Cloud Build trigger for ledger broken
bourgeoisor opened this issue · 3 comments
The Cloud Build trigger for the ledger seems stuck at the following:
Step #1 - "build-and-push-images": Checking cache...
Step #1 - "build-and-push-images": - e2e-tests: Not found. Building
Step #1 - "build-and-push-images": - balancereader: Error checking cache.
Step #1 - "build-and-push-images": getting hash for artifact "balancereader": getting dependencies for "balancereader": could not fetch dependencies for workspace /workspace: initial Jib dependency refresh failed: failed to get Jib dependencies: running [/workspace/mvnw jib:_skaffold-fail-if-jib-out-of-date -Djib.requiredVersion=1.4.0 --projects src/ledger/balancereader --also-make jib:_skaffold-files-v2 --quiet --batch-mode]
It sounds like there is existing cache conflicting.
I have tried:
- Setting the
$_CACHE
env to something else (as a cache-buster) - Replacing the
$_CACHE
variable by a hard-coded cache-buster - Looking to see if there were a cache hiding in
$HOME/.skaffold
- Running
skaffold build
multiple times in a row
None of those methods seem to have any effect. (see various tests in this branch)
The same skaffold build
command works fine on my machine:
Generating tags...
- e2e-tests -> gcr.io/obourgeois-sandbox-10/bank-of-anthos/e2e-tests:v0.5.1-717-g594d53d
- balancereader -> gcr.io/obourgeois-sandbox-10/bank-of-anthos/balancereader:latest
Checking cache...
- e2e-tests: Not found. Building
- balancereader: Not found. Building <--- not found means no errors! yay!
Starting build...
Building [balancereader]...
[. . . and then it actually builds! . . .]
Note that PR pipelines work as expected. What is broken is the individual service pipelines that are triggered one new commits on main
. Looking quickly at the list of commits, you can see that those 3 ledger pipelines has never passed since they were implemented, so this is not a regression; they just never worked. The other services pipelines (frontend, contacts, userservice) on the other hand work fine!
@aablsk any potential insight is deeply appreciated!
Dear @bourgeoisor, sorry for the late reply, I'm currently pretty swamped with onboarding and getting my new apartment set up.
I think this problem was related to this issue (which includes the workaround I was using before 2.1.0): GoogleContainerTools/skaffold#7409
I think it would be valuable for the skaffold team to be aware of this potential regression. (I recommend letting @aaron-prindle know)