Create Syndesis Standalone

[KurtStam]

This would be a downloadable zip file that people can download and then startup as a standalone Java process. It would run all design time components. It would facilitate the adoption in the Community as it would be much simpler to get going.

WIP: https://github.com/KurtStam/syndesis/tree/standalone/app/standalone

[KurtStam]

• Take care of authentication running w/o auth proxy, setting the headers

  • https://github.com/syndesisio/syndesis/blob/master/app/ui-react/syndesis/src/setupProxy.js#L37-L38

  • https://github.com/syndesisio/syndesis/blob/master/app/server/runtime/src/main/java/io/syndesis/server/runtime/SecurityConfiguration.java

• Switching to H2 DB with PostgreSQL compatible mode
• Adding correct application.yml configuration to both server & meta process
• Have the server manage the meta process (startup/restart/shutdown)
• No metrics
• Noop for OpenShift calls
• Run on random ports?

[lburgazzoli]

I feel this issue need a little bit more work in term of defining the scope, the success criteria and some UX consideration as to me those are not really clear i.e.:

• a user looking for a designer would probably go for something simpler, like a VSCode extension and the possibility to store its integration files and projects in a source core repository (see bpmn-vscode-extension) with the ability to deploy such integration straight to k8s
• a user willing to test Syndesis would need to have a k8s cluster somewhere in any case to run the integrations so it would probably better to invest in an easier way to install Syndesis on k8s than having a partial platform running off-line
• as a developer I would certainly not download a zip as a getting starter experience to hack on Syndesis if the goal of this effort is also what described here

Another relevant point is about the constraints that making Syndesis working also "offline" would put to the evolution of the platform itself i.e. if we want to leverage camel-k more and more as example to materialize the meta service (#7686), then the need to make it running offline would require to implement some optional bits that would increase the maintenance costs.

[KurtStam]

@lburgazzoli. This feature has been deemed critical for the Syndesis community efforts. This work will have no direct relance to the product relevance, but it will allow people to onboard with Syndesis w/o having to install cloud infrastructure. Success is defined by being able to have the UI be functional at a bare minimum level. The Syndesis community agreed on this. You're welcome to join the community meetings.

[Delawen]

Duplicates #6200 ?

[lburgazzoli]

don't think so, this issue is about running without k8s at all

[Delawen]

Oh, right! it is even more decoupled. Good.

[kevinmershon]

+💯 for this. Syndesis looks quite promising but requiring it to be deployed on OpenShift is greatly holding it back from adoption.

As a non-devops engineer looking to trial Syndesis on GCP for a replacement/upgrade for an ETL pipeline, I ran into so many problems trying to set up Syndesis on our cloud that I've recommended against Syndesis in our org. I wanted to put together a 5-minute realworld demonstration of an FTP scrape to DynamoDB pipeline, but I would have needed to run it in GCP because of our infrastructure firewalls. But I couldn't, due to a number of issues.

Firstly, after 3 days of researching and trial/error, I was finally able to sort out that Syndesis must run on OpenShift in a production cloud environment, and OpenShift requires a multitude of VMs to run a cluster AND a paid OpenShift license to Red Hat on top of cloud server costs. This is significantly more expensive than our current ETL pipeline on Lambda, and takes away from the main selling point of Syndesis vs. industry IPaaS software like Dell Boomi, which is the perceived price point. Syndesis is touted as free everywhere, but it's not if you have to pay for OpenShift.

Granted, Syndesis is more featureful, so I did press ahead with the 60 day free trial, and attempted to install Syndesis both from the OperatorHub CRD and separately from my checked-out repository, against both OpenShift 4.3 and 4.2. In all cases, Syndesis failed to start up completely due to underlying issues with syndesis-db, and the interface in OpenShift was so unstable it would never let me view logs. I found open tickets on RedHat.org and here on the Issues list, months old, and unanswered, detailing this exact issue. Not a good look.

In the end we are choosing to not go with Syndesis and OpenShift, and instead will be using Apache Nifi which is actually free, and much more lightweight.

[stale]

This issue has been automatically marked as stale because it has not had any activity since 90 days. It will be closed if no further activity occurs within 7 days. Thank you for your contributions!