As I announced in a post on the mailing list, we are having a kickoff meeting for a new Containerization Working Group in a couple weeks on March 21st. Come one, come all! Please bring your passion for Docker and all things containers! :happy:
I'm planning on suggesting that we use this #containers stream to communicate between meetings.
There's no need to wait for the kickoff meeting. Please go ahead and create a new topic in the #containers stream if you like.
In this "kickoff meeting" topic, please feel free to RSVP, or suggest something you'd like to discuss at the kickoff meeting. Oh and please spread the word! Thanks! :rocket:
Hopefully people saw my email but I just had a thought: I could send another email asking people to RSVP by adding their names to the bottom of the agenda/notes doc for the kickoff meeting: https://docs.google.com/document/d/1r_GMYqYzUyQR2LQ2cqSLWPDuFRzd18PJ-4UFC4Nvnpw/edit?usp=sharing
People are used to me encouraging them to do this at the start of the meeting. If we encourage this earlier, after some time, we can reach out to people who we think would be interested but who may have missed the announcement.
I just added this to the doc (feedback welcome):
Screen-Shot-2023-03-10-at-11.25.59-AM.png
I did go ahead and ask people to RSVP to the kickoff: https://groups.google.com/g/dataverse-community/c/U2LHWoNN-X8/m/WLSZw1NVCAAJ
19 RSVPs so far! Not bad!
In that post I also asked people to give feedback on the agenda, if they are so moved. :happy:
Oh, interesting, in the doc under "website" there's a bullet from @Slava Tykhonov :
"(Slava) is this group only about βvanillaβ Dataverse or infrastructure built on top of Dataverse (Traefik, Minio, microservices)?"
I would say Minio is on topic. Right?
Philip Durbin said:
I would say Minio is on topic. Right?
I find it somewhat odd to say "built on top", but yes, Minio is certainly a related product. We probably should also include SeaweedFS, but probably Slava simply meant "are we focusing on the core or also taking into account the larger context"
At the very least, we could do some automated testing of S3 code using Minio. I think we've talked about this a lot.
I was going to write that just now. :-D
Microservices, yes. Do we want Traefik?
Well I dunno what he is referring to when talking about "Microservices". This is an architectural pattern, so might be anything. Certainly the core of Dataverse is not built of microservices. I have reluctance about making a rearchitecturing of Dataverse to microservices a topic for this working group. But certainly we could speak about how other components are better integratable with Dataverse when running in containers, e.g. with adding OIDC API access etc
@Slava Tykhonov is a big fan of running services along side Dataverse, from what I understand. I doubt he's talking about the core.
WRT Traefik: this is a reverse proxy. There are a lot of reverse proxies out there and it highly depends on your cloud context what you might want / need to use. So maybe we could include the general topic and also add docs about how to use Traefik with things. Not sure when things start to be out of scope. Adding docs about how to create good working reverse proxies for Dataverse certainly is in scope, but then we probably need to add HAProxy, Traefik, NGINX, Apache and some more cloud vendor specifics. Maybe it helps to provide general docs and have some example configs around.
Right. Similar to HAProxy, I think.
Back to microservices, for a moment, what about RServe? It could be considered a microservice, right?
Can you point me to your definition of microservices please? I seem to have a very different understanding of the term.
Anyway, for now I'm mostly thinking of services that run along side Dataverse that may not be strictly required, if you don't need that functionality.
Maybe I have an odd understanding of microservices. :sweat_smile:
Let's ask Slava what he means.
There is no single definition for microservices.
https://en.wikipedia.org/wiki/Microservices
I find it hard to speak of Dataverse as a microservice. But that's maybe just me.
I agree.
Rserve is a service that Dataverse uses to accomplish certain tasks. It is not an integral part of Dataverse, we don't control it (we're only reusing it), it's not integrated in terms of AA(A) and it's not of very small size. What's left is that we communicate with it over a lightweight protocol, it's loosly coupled and that it kind of only serves one purpose. Personally I find it hard to speak of "microservices" here. I would agree though if someone speaks of a SoA.
"Optional service if you want R stuff" :happy:
Just in time for tomorrow, we got the first iteration of app image merged, the GDCC is sponsored at Docker Inc and we're really to blow some steam! Really looking forward to tomorrow's meeting!
Me too. Should be fun.
Phew! Good meeting, I thought, but I need some lunch and a walk outside. Any thoughts on the meeting? What worked well? What didn't work so well? :sweat_smile:
I just summarized the kickoff meeting and posted a link to the recording: https://groups.google.com/g/dataverse-community/c/3vJWcAK5F7M/m/h63JizJKBgAJ
Philip Durbin has marked this topic as resolved.
From the people that opted in to be part of the WG by putting an asterisk to their name in the list, only Peter Pallinger is missing from the #containers stream. @BalΓ‘zs Pataki could you nudge him to join us here?
Thanks for the reminder, he will join soon!
Last updated: Oct 30 2025 at 05:14 UTC