Add Data Out's keyserver to our setup

[dtitov]

It can be enabled using flag --keyserver ega while bootstrapping.

[silverdaz]

What does the data-out keyserver do that the data-in keyserver doesn't ?

[blankdots]

if the profile is no-oss I assume it does not require cega-eureka ?

[dtitov]

Good point. It really can be removed in this setup.

[blankdots]

@silverdaz

What does the data-out keyserver do that the data-in keyserver doesn't ?

We have been trying to align the APIs in EGA-archive/ega-data-api#4 however the data-in keyserver (in my understanding) cannot be used by the data-out (due to some missing endpoints/functionality), thus resulting in a setup where we would have two keyservers. With this PR we are exploring the option where we would have one keyserver, in a similar idea where we would have one database EGA-archive/ega-data-api#10

Another option is to match the functionality of the data-out keyserver in the data-in so that the implementations are interchangeable.

[nanjiangshu]

Hi @dtitov , I've tested with the ega keyserver and it works 👍 But it is not very straightforward to know where the flag "--keyserver ega" to be added before one look again at the Makefile. I would like to add

make "ARGS=--keyserver ega" bootstrap

to the document for bootstrapping if the ega keyserver is going to be used.

[dtitov]

Thanks for testing, I'm glad that it works. I made a flag based on the existing flag for Inbox: --inbox min. But feel free to bring any changes.

[nanjiangshu]

I mean it is fine with the such flags but added explicitly in the README. I will add it. So probably you can approve and merge it?

[blankdots]

see comments. also all Integration tests should pass even the Makefile-based.

[blankdots]

flag here as well: if [[ $KEYSERVER == 'ega' ]]; then

[silverdaz]

I think the better approach is to update the current keyserver with the missing endpoints.

[jrambla]

Hi gentlemen, I'm confused that we are having non-coherent Key servers, as this task was 'done' in a previous sprint, IMU... Could someone clarify which is the situation? Again, such changes should be discussed at the Sprint review/planning meeting. I consider switching components a major change and would like to discuss it in depth. Best! Obtén l'aplicació Outlook per a l'iOS<https://aka.ms/o0ukef>

…

________________________________ De: Frédéric Haziza <[email protected]> Data d'enviament: dilluns, de setembre 17, 2018 2:38 p. m. Per a: NBISweden/LocalEGA A/c: Subscribed Assumpte: Re: [NBISweden/LocalEGA] Add Data Out's keyserver to our setup (#355) I think the better approach is to update the current keyserver with the missing endpoints. — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub<#355 (comment)>, or mute the thread<https://github.com/notifications/unsubscribe-auth/AHsiOiT6sXrkb98FMfyTyuB-82IGVRysks5ub5fVgaJpZM4WpQyj>.

[blankdots]

@jrambla
This PR is part of a sprint goal we set (and agreed, for sprint 36): "Start stitching a general docker-compose for LocalEGA (Data Ingestion and Data API)" and it constitutes work towards achieving that goal.

[blankdots]

It would be nice to have the documentation attached to it, instead of doing it in another PR: #356

[blankdots]

Reviewed and ega data api keyserver can be used.
With this PR we are providing the option to bootstrap with it.
Not a subject of this PR, but worth mentioning:

• Regarding the integration tests on travis, increasing sleep is not a good strategy. Waiting for the services to be up in the integration tests sounds like a more viable solution.*
• ega-data-api keyserver should have a health endpoint