IBM Storwize V7000, Problem Determination Manual

Page: 266 / 314

- If there is no error data, the error is because there are insufficient
connections between nodes over the Fibre Channel network. Each node
must have at least two independent Fibre Channel logical connections, or
logins, to every node that is not in the same enclosure. An independent
connection is one where both physical ports are different. In this case, there
is a connection between the nodes, but there is not a redundant connection.
If there is no error data, wait 3 minutes for the SAN to initialize. Next
check that the following items:
v
That there are at least two Fibre Channel ports that are operational and
connected on every node.
v
That the SAN zoning allows every port to connect to every port on every
other node
v
If redundant SANs are being used, that all of them are operational.
After verifying these things, if the 550 error is still present, restart the node
by clicking
Restart Node from the service assistant.
Note: If after resolving all these scenarios, half or greater than half of the
nodes are reporting node error 578, it is appropriate to run the recovery
procedure. You can also call IBM Support for further assistance.
– For any nodes that are reporting a node error 550, ensure that all the missing
hardware that is identified by these errors is powered on and connected
without faults.
– If you have not been able to restart the system and if any node other than the
current node is reporting node error 550 or 578, you must remove system
data from those nodes. This action acknowledges the data loss and puts the
nodes into the required candidate state.
v
Do not attempt to recover the system if you have been able to restart it.
v
If back-end MDisks are removed from the configuration, those volumes that
depended on that hardware cannot be recovered. All previously configured
back-end hardware must be present for a successful recovery.
v
Any nodes that were replaced must have the same WWNN as the nodes that
they replaced.
v If any of the node canisters were replaced, they must not have participated in
any other system. You can resolve this issue by performing a node rescue on the
affected canister by using the service assistant. Do not perform this action on
any of the other node canisters.
v
The configuration backup file must be up to date. If any configuration changes
had been made since the backup was taken, the data is inconsistent and further
investigation is needed. Manual changes are required after the system is
recovered.
v
Any data that was in the cache at the point of failure is lost. The loss of data can
result in data corruption on the affected volumes. If the volumes are corrupted,
call the IBM Support Center.
Fix hardware errors
Before you can run a system recovery procedure, it is important that the root cause
of the hardware issues be identified and fixed.
Obtain a basic understanding about the hardware failure. In most situations when
there is no clustered system, a power issue is the cause. For example, both power
supplies might have been removed.
242
Storwize V7000 Unified: Problem Determination Guide Version
|
|
|
|