Google Compute Platform
I've got an Angular (2) app and a Node.js middleware (Loopback) running as Services in an App Engine in a project.
For the database, we have a Compute Engine running PostgreSQL in that same project.
What we want
The testing has gone well, and we now want to have a test version (for ongoing upgrade testing/demo/etc) and a release deployment that is more stable for our initial internal clients.
We are going to use a different database in psql for the release version, but could use the same server for our test and deployed apps.
Should we....?
create another GCP project and another gcloud setup on my local box to deploy to that new project for our release deployment,
or is it better to deploy multiple versions of the services to the single project with different prefixes - and how do I do that?
Cost is a big concern for our little nonprofit. :)
My recommendation is the following:
Create two projects, one for each database instance. You can mess around all you want in the test project, and don't have to worry about messing up your prod deployment. You would need to store your database credentials securely somewhere. A possible solution is to use Google Cloud Project Metadata, so your code can stay the same between projects.
When you are ready to deploy to production, I would recommend deploying a new version of your App Engine app in the production project, but not promoting it to the default.
gcloud app deploy --no-promote
This means customers will still go to the old version, but the new version will be deployed so you can make sure everything is working. After that, you can slowly (or quickly) move traffic over to the new version.
At about 8:45 into this video, traffic splitting is demoed:
https://vimeo.com/180426390
Also, I would recommend aggressively shutting down unused App Engine Flexible deployments to save costs. You can read more here.
Related
We have an older application based on the AppEngine SDK (now deprecated) and the superseded Cloud Datastore.
In the process of migrating to Google Cloud SDK we also decided to move from JPA/Datanucleus to Objectify.
Given that the Cloud Datastore will be automatically upgraded to Cloud Firestore in Datastore mode sometime in the future, we decided to test our application as described at the bottom of this page:
https://cloud.google.com/datastore/docs/upgrade-to-firestore#testing_an_existing_application
1) Create a new project. In this project, create a Cloud Firestore in Datastore mode database.
2) Using the managed export service, export some of your application's data to Cloud Storage.
3) Using the managed import service, import your application's data to your new project.
4) Copy app logic you want to test to the new project or simulate app behaviour against the new project.
That's what we did and after some issues we could make a portion of our application run fine with the new datastore in a separate test project.
Now to the actual issue...
We wanted to test if the updated application could also run with the superseded Cloud Datastore, so we won't have to worry when the automatic upgrade occurs (as our app will be already ready).
So we deployed it as a new version of the existing AppEngine project (v2-dot-.....): unfortunately running the new version throws a permission error as soon as the app tries to read the datastore:
com.google.cloud.datastore.DatastoreException: Missing or insufficient permissions
So the questions are:
- could this be related to the Cloud Datastore not being upgraded to Cloud Firestore in Datastore mode for our project yet?
- is there anything we can do (add specific permissions maybe) to make it work anyway?
Our concern is that we need to have the new version of the app deployed before July 2020 (that's when the older AppEngine SDK will stop working), and we are worried that the automatic upgrade of the datastore will occur later.
Thank you for your help.
Turned out our project doesn't have the PROJECT_ID#appspot.gserviceaccount.com member in IAM.
It has a pletora of other members (for example: PROJECT_NUMBER-compute#developer.gserviceaccount.com, PROJECT_NUMBER#cloudservices.gserviceaccount.com, service-PROJECT_NUMBER#compute-system.iam.gserviceaccount.com, etc) which I guess are legacy members used in previous versions of Google App Engine.
Adding PROJECT_ID#appspot.gserviceaccount.com with the role Editor fixed the issue: now the new version can be deployed to the old projects and it works fine even if the datastore has not yet been converted to Cloud Firestore in Datastore Mode.
I just ran into this issue and spent way too much time troubleshooting it. Nine times out of ten if you're running into this issue it's because the default App Engine service account doesn't have permission to edit Cloud Datastore. The default App Engine service account is used by default if you're doing a simple gcloud app deploy and nothing else fancy. I solved the problem by giving the default App Engine service account the roles/datastore.owner role with the following:
gcloud projects add-iam-policy-binding PROJECT_ID \
--member="serviceAccount:PROJECT_ID#appspot.gserviceaccount.com" \
--role="roles/datastore.owner"
I've read about partial upgrade, but it always requires to change some parts of the packages application. I'd like to know if there's a way to redeploy a package without version change. In a way, similar to what VS is doing when deploying to the dev cluster.
On your local dev cluster, VS simply deletes the application before it starts the re-deployment. You could do the same in your production cluster, however this results in downtime since the application is not accessible during that time.
What's the reason why you wouldn't want to use the regular monitored upgrade? It has many advantages, like automatic rollbacks and so on.
I'm trying to learn about Heroku's deployment capabilities. I've used Google App Engine in the past and have found its support for multiple simultaneously-accessible live application versions useful, for example, in order to test a new version in the production context before making it the default, running a specially-instrumented version in parallel with the default version to debug a problem specific to the production environment (e.g. production data), for A/B testing, traffic splitting, gradual rollout, etc.
Does Heroku have support for this kind of thing? My Googling has not turned up anything.
No, it doesn't offer support for this right now.
I use Glassfish Java, and JSP over MySQL for my web applications. Many online people uses this web application and that web-site should not be down.
When I want to deploy a new war file, I should undeploy and deploy the new one for my application at server.
My question is that;
Is there any technology that doesn't need to undeploy my application and just change the appropriate classes so no need to redoploy it again?
There are java technologies that would allow you to replace classes on the fly (like JRebel). But since you're using Glassfish already, you should just start using clustering which is built into glassfish. You'll need either 2.1 or 3.1, as 3.0 does not support clustering. With a Glassfish cluster, you have a load balancer (Apache, Sun Web Server, hardware (Big IP, Coyote), etc) distribute the load among your cluster nodes. When you want to upgrade the app, you can technically do it one node at a time. Setting up the cluster is not the easiest thing in the world, but it is doable and it would get you some great benefits. You'll be able to scale the load by adding new hardware and even using Amazon (or whoever) cloud services. You'll be able to keep your site running even if the hardware fails on one of the nodes.
Personally I'm in the middle of converting from Glassfish 2.1 to 3.1. So far I like the management of the Glassfish 3.1 cluster much better, but I can't personally vouch for how it will run in production, though I have high expectations.
http://download.oracle.com/docs/cd/E18930_01/html/821-2432/gktqx.html#gktob
Jim is right, the best solution is currently to use a cluster and perform a manual rolling-upgrade.
But there is actually work ongoing to address your needs. We are working on a rolling-upgrade feature in a single standalone instance. To sum up in a nutshell (as the specifications have not been published yet), it will let you switch from an application version to another (see application versioning and the enable command) with no downtime. Stay tuned.
I love Heroku but I would prefer to develop in Scala rather than Ruby on Rails.
Does anyone know of any services like Heroku that work with Scala?
UPDATE: Heroku now officially supports Scala - see answers below for links
As of October 3rd 2011, Heroku officially supports Scala, Akka and sbt.
http://blog.heroku.com/archives/2011/10/3/scala/
Update
Heroku has just announced support for Java.
Update 2
Heroku has just announced support for Scala
Also
Check out Amazon Elastic Beanstalk.
To deploy Java applications using
Elastic Beanstalk, you simply:
Create your application as you
normally would using any editor or IDE
(e.g. Eclipse).
Package your
deployable code into a standard Java
Web Application Archive (WAR file).
Upload your WAR file to Elastic
Beanstalk using the AWS Management
Console, the AWS Toolkit for Eclipse,
the web service APIs, or the Command
Line Tools.
Deploy your application.
Behind the scenes, Elastic Beanstalk
handles the provisioning of a load
balancer and the deployment of your
WAR file to one or more EC2 instances
running the Apache Tomcat application
server.
Within a few minutes you will
be able to access your application at
a customized URL (e.g.
http://myapp.elasticbeanstalk.com/).
Once an application is running,
Elastic Beanstalk provides several
management features such as:
Easily deploy new application versions
to running environments (or rollback
to a previous version).
Access
built-in CloudWatch monitoring metrics
such as average CPU utilization,
request count, and average latency.
Receive e-mail notifications through
Amazon Simple Notification Service
when application health changes or
application servers are added or
removed.
Access Tomcat server log
files without needing to login to the
application servers.
Quickly restart
the application servers on all EC2
instances with a single command.
Another strong contender is Cloud Foundry. One of the nice features of Cloud Foundry is the ability to have a local version of "the cloud" running on your laptop so you can deploy and test offline.
I started working on the exact same thing as what you said a few weeks ago. I use Lift, which is a great framework and has a lot of potential, on top of Linux chroot environment.
I'm done with a demo version, but Linux chroot is not that stable (nor secure), so I'm now switching to FreeBSD jail on Amazon EC2, and hopefully it'll be done soon.
http://lifthub.net/
There are also other Java hosting environment including VMForce mentioned above.
If you are looking for a custom setup which also has the ease of deployment that heroku offers: http://dotcloud.com. They are invite only right now but I was given access in under three days. I am working on a Lift/MongoDB project there and it works well.
Off the top of my head, only VMForce comes to mind, but its not available yet. This will be a Java-oriented service, so that probably means you'll have to spend a wee bit of time figuring out how to package the app.
For more discussion, there was a debate about this in 2008.
I'm not entirely sure if it's really suitable or not, but people have deployed Scala applications to Google App Engine, for example http://mawson.wordpress.com/2009/04/10/first-steps-with-scala-on-google-app-engine/
Actually you can run scala on heroku right now. You don't believe it?
https://github.com/lstoll/heroku-playframework-scala
I'm not sure the tricks lstoll has used are legit but using the
new cedar platform where you can run custom processes and some
ingenious Gemfile hacking he has managed to bootstrap the Java
play platform into a process. Seems to work as he has a live
site running a test page.
Stax cloud service offers preconfigured lift project skeleton. Also, there is a tutorial on how to deploy lift project to appengine.