I'm still trying to learn the whole hierarchy of cluster > pod > node and how they all interact. There are so many docs, it's overwhelming so I just learn them as I need to to solve a current problem.
As far as cluster -> node -> pods, the cluster is made up of nodes, and the k8s scheduler chooses which node to run each pod on based on a set of criteria that includes the memory and cpu limits, but also taints and tolerations.
BTW, you don't need to nuke the whole cluster. From the console, you can actually create a new pool of nodes, or change the number of nodes in the pool (though I don't think you can change the type of nodes). You can even have multiple pools.
I forgot to add the storage-ro scope to a custom cluster and had to figure that out since without storage-ro (which is the default, BTW). Sadly, you cannot change the access scopes on an existing pool so I had to figure out how to make a new pool and delete the old one. If you have some time I can walk you through it tomorrow on a zoom/google meet if you want.
I was hesitant to create new pools in the my existing cluster because I think I tried that before and my $26/month VM became $52.
It's so easy to make a little tweak and dramatically increase cost.
Oh yeah, we got hit with a $230 bill last month because we forgot to shut down a couple small clusters.
If you delete the old pool and choose the right instance sizes you should be fine though. However, none of the providers really make it easy to see your instantaneous costs.
Oh they do, I hadn't noticed that. I'll have to keep an eye on it.
Hmm, is that because of storage, do they give you a break down?
I'm checking it out now
Ahh
That is a really nice feature. I wonder if AWS has something similar. They have a manual calculator, I haven't seen an automated one like the GCP one, but it would be easy to miss in AWS.
I want to start setting up my postgres db and I noticed in
deploy/styles.tsx the test function k8sTestStyle() has a mockDbName and mockDataPath... is this a functioning db I can read and write to? Not sure how to connect to it, or id I have to do a prod deploy for postgres to work.
The postgres db that's deployed by
k8sTestStyle is a fully functioning postgres database that should work just fine for development and testing purposes. BUT, it's not deployed in a way that's suitable for production. It does not create any persistent storage, so when you destroy your deployment, you destroy your data too (just like you'd typically want for dev or testing). That style also is able to pre-load data from a .sql file or files, which is what mockDataPath is.
Right now, for more long-term data and definitely for production, we'd recommend using a database service from one of the major vendors, like Cloud SQL from Google. To use that, you'd create that database service separate from Adapt. (In the future, we'd like to add components for that.) Then there's an example in the Adapt styles.tsx of a Postgres component that just takes the connection & login information for the prod style. If you need some help with that component, I can go into more detail.
I created a PostgreSQL instance in my gcloud project. I'm not too worried about having separate testing and prod databases, so I'll just use the one for everything. I'm going to spend a few days doing some Lynda (now Linked In Learning) tutorials on how to connect to it from the app. I also want to learn how to use .env file to store the username and password so they are safe. Once I know how to create, edit and delete records, I'll be at the first rung of the ladder of knowledge :) I installed pgAdmin on my PC so I'll mess with that and see if I can connect, then try to move some of those settings over to the app.
also, i searched in adaptjs.org but i couldn't find any documentation about canary deployment using adaptjs..
However, the adapt directory that is referenced in the error has to be preserved across runs, either by copying it to and from GCP, or using some other mechanism. Otherwise adapt update will fail.
In theory, if everything is working you should be able to rerun adapt run from scratch and adapt should adopt existing infrastructure it created, but we don't extensively test this case, and it is easy to have subtle problems with components that cause issues with this.
So, my suggestion with cloud build is to run a step where you copy /builder/home/.local/share/adapt to GCP after a build, and from GCP before a build. But you have to get the state there first, by doing an adapt run.
We do have plans to make this easier in the future but have been focusing on some other projects since we've worked around this issue ourselves.
If you cook up a solution, I'd love to see your cloudbuild.yaml so others can benefit from it.
Let me know if this helps.