Deploy to Heroku isn't working


#1

Hi!

When I click Deploy to Heroku on this page:

I get a message from Heroku saying:

“The given URL (github.com/metabase/metabase-deploy) is not a valid GitHub repository.”

Can’t wait to try metabase out, thanks for creating it!


#2

The link should be fixed now. It should point to https://dashboard.heroku.com/new?button-url=http%3A%2F%2Fdownloads.metabase.com%2Flaunch-heroku.html&template=https%3A%2F%2Fgithub.com%2Fmetabase%2Fmetabase-deploy

Also be aware that sometimes the application takes too long to start and Heroku times out. If that happens just try again. We’re working on making this more robust.


#3

Hi Tom, when you say “try again”, do you mean try deploying again? I tried restarting the app several times to no avail. I’m also not sure how you increase the timeout for Java apps sans a gem :confused: Thanks in advance. I’m a long time Looker user - excited to give this a whirl.


#4

I was able to deploy to heroku, but I keep getting an Application Error when I try to visit the app setup url.


#5

We’ve had a couple folks mention that they are running into memory limitation issues when using 512m dynos. It’s possible that’s what you’re running into here.

We haven’t hardened the Metabase deployment on Heroku (yet), so we are still finding out about some of these issues. We love the simplicity of Heroku though, so we’ll definitely be working to get these problems resolved as soon as we can.


#6

The memory issue is definitely what’s stopping it working on Heroku for me at the moment. Would your advice be to run it on a bigger instance for now?


#7

Yeah, if you have the ability to provision some more resources for the instance that’s your best bet for now.

We try to keep things as lean as possible as we develop, but we haven’t spent focused time trying to optimize for small memory scenarios. We run quite a few Metabase deployments on AWS using t2.small instances (2gb total memory) and we’ve never had any issues, so that’s a good benchmark. 512mb is a pretty small amount of memory for running this type of application.


#8

That seems a bit crazy, that “this type of application” needs so much memory. On Heroku (on which standard dynos with 512mb run thousands of production apps), getting to above 2gb would cost $250 a month. That’s quite a steep entry point


#9

Just to be clear, I was not saying that Metabase requires 2GB of RAM, simply that we hadn’t spent time trying it in smaller memory situations because frankly it hadn’t come up.

We run Metabase on AWS quite a bit and estimate that if you want to use ElasticBeanstalk with t2.small instances plus an RDS database your total cost is roughly $50/mo, and that’s for what I consider a production deployment. You can certainly downgrade that to micro instances and be within Amazon’s free tier quite easily. Even in the AWS free tier you get at least 1GB of ram.

In any case, we are looking at the Heroku setup and there don’t seem to be any issues running the app within the 512mb limit on standard dynos, so we’ll have that fixed up soon … https://github.com/metabase/metabase-deploy/pull/5


#10

Hello! I wasn’t able to launch on Heroku today. I want to be super-lazy, and don’t want to dig into the Heroku CLI to tail the log to figure it out.

It looks like the Heroku free/trial deployment has been problematic.

Is there a Demo instance that refreshes itself every X minutes? That might be better or everyone.

Yeah, Thanks for Metabase, looks superb!