mirror of
https://github.com/google/nomulus.git
synced 2025-07-04 10:13:30 +02:00
We have a cron job that runs the RDE upload action every 4 hours for all TLD. Normally this should be a no-op beacuse a RDE upload is scheduled after RDE staging is completed, and when it fails with non-2XX status it will retry. However if for some reason it failed due to 20X status (like waiting for the SFTP cursor), it will not retry but rely on the cron job to catch up. With the BEAM RDE pipeline every staging job saves all its deposits in a uniquely named folder to avoid the need to use a lock, which is not practical in BEAM. However the cron job has no way of knowing what the prefixes are for each TLD so it will fail in SQL mode. In this PR we implemented a logic to guess what the prefix should be and use it, if we are in SQL mode and a prefix is not provided. <!-- Reviewable:start --> --- This change is [<img src="https://reviewable.io/review_button.svg" height="34" align="absmiddle" alt="Reviewable"/>](https://reviewable.io/reviews/google/nomulus/1574) <!-- Reviewable:end --> |
||
---|---|---|
.. | ||
gradle/dependency-locks | ||
src | ||
build.gradle | ||
Dockerfile |