Users Guide

● If one of the source SPs reboots but the surviving SP can service the source volume, both import and I/Os continue without
any impact.
Depending on the nature of the fault, the import orchestrator handles them so that there is minimum impact on the host I/Os. In
all these cases, error messages, or alerts, or both describe the issue to the user.
Based on system load, it is possible that the errors between the PowerStore cluster and the source network result in I/O errors
to a user application, even when an automatic cancel occurs. For such rare cases, user intervention may be required to restart
the I/Os to the source volumes. If the problem persists, contact your service provider for resolution.
Source disaster recovery
If the source system has a disaster recovery (DR) incident in which it becomes entirely unreachable during import, the following
applies:
● If a different host is used in the DR site for recovery, then the existing source system-specific DR procedure should work.
The import-specific states are only for the primary site system volume. These states have no bearing on the system volume
of the secondary or DR sites.
● If the same host is used to run the applications when the primary system is impacted in a disaster, then the following applies:
○ When the primary (source) system fails, the import process initiates an automatic cancel operation. Check whether
the automatic cancel is successful. Once this operation successfully completes, continue with the source-specific DR
recovery procedure.
○ If automatic cancel was not successful, for example, it failed due to network connectivity issues, then additional steps
might be required based on the failure. The recovery steps are specific to the error. Contact your service provider for
more information about recovery.
○ EqualLogic PS system have some limitations which affect DR during import. Contact your service provider for resolution.
Double fault
Double faults are defined as two separate fault events happening around the same time, such as the ones in the following
examples:
● Paths from the host to the source go down and, around the same time, the connection between the PowerStore and the
source system also go down.
● Both SPs of the source system go down during import.
For such rare case double faults, recovery is not fully automatic and manual intervention is needed. The recovery depends on
the specific fault condition. Contact your service provider for resolution.
A single problem may manifest as a double fault where the recovery needs some manual intervention. For example, if the source
system or volume goes out-of-service (for example, a source volume goes offline) during import, this problem manifests as a
double fault. First, the import fails because PowerStore is not able to reach the source volume in the SAN path. The subsequent
automatic cancel does not fully succeed since the host is not able to reach the source volume, it is offline. To recover, repair the
source volume state, do any required cleanup, then try the import again.
NOTE:
This error, which is due to a source volume going offline, results in data being unavailable temporarily, even when
import is not occurring.
PowerStore is a dual-node, multiappliance scale-out cluster. Typically, most failures in data unavailable are not fatal and handled
by the inherent PowerStore cluster high availability design. The import orchestrator runs inside the PowerStore appliance and so
the orchestrator is as highly available as PowerStore itself.
However, should there be an extreme case where a PowerStore cluster goes down during import before cutover, an option
exists to recover and go back to the source system. Until this recovery, data is temporarily unavailable. The general idea is
that you must do some manual intervention at the source system to clear the migration-specific states and then run the host
platform-specific steps. Contact your service provider for such a recovery.
If an entire appliance is down, both nodes of the appliance are down. However, in a multiappliance cluster, ongoing imports
continue in other running appliances. The imports, for which the destination volume is on that specific appliance which is down,
get stalled. It is recommended that the appliance be brought up first before trying to cancel this stalled session.
66
Fault handling by PowerStore import