Version 4.0.8 ubuntu, on a web app, linux B2 plan currently.
Does anyone else have an issue with memory consumption running in an azure web app?
When I have a single app running in PSU my memory use is around 1.8GB, there’s nothing fancy going on either in my code, just a couple of pages with some data being pulled from sql, but nothing vast, there’s only a single user session open which is my own.
If I add a new app, and select the multiple pages demo, so really not much there. The moment the app runs, it jumps from 1.8GB to 3.7GB usage, which seems incredibly high for a single additional app. I can confirm this is the case by deleting the app and watching it drop back down to 1.8, and repeating the process I get repeatable results.
If this is the norm, it might not be great in scaling long term, just wondered if I’m maybe doing anything wrong or what others experiences were around scaling and azure.
A bit more context around my environment, I’m using a fileshare/storage account.
Currently on a linux B2 tiered web app.
Are you by chance running the apps/dashboards using the Integrated environment?
Because that exact behavior is what I saw when I had the Integrated environment as default for all apps/dashboards.
As soon as I changed them all to use anything but the integrated environment, RAM usage dropped to max 2GB for 7 medium sized dashboards with an average of 12 user sessions.
There a small section about this in the docs, Best Practices section:
I did actually see this, and was initially doing that. I can say that my upfront ram usage does drop considerably lower when not using integrated, so thats good.
But the ram usage per new app still seems quite high, for example with a single app which has multiple pages being dot sourced etc I’m using 283mb, when I add a blank template test app with a single page, it goes up to 1.6GB.
7 dashboards and below 2GB sounds ideal, whats your setup out of interest?
I’m using an azure web app, with mounted fileshare.
I actually backed out of using sql/git sync initially since git was not bundled in earlier versions and I was seeing even higher memory use from that setup too. The more I read into using sql, the more it seemed specifically for HA so I decided to run with a file share instead.
I’m running universal as Azure Web App (Windows) with SQL (Azure SQL) and Git in Push-Only.
Something I didn’t think about when writing my first comment, I’m on 3.9.14 still - since v4 has been anything but forgiving to me in the multiple tests I’ve done thus far.
Each dashboard uses it’s own custom module for common functions and all external information is gathered in their own respective runspaces through Endpoints or Scripts.
My aim has been to not rely on any external modules from within each dashboard. So no Az.*'s, no bloated dbatools or similar modules will be used when spinning up the environment runspace for said app/dashboard.