Update SWOT image to latest pangeo image

[choldgraf]

cc'ing @yuvipanda so he can confirm that this is how the Pangeo hub image updates should work.

Also - should we suggest that the SWOT team try the configurator?

[yuvipanda]

We should definitely ask them to try out the configurator! That frees us from having to do this kinda work. The configurator needs some work though.

#569 will enable automatic deployment for the SWOT (aka meom-ige) hub. After that is merged, this can be merged. Regardless of what happens with the configurator, we should get this merged

[sgibson91]

We merged #582 instead, so merging this PR should only trigger a deploy for the meom cluster. Actually, this PR is updating the 2i2c.cluster.yaml file. But @yuvipanda you thought it would be meom-ige?

[yuvipanda]

oh, you caught that @sgibson91. The meom-ige folks used to have a hub under 2i2c.cluster.yaml but we moved it to their own project. Guess we didn't decomission the old cluster!

Should definitely be in the other file.

[sgibson91]

Guess we didn't decomission the old cluster!

I did see the low-touch-cluster still going this morning in the GCP web GUI!

[choldgraf]

@sgibson91 mmm, so does that mean that this PR is not correct? It would be a different cluster yaml file?

[sgibson91]

@sgibson91 mmm, so does that mean that this PR is not correct? It would be a different cluster yaml file?

I think so, I think it should be meom-ige.cluster.yaml but it also sounds like there's some historical confusion and switching about where this config lives.

[sgibson91]

If I understand correctly, the changes in your PR should be applied here

https://github.com/2i2c-org/pilot-hubs/blob/d6b109e425885232312e462d15dbcc81bf56ca23/config/hubs/meom-ige.cluster.yaml#L89-L91

And the block you have actually modified should be removed?

[choldgraf]

@sgibson91 OK I think the latest commit does this, does that look right?

I assume we'll have to manually delete the old cluster? I am not sure how to do this these days though

[sgibson91]

I think the latest commit does this, does that look right?

Looks right to me, but I'll let @yuvipanda officially sign off on it. We don't need to delete the 2i2c cluster, but we will need to delete the grenoble deployment.

Something like:

helm delete grenoble --namespace grenoble
helm delete namespace grenoble

But we should do that after this has been successfully deployed and the SWOT folk confirm.

[damianavila]

but we will need to delete the grenoble deployment.

There might be some other old stuff (other namespaces) in the cluster (in fact I have an internal note to open a ticket about that), it is probably worth checking if there is anything else old (and not used) and deletes it as well...

[yuvipanda]

Let's open another issue to tackle decomissioning the existing hub fully? Otherwise this is good to go.

[sgibson91]

I opened #612

[yuvipanda]

Great! @choldgraf do you feel comfortable merging (and hence deploying) this and checking out if the new image works?

[choldgraf]

OK let's see if this works 🙃