Troubleshooting Specific Issues; Software Services; Failover - HP StoreAll 9730 Administrator's Manual

Troubleshooting specific issues

Software services

Cannot start services on a file serving node, or Linux StoreAll client
SELinux might be enabled. To determine the current state of SELinux, use the getenforce
command. If it returns enforcing, disable SELinux using either of these commands:
setenforce Permissive
setenforce 0
To permanently disable SELinux, edit its configuration file (/etc/selinux/config) and set
SELINUX=parameter to either permissive or disabled. SELinux will be stopped at the next
boot.
For StoreAll clients, the client might not be registered with the Fusion Manager. For information on
registering clients, see the

Failover

Cannot fail back from failover caused by storage subsystem failure
When a storage subsystem fails and automated failover is turned on, the Fusion Manager will
initiate its failover protocol. It updates the configuration database to record that segment ownership
has transferred from primary servers to their standbys and then attempts to migrate the segments
to the standbys. However, segments cannot migrate because neither the primary servers nor the
standbys can access the storage subsystem and the failover is stopped.
Perform the following manual recovery procedure:
1. Restore the failed storage subsystem (for example, replace failed Fibre Channel switches or
replace a LUN that was removed from the storage array).
2. Reboot the standby servers, which will allow the failover to complete.
Cannot fail back because of a storage subsystem failure
This issue is similar to the previous issue. If a storage subsystem fails after you have initiated a
failback, the configuration database will record that the failback occurred, even though segments
never migrated back to the primary server. If you execute ibrix_fs -i -f FSNAME, the output
will list No in the ONBACKUP field, indicating that the primary server now owns the segments,
even though it does not. In this situation, you will be unable to complete the failback after you fix
the storage subsystem problem.
Perform the following manual recovery procedure:
1. Restore the failed storage subsystem.
2. Reboot the primary server, which will allow the arrested failback to complete.
StoreAll client I/O errors following segment migration
Following successful segment migration to a different file serving node, the Fusion Manager sends
all StoreAll clients an updated map reflecting the changes, which enables the clients to continue
I/O operations. If, however, the network connection between a client and the Fusion Manager is
not active, the client cannot receive the updated map, resulting in client I/O errors.
To fix the problem, restore the network connection between the clients and the Fusion Manager.
150 Troubleshooting
HP StoreAll Storage Installation Guide.