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
In the event of an upload failure, the concourse pipelines can leave a "broken" repository for an hrev.
We should likely investigate an on_failure step which cleans up the hrev folder in s3 if the upload fails.
However, in the event that s3 is 100% unavailable, this may not work.
The better solution is hpkgbouncer needs to validate the "completeness" of the hrev repository directory before sending users there. This would likely be accomplished by:
Examining each repo file
Validating that all packages exist for the given repo file, and the sha256 checksums are valid
Adding to RepositoryCache
hpkg bouncer needs a native rust crate that can read hpkr files however to accomplish this.
The text was updated successfully, but these errors were encountered:
kallisti5
changed the title
[concourse:
[concourse]: cleanup on failure
Jan 8, 2024
kallisti5
changed the title
[concourse]: cleanup on failure
[concourse]: cleanup repositry directory on failure
Jan 8, 2024
A note for future viewers: The workaround is to kick off a new concourse job to re-build and re-upload the failed hrev repo. (or manually delete the folder from s3)
In the event of an upload failure, the concourse pipelines can leave a "broken" repository for an hrev.
We should likely investigate an on_failure step which cleans up the hrev folder in s3 if the upload fails.
However, in the event that s3 is 100% unavailable, this may not work.
The better solution is hpkgbouncer needs to validate the "completeness" of the hrev repository directory before sending users there. This would likely be accomplished by:
hpkg bouncer needs a native rust crate that can read hpkr files however to accomplish this.
The text was updated successfully, but these errors were encountered: