Solution for STOP 7B error when replacing a disk system controller. The use of virtual machines in data recovery
Disclaimer: over the years, the exact model of servers and controllers could not be established, so there will be reservations in the text when mentioning the model and, of course, errors in details are possible. This does not change the essence - the procedure for overcoming the lack of a new controller driver without Recovery Console and the “deep” knowledge of Windows 2003 is described below. Probably, more optimal solutions to such a problem are possible than using a virtual machine.
Problem : The controller broke down, two SCSI disks fell out, and then the whole server died. It was an HP Proliant, presumably a Proliant DL380 G3 with a Smartarray 6i controller. By morning, you need to make Exchange work and several domains that were spinning on the server.
Raised the wheels, made sector-by-sector images, made the image of the entire raid. We restored the data from the image, but the customer also wants to restore the server. There is no server with the same RAID controller, there is no time to search - it is a controller of several domains, not all domains have BDC. Plus, there is spinning Exchange. By morning, everything needs to be put into operation - several people's offices will come and will not be able to work. There is another server with another controller - Smartarray P410 on SAS disks. We will try to make it work on it. The full image of the array is poured onto the new RAID on the new controller in the hope that the drivers are compatible, but the system does not want to boot - it crashes with the error STOP 0x0000007B INACCESSIBLE_BOOT_DEVICE. There is no Recovery Console. How to implement it in a non-working server is not clear. We tried HP SmartStart in the hope that it would install the drivers in the new system, but it did not help.
In the process, we tried to start the server in a virtual environment - it works, but slowly and sadly - users will tumble it down in the morning. We tried to make backup-restore from a virtual machine to a new one using ntbackup. Backup will not be done until morning - slowly. We thought about the image of Acronis, but there is not a single PCI in the new server, and all the PCIe controllers available at hand do not want to work in the new server, so there is no way to connect the image to the disk from which you wanted to restore this image to the new RAID.
We are trying to install the driver of the new controller in VMware (this is an sfx archive). He says that such equipment is not in the system. Unpack the archive and try to install the driver via the context menu with the right mouse button, boot, it does not help - STOP 7B. We start digging the driver itself, or rather the inf file, after studying the InstallHinfSection Function . It does not have a Default section or anything like that, apparently because of the installation by the right mouse button that didn’t help. We start to try all sections in a row with the team
RUNDLL32.EXE SETUPAPI.DLL,InstallHinfSection DefaultInstall 4 C:\WINDOWS\INF\SHELL.INF
where DefaultInstall is the name of the section, 4 is “If a reboot of the computer is necessary, ask the user for permission before rebooting". I was especially interested in the sections that mentioned something like copying and installing the driver
[HpCISSs2_Inst.ntx86] CopyFiles = @HpCISSs2.sys [HpCISSs2_Service_Inst] ServiceType = %SERVICE_KERNEL_DRIVER% StartType = %SERVICE_BOOT_START% ErrorControl = %SERVICE_ERROR_NORMAL% ServiceBinary = %12%\HpCISSs2.sys LoadOrderGroup = SCSI Miniport AddReg = pnpsafe_pci_addreg AddReg = boot_ctrl_addreg AddReg = disk_timeout_addreg
after executing a successful command, a proposal was received to reboot, an image was made and the system was loaded. Users successfully earned in the morning, and then DC and Exchange replicas were made to the new servers, but customers did it themselves.
Practical benefits for the masses
Suppose you change the motherboard after the old one breaks down - you can safely try to put a disk into it after the driver has been introduced. To do this, you need VMware, Winhex (to disable / enable the boot sector - it helps to avoid NTFS damage if it is opened simultaneously by host and guest systems) and new drivers. It may be worthwhile to describe this process step by step in a separate article.
This order is described on our website in the data recovery cases section. Another case -virtualization of an old Novell server after a controller failure .
UPDATE of practical use for the masses: mafet rightly noted that in the same way (via RUNDLL32.EXE SETUPAPI.DLL, InstallHinfSection), you can add the drivers of the new system before migrating to it