ingress issues in CI
# 💻|contributing
a bunch of our VM tests have been failing in CI recently
i was creating this PR to see if changing the hostnames from possibly-registered valid ones to invalid ones would change anything but it's unrelated - the tests were failing before and are failing after
ah, these tests were not enabled before
Didn't @calm-family-29323 already post a fix PR?
@alert-helicopter-61082 can I merge that fix or will you look into a more clean way of fixing the issue?
@calm-family-29323 i'm tempted to say let's just merge it to get our CI back to green, we can always make another followup PR where needed
@alert-helicopter-61082 I get this on platform now
Copy code
âś– Validation error(s) when reading configuration file at /Users/orz/work/garden-platform/.garden/local-config.json:
      "code": "invalid_type",
      "expected": "string",
      "received": "undefined",
      "path": [
      "message": "Required"
I'm also unable to set a default env via the command (get the same error) Fixed it for myself by manually adding the
entry to the file.
Yeah, we need a default empty string value there in the local config schema. Could you fix that?
oh this was an unrelated thread, my bad
bah. the tests passed on your PR, but now on 0.13 the VM tests are failing with timeouts. related or not? 🤔
these integ tests are making me feel strong feelings
I think it's the same imagePull issue
Those same tests are also failing from my machine directly
Actually, more specifically, they fail on the first run, not the second. So basically they don't work on a clean cluster.
We really need to fix these integ tests... We're back to "no clue whether I broke something in my PR" territory here.
I can have another go at it, but I've failed to reproduce the issue locally even on a reset cluster, deleted images, cleared cache, etc. Also had Anna run these on her laptop and they passed just fine. I'd appreciate a higher log level output if you can still reproduce it on your machine. If not I'll just try to reproduce it again or debug in the ci.
I suspect it's all the same issue somehow... Maybe changing the integ tests to log aggressively and break after first failure would be good.