No, Mars 400 can keep his services while replacing any Arm microserver module, switch module, disk drive, M.2 SSD, or power supply module.
All Mars 400 modules are all hot-swappable.
- Microserver: Microservers are independently installed on a passive backplane. Every microserver has a power supply control switch, which is controlled by the BMC. When replacing a faulty microserver module, you can just turn off the power for the failed node.
- In-chassis switches: There are two in-chassis network switches in every Mars 400 appliance. Their internal ports connect to the bonded ethernet on microservers. This design provides a redundant network for both Ceph public and in-cluster communication. If one of the switches has a problem, all microservers in that Mars 400 can keep their network connection through the healthy switch. The switch is hot-swappable. Replacing the in-chassis switch will not stop the whole appliance.
- Redundant power supply: The power supply unit has two power modules work as the active-active redundant mode.
- Replacing the M.2 SSD: Each microserver has an M.2 SATA SSD for the usage of blue store DB and WAL or is used as the log storage of monitor nodes. You can replace a failed M.2 SSD without interrupting the service of the other seven nodes.
- Replacing the data disk drive: Data disks are hot-swappable. You can Replace the disk drive by pulling the chassis out of the rack. Every Mars 400 comes with a pair of slide rail. You can effortlessly slide out the appliance.
