76 virtual machines upgraded across five major versions while flight operations continued.
PROJECT AT A GLANCE
ACM Air Charter
THE TASK
Upgrade Proxmox and Ceph while flight operations continue.
OUR CONTRIBUTION
Check version compatibility, upgrade node by node and document test results.
THE OUTCOME
Flight operations remained available. ACM operates the upgraded environment itself.
THE PROJECT
Starting point and requirements.
ACM Air Charter provides charter flights, aircraft management and maintenance at Baden Airpark. Its systems are needed around the clock.
The Proxmox and Ceph clusters were stable but several major releases behind. The gap widened each quarter. ACM chose to upgrade while the work could still be planned.
NON-NEGOTIABLE REQUIREMENTS
Flight operations continue. There is no maintenance window.
IMPLEMENTATION
What Sysfacts took on.
Recorded the starting state before any changes: versions, quorum, OSD topology, network paths and running workloads
Planned staged Proxmox VE and Ceph upgrades. Compatibility was checked at every intermediate version
Worked node by node in the live cluster and live-migrated VMs. Each subsequent change followed a cluster health check. The upgrade path and test results were documented for the operations team
Data integrity took priority in storage work. Rebalancing was controlled to keep data available throughout
AFTER THE PROJECT
What changed for the client.
OUTCOME AND OPERATIONS
Charter operations, AOG readiness and maintenance remained available. ACM operates the upgraded environment itself and has documentation of the changes.
PROJECT DATA
virtual machines on three nodes
76
maintenance windows over two weeks
0
Technical context +
01
Original environment
76 virtual machines ran on a three-node Proxmox and Ceph cluster with 30 OSDs. The platform was several major versions behind. No additional host was available to absorb workloads.
02
Version upgrade
Proxmox VE was upgraded from 7 through 8 to 9. Ceph moved from 16 through 17 and 18 to 19 (Squid). The sequence accounted for compatibility between intermediate versions.
03
Work in the live cluster
Cluster health, quorum, data placement and network paths were checked before every change. Virtual machines were moved between nodes before each host was upgraded.
04
Storage and verification
Ceph rebalancing was controlled during the upgrades. Each subsequent step followed checks of the current state. The upgrade path and results were documented for the internal operations team.
INSIDE THE PROJECT
THE THREE NODES IN THE RACK · NVME IN EACH CHASSIS
“We needed to return to a current version without stopping flight operations. Sysfacts completed the upgrade in two weeks, with the knowledge these Proxmox and Ceph version changes require. The key was that cluster health and data integrity remained reliable throughout. That is exactly how it went.”
Alexander Schmidt IT Lead · ACM AIR CHARTER Translated from German
