Retrieving diagnostics information in an N-way clustered RAID subsystem

摘要

A method and system for reading exception data by a storage server from a storage controller. An exception event is detected at the storage server. The storage server registers with the controller, such that only one storage server can read the exception data at a time. If the storage server is registered with the controller, the storage server reads the exception data. If the storage server does not successfully register with the controller, the controller keeps track of which storage servers have not read the exception data, to ensure that all storage servers can read the data.

主权项

1. A method for reading exception data by a storage server from a storage controller communicating with a plurality of storage servers that provide storage services to network storage clients, comprising:
communicating by the plurality of storage servers with the storage controller for detecting an exception event generated by the storage controller managing a plurality of storage devices, the exception event being associated with at least one of an application specific integrated circuit (ASIC) specific failure, a memory failure, a processor failure, or a disk failure; storing the exception data related to the exception event and a status for the exception event by the storage controller; detecting the exception event at the storage server after the storage controller updates the status of the exception event; after detecting the exception event, attempting to register the storage server with the storage controller by sending a unique token from the storage server to the storage controller, whereby one storage server is registered with the storage controller at a time; determining whether another storage server is already registered with the storage controller; denying registration if the other storage server is already registered with the storage controller; allowing registration if no other storage server is registered with the storage controller; if the storage server is registered with the storage controller, then reading the exception data by the storage server; tracking if another of the plurality of servers is waiting to read the exception data; based on the tracking, registering another server after the storage server has read the exception data or if the storage server becomes inactive as indicated by a timer; and indicating that the exception event is inactive by the storage controller after all the plurality of servers have read the exception data.