#meeting:fedoraproject.org log

<@humaton:fedora.im>

14:02:14

!startmeeting Git Forge Meeting

<@meetbot:fedora.im>

14:02:15

Meeting started at 2025-01-22 14:02:14 UTC

<@meetbot:fedora.im>

14:02:16

The Meeting name is 'Git Forge Meeting'

<@amoloney:fedora.im>

14:03:02

!hi

<@zodbot:fedora.im>

14:03:03

Aoife Moloney (amoloney)

<@dherrera:fedora.im>

14:03:07

!hi

<@zodbot:fedora.im>

14:03:08

Diego Herrera (dherrera) - he / him / his

<@humaton:fedora.im>

14:03:08

!info this is meeting about the migration effort of Fedora gitforge

<@t0xic0der:fedora.im>

14:03:12

!hi

<@zodbot:fedora.im>

14:03:13

Akashdeep Dhar (t0xic0der) - he / him / his

<@humaton:fedora.im>

14:03:16

!topic init

<@ekidney:matrix.org>

14:03:17

!hi

<@zodbot:fedora.im>

14:03:20

emma kidney (ekidney) - she / her / hers

<@phsmoura:fedora.im>

14:03:23

!hi

<@zodbot:fedora.im>

14:03:24

Pedro Moura (phsmoura) - he / him / his

<@t0xic0der:fedora.im>

14:03:41

!link https://docs.google.com/document/d/1DpiOvGbqcVrrafKmt9B1-bPrMthqoA0H31p-HzG0ywA/edit?usp=sharing

<@patrikp:matrix.org>

14:04:04

Hi. 👋

<@humaton:fedora.im>

14:04:10

!hi

<@zodbot:fedora.im>

14:04:12

Tomáš Hrčka (humaton) - he / him / his

<@t0xic0der:fedora.im>

14:04:27

I wonder if we should retain meeting notes separately or would the meeting summary/logs from the meetbot be sufficient. Thoughts?

<@humaton:fedora.im>

14:04:29

Can't access item

<@humaton:fedora.im>

14:04:45

notes will be kept by the bot here wont they?

<@amoloney:fedora.im>

14:04:57

do we need the doc? The bot should do just fone

<@amoloney:fedora.im>

14:05:01

*fine

<@t0xic0der:fedora.im>

14:05:24

Cool - Lets stick to the meetbot logs/summary then - I have added you to the doc though jednorozec

<@humaton:fedora.im>

14:05:38

we have a repo for the meeting structure and org stuff https://codeberg.org/fedora/gitforge-migration/src/branch/main/meeting.md

<@farchord:fedora.im>

14:05:59

Yeah maybe not a good idea to use a google doc that's locked out XD

<@lenkaseg:fedora.im>

14:06:26

!hi

<@zodbot:fedora.im>

14:06:28

Lenka Segura (lenkaseg)

<@humaton:fedora.im>

14:06:43

we will have meeting notes if there is structure to the meeting.

<@mathstuf:matrix.org>

14:06:53

!hi

<@zodbot:fedora.im>

14:06:54

No Fedora Accounts users have the @mathstuf:matrix.org Matrix Account defined

<@mathstuf:matrix.org>

14:07:13

hrm…will have to connect that up i guess

<@humaton:fedora.im>

14:07:41

!link https://codeberg.org/fedora

<@humaton:fedora.im>

14:08:11

there is a organization on Codeberg with repository for the meeting template

<@humaton:fedora.im>

14:08:23

feel free to propose any changes/content

<@humaton:fedora.im>

14:08:51

lets move on

<@humaton:fedora.im>

14:08:52

!topic Tickets and Updates

<@humaton:fedora.im>

14:09:02

!link https://pagure.io/fedora-infra/forgejo-deployment/issues

<@mathstuf:matrix.org>

14:09:09

!hi

<@zodbot:fedora.im>

14:09:11

None (mathstuf)

<@humaton:fedora.im>

14:09:12

We have few Epics to look at

<@Zlopez:matrix.org>

14:10:22

!hi

<@humaton:fedora.im>

14:10:23

!link https://pagure.io/fedora-infra/forgejo-deployment/issue/1

<@zodbot:fedora.im>

14:10:26

Michal Konecny (zlopez)

<@humaton:fedora.im>

14:11:01

SO we need to build or reuse iamges for our FOrgejo deployment

<@humaton:fedora.im>

14:11:14

SO we need to build or reuse images for our Forgejo deployment

<@humaton:fedora.im>

14:11:57

I was thinking about using konflux to do that, there is a demo pipeline in the ticket

<@humaton:fedora.im>

14:12:17

naming is up for change I just wanted to get thing flowing from codeberg->konflux

<@humaton:fedora.im>

14:12:48

WDYT? how/where should we get our images for the deployment

<@Zlopez:matrix.org>

14:13:43

Could we use Fedora images as a base for Forgejo

<@Zlopez:matrix.org>

14:13:45

<@t0xic0der:fedora.im>

14:13:56

Sounds good to me. Gives us a chance to take Konflux for a spin as well.

<@thisisyaash:fedora.im>

14:14:06

!hi

<@zodbot:fedora.im>

14:14:06

Yashwanth Rathakrishnan (thisisyaash)

<@humaton:fedora.im>

14:14:14

Well what do you mean?

<@humaton:fedora.im>

14:14:31

the demo uses rawhide, so for sure we want to use fedora as base image

<@humaton:fedora.im>

14:14:44

but we need postgres image we need nginx for proxies

<@humaton:fedora.im>

14:14:49

we need a lot of service images

<@Zlopez:matrix.org>

14:14:55

That answers my question

<@dkirwan:fedora.im>

14:15:10

If its open to discussion, might be better to use rhel imo

<@humaton:fedora.im>

14:15:36

sure why? what? how?

<@dkirwan:fedora.im>

14:15:48

Much longer support cycles

<@humaton:fedora.im>

14:16:02

usinig UBI images and building everything? Usuing the sclog images linked in the issue?

<@dkirwan:fedora.im>

14:16:03

production grade

<@humaton:fedora.im>

14:16:29

that is not real answer

<@humaton:fedora.im>

14:16:43

We need to setup postgres cluster

<@Zlopez:matrix.org>

14:17:06

If the forgejo is not built for RHEL, it doesn't make much sense to use it

<@Zlopez:matrix.org>

14:17:30

Could we use existing postgres db we have in Fedora Infra?

<@humaton:fedora.im>

14:17:36

its a golang application it can be build on anything...

<@t0xic0der:fedora.im>

14:17:36

Or we'd have to put in extra effort to package em

<@t0xic0der:fedora.im>

14:17:36

These dependencies can be difficult to find in RHEL https://codeberg.org/fedora/oci-image-definitions/src/branch/main/forgejo/Dockerfile#L3

<@t0xic0der:fedora.im>

14:17:55

Which is prolly not going to be worth it - but I understand where David Kirwan is coming from

<@Zlopez:matrix.org>

14:17:56

Or do we want to have separate postgres for Forgejo?

<@humaton:fedora.im>

14:18:27

clang and llvm are available in the appstream

<@humaton:fedora.im>

14:19:26

Yes we do want to have a postgres cluster deployed in openshift for forgejo

<@humaton:fedora.im>

14:19:37

1 or 2 replication hosts

<@t0xic0der:fedora.im>

14:19:55

jednorozec: Is it worth exploring if RHEL can be used or are we sticking to Fedora base image?

<@humaton:fedora.im>

14:20:00

if you look at the help deployment I mentioned last week there is a section about it

<@Zlopez:matrix.org>

14:20:29

I wasn't here last week, so I probably missed that

<@t0xic0der:fedora.im>

14:21:08

The notes were shared but I will forward those to you - No problems at all

<@humaton:fedora.im>

14:21:09

sure it is. If we have supported rhel image. But I think we will still need to put layers on top of it with our configuration os postgres

<@t0xic0der:fedora.im>

14:21:41

Oooh - natively installed Postgres? Not postgres on a different container? Interesting.

<@Zlopez:matrix.org>

14:21:42

I will find them somewhere in e-mail

<@humaton:fedora.im>

14:21:43

if you look at the helm deployment I mentioned last week there is a section about it

<@humaton:fedora.im>

14:22:10

!link https://code.forgejo.org/forgejo-helm/forgejo-helm#configuration

<@humaton:fedora.im>

14:22:39

helm configuration referencing the deployment of code.forgejo.org

<@humaton:fedora.im>

14:23:28

this is the postgres configuration they are running https://github.com/bitnami/charts/tree/main/bitnami/postgresql-ha

<@dkirwan:fedora.im>

14:24:51

AWS RDS with a postgresql engine could be explored also ?

<@Zlopez:matrix.org>

14:25:32

I don't think we want to host this on AWS, but rather in our own infrastructure

<@humaton:fedora.im>

14:29:37

David Kirwan: well yes but I have problem that we are then exposed to cost of AWS if we would ever lose sponsorship from Amazon

<@humaton:fedora.im>

14:30:08

if we deploy it in openshift based on rhel9 the cost is the cost of operating the cluster

<@humaton:fedora.im>

14:31:11

there may already be replicated cluster template in openshift

<@dkirwan:fedora.im>

14:31:52

I'd imagine theres a few supported ways. might be an operator even

<@humaton:fedora.im>

14:32:39

we will need few more services

<@humaton:fedora.im>

14:32:53

so that os PostgreSQL, Redis, Opensearch

<@humaton:fedora.im>

14:33:39

we need to build from source on the Forgejo container image

<@humaton:fedora.im>

14:33:50

we need to build from source only the Forgejo container image

<@humaton:fedora.im>

14:34:28

so that is PostgreSQL, Redis, Opensearch

<@humaton:fedora.im>

14:36:46

but i think we should do something else first

<@humaton:fedora.im>

14:37:33

!topic Openshift app space

<@humaton:fedora.im>

14:37:37

!link https://pagure.io/fedora-infra/forgejo-deployment/issue/2

<@humaton:fedora.im>

14:38:03

we need someone to define the openshift app forgejo for staging and prod cluster

<@humaton:fedora.im>

14:38:50

and apply the changes so we can delegate access to the staging app to other people

<@gwmngilfen:fedora.im>

14:40:28

i don't know enough yet to volunteer for that, but I'd like to follow along for sure

<@lenkaseg:fedora.im>

14:40:53

+ 1, I'd like to follow along here too

<@lenkaseg:fedora.im>

14:41:18

//+1, I'd like to follow along here too

<@gwmngilfen:fedora.im>

14:41:35

(plus FOSDEM is going to occupy the next 2 weeks)

<@gwmngilfen:fedora.im>

14:42:55

i guess my shortcoming is knowledge of our openshift clusters, but I'd be happy to spin up minikube locally and at least test the helm chart - no idea if that's actually useful though?

<@dkirwan:fedora.im>

14:43:07

Are we planning to deploy via Helm?

<@dkirwan:fedora.im>

14:43:11

or we're not sure yet ?

<@humaton:fedora.im>

14:44:26

We discussed it last week on infra weekly, and Kevin had a good note. We should not be afraid to deploy new things and try different approaches

<@amoloney:fedora.im>

14:44:30

I would suggest using the !proposal or even !agreed to note the plan for deployment, even if its not fully agreed here today as it can be sent out post-meeting for wider feedback

<@humaton:fedora.im>

14:44:39

so I \think we should try to make the helm deployment happen

<@humaton:fedora.im>

14:45:09

Using helm means it can be running by the end of the week But with upstream as base images

<@humaton:fedora.im>

14:45:30

I dont care for staging and it even might be a good idea to do that

<@dkirwan:fedora.im>

14:45:39

So a minimal ansible playbook/role to launch helm...

<@humaton:fedora.im>

14:45:54

But we definately need a playbook https://pagure.io/fedora-infra/ansible/blob/main/f/playbooks/openshift-apps that will define the space for us

<@dkirwan:fedora.im>

14:46:35

We'll need helm configured and available on the os-control and os-control.stg machines then also

<@humaton:fedora.im>

14:46:44

yup, we can configure the heml by variables to some extend. Maybe if replacing its images with services based on openshift operators?

<@humaton:fedora.im>

14:47:09

Does it require some extra work?

<@humaton:fedora.im>

14:47:50

My idea was to fork the forgejo helm under fedora org and do some changes there if needed.

<@dkirwan:fedora.im>

14:47:52

Can be installed manually just to get running, but ideally it also needs to be added to the ansible role managing the os-control systems

<@humaton:fedora.im>

14:48:06

right

<@dkirwan:fedora.im>

14:48:24

might be easy as adding helm to the package install tasks vOv

<@humaton:fedora.im>

14:49:13

David Kirwan: so since you know what need to be done for helm to be available and for the openshift playbook to work.

<@humaton:fedora.im>

14:49:41

can you create a Ticket for it under fedora-infra and link it in comment https://pagure.io/fedora-infra/forgejo-deployment/issue/2

<@dkirwan:fedora.im>

14:49:58

Yeah i can take that on, maybe lenkaseg and Gwmngilfen wants to join ? anyone else interested too of course

<@lenkaseg:fedora.im>

14:50:10

I'm in!

<@gwmngilfen:fedora.im>

14:50:12

i'll certianly follow along

<@gwmngilfen:fedora.im>

14:50:15

thanks!

<@humaton:fedora.im>

14:50:23

!agreed David will create tracking issue and ping interested parties

<@t0xic0der:fedora.im>

14:50:23

Let me know how I can be of help as well

<@humaton:fedora.im>

14:50:51

hum

<@humaton:fedora.im>

14:50:57

we have 10 mins

<@humaton:fedora.im>

14:51:04

but we have progress!

<@t0xic0der:fedora.im>

14:51:25

Wanna talk about the Pagure Exporter port to Forgejo or should we keep it for the next meetings?

<@humaton:fedora.im>

14:51:55

!topic Q&A

<@humaton:fedora.im>

14:52:08

ok so exporter Akashdeep Dhar ?

<@t0xic0der:fedora.im>

14:52:27

If we have enough time - Or else, Q&A is fine

<@humaton:fedora.im>

14:52:47

Well the meeting has no structure set in stone

<@humaton:fedora.im>

14:52:49

but

<@humaton:fedora.im>

14:53:09

talking about exporter is interesting since we still dont have the service to import it into...

<@t0xic0der:fedora.im>

14:53:17

<@t0xic0der:fedora.im>

14:53:39

Very early ideation but folks can find the initiative proposal here

<@t0xic0der:fedora.im>

14:53:41

!link https://pagure.io/cpe/initiatives-proposal/issue/32

<@t0xic0der:fedora.im>

14:53:53

Lets wait until we have some deployment ready before I hammer it with the exports

<@t0xic0der:fedora.im>

14:54:00

Fin

<@amoloney:fedora.im>

14:54:46

if you have a few mins, could someone summarise the intended plan for deployment please?

<@amoloney:fedora.im>

14:54:55

for the notes, wont someone please think of the notes :D

<@t0xic0der:fedora.im>

14:57:04

Not sure of the summary here but I can forward the meeting summary/logs to Discussion FPO thread to funnel people in

<@t0xic0der:fedora.im>

14:57:17

We can use all the visibility and assistance we can get

<@amoloney:fedora.im>

14:58:10

Ideally there would be a !proposal: Helm will be used as the deployment mechanism for Forgejo. We will use a base image from Fedora Rawhide and the managament of Helm will be configured through opeshift etc

<@amoloney:fedora.im>

14:58:21

I cant do this part, I dont have the technical fluency

<@amoloney:fedora.im>

14:59:10

but it would be helpful to have an intention stated for the meeting that 'this is what were doing' so if someone reads the logs and thinks of a way to help/do it better, they can engage with us to improve things

<@humaton:fedora.im>

15:00:08

I will put your feedback into the meeting template Aoife Moloney

<@humaton:fedora.im>

15:00:18

so we have better structure next week

<@humaton:fedora.im>

15:00:40

and we are at the hour

<@amoloney:fedora.im>

15:00:51

thank you! Ive seen it done in other meetings and its very helpful to just jump straight to that outcome, and then re-read the conversation if needed

<@humaton:fedora.im>

15:00:56

!endmeeting