In https://hibernate.atlassian.net/browse/HSEARCH-4684 , we introduced a workaround to accurately detect that postgres really was up. In essence we would wait for postgres to tell us it’s ready twice , because for some reason it would, and the first time it wasn’t really ready. Fast-forward to today, and postgres no longer tells us it’s ready twice:
See https://ci.hibernate.org/blue/organizations/jenkins/hibernate-search/detail/main/493/pipeline I don’t know what’s going on, but we need to have another look. Maybe simply upgrading to the latest postgres would lead to a more… stable… behavior? |