<!-- NOTICE: This Issue tracker is for the GitLab Helm chart, not the GitLab Rails application. Support: Please do not raise support issues for GitLab.com on this tracker. See https://about.gitlab.com/support/ --> ## Summary After upgrading Gitlab from 17.11.7 to 18.5.2, we have started to encounter errors in the backup when creating the artifacts backup. After some time we get the error described below where the backup process seems to no longer be able to make connections to Azure blob storage. At this point, the backup stops with artifacts and moves on leaving the backup incomplete. It appears the something in the way the backup works has changed and is exhausting the available ports (too many open connections). We looked back through all our logging and this error is only seen on the next backups after the upgrade to 18.5.2 ## Steps to reproduce Upgrade to 18.5.2 and run the backup with large numbers of artifacts. ## Current behavior The artifact backup is failing with a "cannot assign address" error. ## Expected behavior The backup routine should retry or backoff until there are connections. ## Versions - Chart: 9.5.2 - Platform: - Cloud: AKS - Kubernetes: - Server: 1.32.7 ## Relevant logs ``` failed to traverse container: cannot list files due to reason Get "https://xxxxxx.blob.core.windows.net/gitlab-artifacts?xxxxxx": dial tcp xx.xx.xx.xx:443: connect: cannot assign requested address ``` <!-- template sourced from https://gitlab.com/gitlab-org/charts/gitlab/-/blob/master/.gitlab/issue_templates/default.md -->