Help with Reverse Proxy Setup and Multiworkspace
#Help with Reverse Proxy Setup and Multiworkspace
fickle dock
stoic abyss
same issue here
fickle dock
apparently activating multiworkspace needs a whole lot of dependencies inside of the workspace container, try docker exec -it twenty-server-1 yarn and docker exec -it twenty-server-1 npx nx database:reset --configuration=no-seed afterwards but make sure to have a backup 🙂
explains a whole lot why the application wasnt loading after this
but the dns wildcard setup is pretty messy 😦
tame saffron
apparently activating multiworkspace needs a whole lot of dependencies inside of...
Hello there
No added dependencies needed to run a multiple workspace instance
Also I would not recommend installing dev dependencies in your production container
Meaning you should not run NX command in prod but rather use shiped in available yarn scripts
Are you still facing issue with your reverse proxy ?
fickle dock
hi @tame saffron - thanks for the quick question.. i would try a fresh reinstallation today without any dev/nx commands.
fickle dock
Are you still facing issue with your reverse proxy ?
kind of.. the multiworkspace urls are kind of hard to predict
tame saffron
kind of.. the multiworkspace urls are kind of hard to predict
Hey no worries
Ah yes for sure they're randomly generated if not set, but you can customize them from your workspace settings
That's why we need a wildcard
fickle dock
is there any possibility to set the multiworkspace urls beforehead? that would save some time because we aren't doing wildcard le certificates.
or at least make them "guessable"
tame saffron
Unfortunately I don't think so 🤔
We're currently working on workspace import export as code
Such a behavior might be available by the end of Q3
fickle dock
yeah, that would be super cool
tame saffron
Even though the first import/export feature will be very workspace scoped
Until now what I would do is programmatically call the api setting a custom domain
Within a script
fickle dock
oh this would totally work for me
tame saffron
With a deterministic payload
fickle dock
Late catchup, solved it like this without Wildcard Subdomains:
#0 Create DNS Records of Reverse Proxy beforehead (app.crm, crm.)
#1 Fresh Install of TwentyCRM without Multiworkspace enabled
#2 Create a first User
#3 Reconfigure for Multiworkspace, Restart Containers
#4 Try opening app.crm.. now - You'll be redirected to a non-existent Subdomain
#5 Add the newly spawned url (email-uid.app.crm.) to Reverse Proxy URL
#6 Login to your new CRM with the new URL
#7 Change the domain in Settings -> Generic -> Change Domain to your liking and add to your Reverse Proxy
#8 Repeat steps 5-7 for each new Multi-Workspace
#9 ???
#10 profit 🙂
Thus, predictable Subdomain URLs for new Workspaces would be a great great thing
sage narwhal
Hi @fickle dock , I think I'm running into the same problem when hosting on AWS ECS with the Twenty docker image. Before enabling IS_MULTIWORKSPACE_ENABLED in the environment variables, I had no issue. After, I'm getting the "app.<aws load balancer string>" url popping up. I found the subdomain for the workspace in DB -> "radiant-bronze-peacock".
What should the url from step #5 look like? "radiant-bronze-peacock.app.<aws load balancer string>" isn't working, neither is removing "app" from the url string.
sage narwhal
It seems like I may need to set up subdomain access in my load balancer as well