A file server everyone depended on, replaced by a cluster with no visible failover
SMB / Samba scale-out cluster (CTDB)
Where they started
All project data lived on one Windows file server with a nightly copy to a NAS. Any maintenance meant a planned outage, and permissions had drifted over a decade.
The problem, precisely
Users navigated by drive letter, so the replacement had to preserve paths and permissions exactly while becoming highly available.
What we did
Reproduced the existing share structure and ACL model first on a cluster that presents a floating address, then migrated data with a synchronised cutover over a weekend.
Delivered
- Three-node cluster presenting a single floating SMB address to users
- Permission model reviewed and rebuilt, mapping Windows ACLs to POSIX correctly
- Data migration with a pre-staged copy and a short final synchronisation window
- Quota and audit policy so storage growth is visible per project
- Node failure test during business hours, with open-file behaviour documented
Technologies used
Measured result
Every number below was measured and reported to the customer. Where we do not have a measured figure, the entry is a fact rather than a percentage — we would rather be precise than impressive.
- 0 user-visible interruptions during migration
- 3 nodes, any one of which can be lost without impact
- 60TB migrated with permission fidelity verified
- 1 permission model, replacing a decade of drift
Preserve the paths, fix the permissions. Users will forgive a new server name; they will not forgive a folder that suddenly becomes read-only for the wrong half of the team.