Add helm test integration
#415
Conversation
seaneagan
force-pushed
the
helm_test_integration
branch
2 times, most recently
from
48f1352 to
c723f86
Compare
seaneagan
force-pushed
the
helm_test_integration
branch
4 times, most recently
from
d05f64d to
b2a0946
Compare
seaneagan
force-pushed
the
helm_test_integration
branch
2 times, most recently
from
d01db3e to
2b9fc59
Compare
seaneagan
force-pushed
the
helm_test_integration
branch
7 times, most recently
from
a84615c to
a2fcab2
Compare
seaneagan
force-pushed
the
helm_test_integration
branch
6 times, most recently
from
b95bbe2 to
6ec9a29
Compare
seaneagan
force-pushed
the
helm_test_integration
branch
from
6ec9a29 to
4fb2f43
Compare
test/e2e/40_tests.bats
Outdated
| poll_until_equals 'release deploy' 'True' "kubectl -n $DEMO_NAMESPACE get helmrelease/podinfo-helm-repository -o jsonpath='{.status.conditions[?(@.type==\"Released\")].status}'" | ||
|
|
||
| if [ $HELM_VERSION = "v2" ]; then | ||
| test_exists_jq_filter='.info.status | has("last_test_suite_run")' |
There was a problem hiding this comment.
there is currently a race condition where the observed generation is set, immediately followed by the lock opening up for the next chart sync interval loop, and the informer cache is not updated yet with the observed generation change, which results in an upgrade, which causes "last_test_suite_run" to be unset, which causes this test to fail. This is resolved by #444 which runs a harmless dry run compare, instead of a direct upgrade.
There was a problem hiding this comment.
Fixed the race condition by using polling in the test.
| @@ -406,14 +433,14 @@ func (r *Release) upgrade(client helm.Client, hr *v1.HelmRelease, chart chart, v | |||
| ResetValues: !hr.GetReuseValues(), | |||
| SkipCRDs: hr.Spec.SkipCRDs, | |||
| MaxHistory: hr.GetMaxHistory(), | |||
| Wait: hr.Spec.Wait || hr.Spec.Rollback.Enable, | |||
There was a problem hiding this comment.
existing behavior is that wait only defaults to true on upgrades (not installs) when rollbacks are enabled. It seemed better to me for the config to be statically determined, and not depend on current action (install vs upgrade), so I implemented that for when tests are enabled, and then I just made the behavior match when rollbacks are enabled. This does mean wait may be turned on for some installs where it wasn't before. Let me know what you think.
There was a problem hiding this comment.
This is 👌, I would have made the same change if I had to further extend the waiting behaviour.
seaneagan
force-pushed
the
helm_test_integration
branch
from
4fb2f43 to
04d8675
Compare
| chart: | ||
| repository: https://stefanprodan.github.io/podinfo | ||
| name: podinfo | ||
| version: 3.2.0 |
There was a problem hiding this comment.
Wouldn't hurt to update things to a newer podinfo version, but can be taken care of in a a separate PR to change the other fixtures as well.
There was a problem hiding this comment.
Lengthy PR! But tidy, clean, and extensively documented.
LGTM @seaneagan 🥇
Fixes #369.
@hiddeco