhibits on tab 6 not showing at all.

4. @KamalCharles reported: the nginx server had reached full capacity so it was unable to parse entries into the access_log thus resulting the page not loading properly. Tested and working on Chrome and Firefox. I've also implemented an 'Auto scaling group' that should bump up the nginx write capacity and also scale up the RiskExplorer machine to 16 cores.

5. @stufraser1 tested with 6 tabs: experienced issues when getting to vulnerability tab (screen hung while loading table). Restarted app and opened 3 instances/tabs, I managed to run a sim on each of those tabs, but when the first sim finished it seemed to show the same results on all tabs even though I simulated three different areas.

[stufraser1]

@KamalCharles have you had chance to look at this issue?

[stufraser1]

Stable environment achieved Q1 2025. @KamalCharles to provide brief description of the environment including scaling capability here, please.