2.3 Cell Behavior During the Initial Boot of a Partition

     

2.3 Cell Behavior During the Initial Boot of a Partition

When we power-on a cell, or a cabinet, or the entire complex through the GSP PE (Power Enable) command, each cell goes through a sequence of tests before booting within a partition configuration, if appropriate. As soon as the cabinet 48V power has stabilized, a hardware register for each cell is set. This register dictates the behavior of the Boot Inhibit Bit ( BIB ) and is commonly referred to as Boot-Is-Blocked. BIB is designed to stop a cell from booting until all appropriate checks have been made to ensure that the cell is functioning properly. Each cell will go through its Power-On Self Test ( POST ), which has various steps such as CPU self tests, Memory self tests, IO Discovery, and Fabric Discovery. During this initial phase, the cells are considered INACTIVE. The amount of cell- related hardware will determine how long the POST will take to complete. We can monitor the POST from the VFP screen within the GSP.

 GSP MAIN MENU:          CO: Consoles         VFP: Virtual Front Panel          CM: Command Menu          CL: Console Logs          SL: Show chassis Logs          HE: Help           X: Exit Connection GSP>  vfp  Partition VFP's available:      #   Name     ---  ----      0)  Partition 0      S)  System (all chassis codes)      Q)  Quit GSP:VFP>   Eindicates error since last boot      Partition 0  state            Activity      ------------------            --------      Cell(s) Booting:    238 Logs   #  Cell state                    Activity   -  ----------                    --------   0  Early CPU selftest            Processor test               238  Logs GSP:VFP (^B to Quit) > 

The POST goes through various phases. (The Logs can be viewed via the GSP SL command. Unless we see an error indicated by the letter E beside the cell number, the Logs are simply Activity Logs .) Once the cell has finished its POST, it reports its hardware configuration to the GSP and is left spinning on BIB . A cell will spin on BIB , waiting for other cells in its partition configuration to finish their POST before being allowed to boot the partition. This makes sense, because we can't have a partition boot while a cell is still performing a POST. While a cell is performing its POST, details of cell-related hardware are not available to the GSP or other administrative commands such as Partition Manager. Once all cells have reach BIB, the GSP will supply the cells with the current version of the Complex Profile, release BIB, and allow the partition to boot. As soon as BIB is cleared the cell is considered to be active. At this stage, the cells are said to have reached partition rendezvous . If a cell does not get to a BIB state within 10 minutes of the initial POST, the GSP will clear BIB for the remaining cells and allow them to boot. This avoids the situation of a partition being blocked due to the failure of a single cell. At this point, the cells coordinate their activities in order to choose a Core Cell, which will proceed to boot the PDC/BCH. This is explained in Figure 2-16.

Figure 2-16. Booting a 2-cell partition.

A cell will remain in a BIB state due to the following reasons:

• The cell has not passed its POST and has some hardware error. This is indicated by the letter E beside the cell number in the VFP. An investigation of the Chassis Logs (via the GSP SL command) would reveal any Error Logs . Logs are time stamped and any new Error Logs should be reported to HP for further investigation.

• The use-on- next -boot flag has been set to NO for this cell. This is a specific partition configuration. We should not see this when creating the Genesis Partition.

• The cell has an incoherent Complex Profile. This normally indicates some form of hardware error whereby the Complex Profile held in NVRAM has become corrupted. This should be reported to HP for further investigation.

Now that we have a Genesis Partition and understand the state of cells during the initial boot of a partition, we can now look at adding/modifying partitions via the Partition Manager software.