You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
The cloud image used to run the staging-with-rebase test suite in CircleCI was last built about a year ago:
$ gcloud --project securedrop-ci compute images list --filter="family:fpf-securedrop AND name ~ ^ci-nested-virt-stretch" --format json | grep -i creationtimestamp
"creationTimestamp": "2019-05-15T20:10:18.310-07:00",
"creationTimestamp": "2019-07-25T09:47:57.609-07:00",
Since we bake in the Xenial vagrant images into the image to avoid pulling them, that means the apt lists inside those vagrant images were also last updates about a year ago. Since the CI run includes an apt-upgrade, let's rebuild the boxes to minimize CI run time.
Since we're heading into QA for 1.4.0, I propose we don't change CI until after the release. It's worth noting that the CI image selection logic will always choose the newest—I propose we start pinning a hardcoded version to make the results a bit more predictable.
We should also update the base image stretch -> buster, in line with the dev docs.
The text was updated successfully, but these errors were encountered:
Seeing run times of the staging-test-with-rebase job that are ~10m shorter in #5360. That's a nice benefit, but the major change is that we're now testing on Buster, rather than Stretch, matching the dev docs. Whether the change also reduces error rate for the job remains to be seen.
Description
The cloud image used to run the
staging-with-rebase
test suite in CircleCI was last built about a year ago:Since we bake in the Xenial vagrant images into the image to avoid pulling them, that means the apt lists inside those vagrant images were also last updates about a year ago. Since the CI run includes an apt-upgrade, let's rebuild the boxes to minimize CI run time.
Since we're heading into QA for 1.4.0, I propose we don't change CI until after the release. It's worth noting that the CI image selection logic will always choose the newest—I propose we start pinning a hardcoded version to make the results a bit more predictable.
We should also update the base image stretch -> buster, in line with the dev docs.
The text was updated successfully, but these errors were encountered: