Ghost device handles in AIX cause longer return time on iSCSI commands
Contributors
Suggest changes
-
PDF of this doc site
-
Administration for Linux
- Role-based access control in SnapDrive for UNIX
-
Installation and Administration for Solaris
- Role-based access control in SnapDrive for UNIX
- Creating and using Snapshot copies in SnapDrive for UNIX
-
Installation and Administration for IBM AIX
- Role-based access control in SnapDrive for UNIX
-
Administration for Linux
Collection of separate PDF docs
Creating your file...
This may take a few minutes. Thanks for your patience.
Your file is ready
Querying AIX devices through iSCSI or FC that have ghost device handles (devices handles in a defined or failed state) can cause SCSI commands to wait for the AIX default timeout which is about 30 seconds.
You can see this issue after switching an AIX system from using FC to using iSCSI if FC child device handles are not cleaned up or a FC HBA is left not attached to a switch.
Workaround:
-
If the Fibre Channel HBA is not attached to a switch, install a Fibre Channel IOA wrap plug in all Fibre Channel initiators not connected to a Fibre Channel Switch.
-
If the Fibre Channel HBA is attached to a switch but has Failed or Defined device handles as child devices, remove the stale ghost device handles with the
rmdev
command.