PowerFlex 3.0: VTree Migration Cannot Be Initiated And MDM Process Restart
Summary: "vtree" migration fails due to timeout.
Instructions
"vTree" migration fails due to timeout.
scli --migrate_vtree --volume_id 8b909f1e00000001 --destination_protection_domain_name PD00 --destination_storage_pool_name SSD00 Error: MDM failed command. Status: A timeout occurred
The primary MDM process restarts
In the MDM events, the migrate_vtree command result error code is Illegal.
2020-05-21 13:51:06.403000:1934119:MDM_CLI_CONF_COMMAND_RECEIVED INFO Command migrate_vtree received, User: 'admin'. [15646155] Migrating vtree of volume : ID: 8b909f1e00000001 to StoragePool Name: SSD00 Protection Domain: PD00 Queue Position: Default 2020-05-21 13:51:11.418000:1934121:CLI_COMMAND_FAILED WARNING Command migrate_vtree was not successful. Error code: Illegal [15646155]
The MDM logs a panic similar to the following.
21/05 14:35:26.924724 Panic in file /data/build2/workspace/ScaleIO-Common-Job@2/src/net/net_rpc.c, line 520, function netRpc_SendResponse, PID 30351.Panic Expression !(pMsg->msgSnd.rpcReq.context == 0) .
Scenario
One or more SDS from the migration source or target protection domains (PD) is disconnected from the system.
Impact
vTree migration cannot be initiated.
MDM process restart, possibly triggering an MDM primary role switchover
Root Cause
When vTree migration is initiated, the MDM checks whether all SDS in the source and target PD are connected.
Due to a software issue, this check is not complete.
By design, when an operation cannot be completed, the MDM process terminates itself.
Workaround
Prior to vTree migration initiation, manually confirm that all SDS in the source and target PD are connected.
Impacted Versions
PowerFlex OS versions lower than 3.0.1.1
Fixed in Version
PowerFlex OS 3.0.1.1