Re: [DISCUSS] Hop Server configuration

[Bart Maertens <ba...@know.bi>]

Hi Matt,

We could have a server that can run in one of two modes: "standalone" or
"dynamic". In standalone mode, we could use a startup option to specify a
project/environment. Ideally, we'd have a rest call to switch between
standalone/dynamic mode, specify a project etc.
In dynamic mode, we could pick up HOP_CONFIG_FOLDER and accept
projects/environments from the clients. If a client sends a project that
already exists locally on the server, we could add a sequence to the
project name, similar to what currently happens to an action or transform
name.

Both modes could have their use cases (e.g. dynamic in dev, standalone in
prd), but there probably will be use cases (uat environments), where mode
switching is required, e.g. dynamic by default, standalone for a limited
period of time during a testing phase. A server restart to change modes
would not always be desirable or possible in those scenarios.

Regards,
Bart

On Sun, Feb 28, 2021 at 12:31 PM Matt Casters
<ma...@neo4j.com.invalid> wrote:

> Dear Hoppiasts,
>
> It was only a matter of time before someone asked about services on Hop
> Server.
> This week that time came and I implemented HOP-2577
> <https://issues.apache.org/jira/browse/HOP-2577> which went much faster
> than I anticipated.  It's just so easy to add more metadata and plug in new
> code.
> For those interested, the documentation is here
> <http://hop.apache.org/manual/latest/hop-server/web-service.html>.
>
> What was uncovered was the need to add configuration details to the Hop
> Server.
> Up until now it's been designed to be doing just fine on its own.  The
> metadata and variables the server needs it receives from clients like Hop
> GUI and Hop Run that want to run something on it.
>
> However, now that we're deploying stand-alond services on Hop Server we
> might want to consider optional configuration options.  For now I added an
> option called metadata_folder which allows you to add metadata specifically
> for the "Web Service" service.
> I would love to come up with something more generic and with a wider
> user-case.
> Should we allow the server to cater for one specific environment (project)
> or multiples? Should we throw errors if a client runs a pipeline from
> project "A" on a server configured for project "B"?
> Too many questions so I would love your input.
>
> The thing is that it won't be long before someone asks about porting the
> JDBC driver to a pipeline and then we might as well have a more general
> answer.
>
> Cheers,
> Matt
> --
> Neo4j Chief Solutions Architect
> *✉   *matt.casters@neo4j.com
> ☎  +32486972937
>