Server backup policy and recovery plan Sepcifically this plan is to address the following items from the * Quality Manual: o Data retention, archiving and providing documents to the public o Defined recovery * Configuration Management Plan: o no specific requirements Backup Policy: The backup server shall comply with the requirements of he server_setup guidelines - specifically it shall satisfy: * Server Selection criteria: * Distribution Selection criteria: * Server setup plan: And be verified by applying the "Post-Install Checklist" as set forth in the server_setup working instructions Periodic Checklist (monthly) * dmesg error reports ? * service problems/error reports ? * unexpected shutdowns ? * server status report (might just be 5 lines) committed to CM Security Checklist: * data integrity checking * login check (basically only CM/QM should be loginning into this box) * review of connection activity (should be more or less idle) It might be advisable to have the backup system off-line and only bring it on-line during backups. Backup (data retention) Task Minimum backup requirements: * at least weekly backup of blessed repository by git clone (as bare repository) to backup server (this really can be a cron job !) * monthly backup (at least) of Developer Repository to this backup server or to a local backup system (but this should be defined) if direct access to the backup server is not practicable (e.g. too remote) * Local branches in developer work trees are in the repository of the team member and adequate backup of these local branches must be done by that team member. Backup Checklist: * checkout from backup repository (it works) * md5sum of checkout tree (TODO: details) and compare Recovery strategy: In case the blessed repository fails * all team members are notified and stop committing to their * developer repository (they can of course continue local work tree commits. * the backup repository is cloned (bare) to the new server * CM pulls individual Developer Repositories to the cloned backup tree and merges again. * CM notifies all team members once all Developer Repositories are pulled and team members can restart regular workflow (push from work tree and pull request to CM) * The incident is recorded in the next team meeting as well as notification of re-start of normal operation after a rebuild