Will the new demo infrastructure setup be done by next week in time for the release?
What is the ETA on the following issues?
Kriti,
Regarding Update “Connect to an external database server” docs:
Sorry, but I don’t think I can take this task solo. I’ll
need help from someone else with a better understanding of
the underlying content. It looks like that page has mostly
been written by Anish. He and I could potentially do some
pairing on it, but given my scheduling challenges, I don’t
think it’s worth trying to schedule that to happen before
the v0.1.5 release. At the very least we need to remove
references to MATHESAR_DATABASES
.
I wonder if Anish could handle that on is own. What do you
think Anish?
At a high level, I have some qualms with this page being hard to follow for a reader who doesn’t already understand the context. I would try to say more here (or in the ticket), but it would be hard via async communication. To express my concerns, I’ll need to ask many small and sequentially-dependent clarifying questions to someone with a better understanding of the docs goals behind this page.
Regarding UI points to missing user documentation when deleting a connection
I opened a PR, Temporarily hide link to missing docs page, which just removes the offending UI. I don’t have time to write the docs content before the release, so I added an issue, Improve user docs about Mathesar schemas, to track that work.
I have some other responses and questions to other points in your email but unfortunately don’t have time to write them right now. I’ll try to respond again tonight with more.
Okay, thanks everyone. We’ll cut the release branch on Monday or Tuesday, do QA, and release by Thursday.
Thanks for the update! We’ll wait to cut the release until your work on those issues is merged. Tuesday is our cutoff though, so if you can’t get those fixes merged in by Tuesday, we’ll release without them.
Also, you’ll be doing some QA next week, so please set aside some time for that.
You said:
I have a branch that fixes the problems there. I’ll raise a PR if we want, but the eventual plan is still to get rid of “Demo Mode” in the actual code and handle it on an infrastructure level.
Is there any point to this if Anish is getting the demo infrastructure done this week? I assume we won’t have enough time to delete the “demo mode” code for 0.1.5, even if Anish does get the infrastructure up and running. We can still deploy the demo with demo mode on the new infrastructure, right?
Also, please review / merge Anish’s work ASAP and be on standby on Mon / Tue to deploy the release to the internal server once Sean has cut the release branch. I also made a Basecamp task for you to help Ghislaine with the installation issue she faced.
Thanks for testing installation, and good to know about the QA scripts. We don’t need the QA scripts for this release, but we can use them if they’re ready.
FYI to everyone else: Sean and I discussed the rest of his points on Matrix and we’re good to go.
Does anyone have any further questions or thoughts? Are everyone’s priorities clear?
Thanks!
I’ve been working on cutting the release but I’m stuck and I need help.
First, I’ll give a few updates, because that will help explain where I’m at:
As I’ve been proceeding with the release process, I’ve been updating our docs along the way. My goal with these updates is to gently nudge our process more towards automation by swapping prose for bash commands that the reader can copy-paste.
I just pushed a pretty big commit to the wiki which re-works our release docs. There’s now a top-level Release process page with many different sub-pages. The release process is now broken up into just a small handful of independent steps — each with their own documented procedure.
“Cutting a release” is the step I’m on now. You can see this page is all copy-paste commands now!
I’m stuck
on the docker login
command. (Or, as I run on my machine, sudo docker login
.)
I’m using the credentials in 1Password. I’ve tried it several times, but I see this:
Error response from daemon: Get “https://registry-1.docker.io/v2/”: unauthorized: incorrect username or password
Any advice?
It’s worth noting that, prior to my docs changes, the old docs just said:
Ensure you’re logged into Dockerhub. (Credentials are in 1Password.)
I’m guessing
this means I need to run docker login
,
but I’m not sure. (Although I helped with the previous release,
Pavish still did this part, so the process is new to me.) As a
troubleshooting step, I actually tried logging into https://hub.docker.com/
using our credentials but it said I can’t log in there using an
organization account.
I’m turning in for the night, but I expect to be back at my desk around 13:00 UTC Tuesday.
- github.com/docker/buildx v0.12.1-desktop.4 6996841df2f61988c2794d84d33205368f96c317
4. Use the cloud builder to build and publish our images:docker buildx create --driver cloud mathesar/release-builder
docker buildx build \
-t mathesar/mathesar-prod:0.1.5 \
--builder=cloud-mathesar-release-builder \
--platform=linux/amd64,linux/arm64 \
--push \
--build-arg PYTHON_REQUIREMENTS=requirements-prod.txt .
- The above command will build our prod setup for amd and arm using our cloud builder, tag it as 0.1.5, and publish it to DockerHub.
- The only change from the command present in our release docs is the value of `--builder=`