Portworx Enterprise - Scenarios Where the Replica Count Label and Actual Replica Count May Differ
KB0022602
Pure Product
Portworx
Question
When a volume is created, it is assigned a label named "repl" whose value indicates the replica count. This KB explains the different scenarios in which a mismatch between the configured replica count and the "repl" label value can occur.
Answer
Scenario 1: Manual Update of Volume Replica Count
- The volume replica count was manually modified using the pxctl volume ha-update command
- The corresponding repl label was not updated, resulting in a mismatch
Scenario 2: Manual Update of Volume Label
- The repl label was manually updated using the pxctl CLI
- The actual volume replica count was not modified to match the updated label
Scenario 3: License Expiry with Replica Node Decommissioned
- The Portworx license had expired, preventing any replica scale-up operations
- One of the replica nodes was decommissioned
- During decommissioning, replicas were reduced first
- Replica re-addition failed due to the expired license
- Volume is currently running with fewer replicas than originally configured
Scenario 4: When License is expired and one node in PX cluster was down for more than 24 hours
- Portworx license had expired, blocking any replica increase operations
- One replica node was in a down state for an extended period (>24 hours)
- Portworx removed replicas from the affected node after the replica move timeout was exceeded
- During the replica movement process, replicas were reduced first
- Replica re-addition failed due to the expired license
- The volume is currently running with fewer replicas than originally configured
Scenario 5: When one replica node is decommissioned and there is node to move the replica as per VPS
- One of the replica nodes was decommissioned
- During decommissioning, replicas were reduced first
- Replica re-addition failed due to the absence of eligible nodes as defined by the Volume Placement Strategy (VPS) attached to the volume
- Volume is currently running with fewer replicas than originally configured