When the Oracle FS System starts, all Controller nodes progress through two stages. The first stage is controlled by the Controller BIOS that resides on the Controller. The second stage is controlled by Pilot software and the Configuration Manager (CM) software component on the Controller.
A Controller node starts by executing page code that resides in its Controller BIOS. You can watch the Controller LEDs to monitor the progress of this stage. (Refer to the Oracle FS1-2 Flash Storage System Installation Guide for an explanation of these LED startup codes.)
First, the page code runs a set of power-on tests.
Next, the code begins monitoring flash backed memory.
Then, the code initializes the private management interface (PMI) on the Controller.
Finally, the code passes control to pxeboot, which downloads the Controller software from the Pilot.
If this stage succeeds, the second stage begins, in which the Controller software components are initialized.
In this second stage, the Pilot performs several checks on each Controller node and, if all is well, puts the node into an Initialization state. When the initialization of the software components finishes, the Pilot then, by sending commands in a defined sequence, brings the Controller software components to a Normal state.
Both the Oracle FS System Manager (GUI) and the Oracle FS CLI products identify the names of these Controller software components with a prefix of PDS_COMP_. This prefix is followed by the initials for the component. For example, CM identifies the Configuration Manager. Support administrators can selectively control the execution of these software components by enabling and disabling halt points on those software components.
If the Support administrator has previously enabled a halt point on one of these software components, the Controller boot process is halted at the specified halt point. The halt points are unique for each component.
When the startup process stops, the Support administrator can take various diagnostic actions, such as dumping a log file or reading an internal table.
After completing those actions, the Support administrator can instruct the startup process to continue. At this point, the Pilot now sends commands to each Controller to resume the startup process from the point at which it was halted. If there are multiple halt points, the system starts up to the next halt point