This document is the "ULTRIX Guide to System Crash Recovery" (Order Number: AA-ME94B-TE), published in June 1990 for ULTRIX Version 4.0 or higher.
It is designed for system administrators responsible for managing and maintaining ULTRIX systems, assuming familiarity with ULTRIX commands and system configuration.
The manual provides comprehensive information on:
- System Crashes and the Dump Process: Explains what happens during a system crash, what physical memory pages are dumped (kernel image, memory allocator, process contexts, page tables), and how to calculate the necessary dump partition sizes for both VAX and RISC processors.
- Maintaining File System Consistency After a Crash: Details how file system inconsistencies occur and how to correct them using the
fsck command, both automatically during reboot via /etc/rc and interactively. It also covers restoring pseudoterminals.
- Generating and Examining Crash Dump Files: Describes how to create crash dump files using the
savecore command (either automatically during reboot or manually), how to copy these sparse files using the dd command, and how to examine them using adb, dbx, or the crash utility to determine the cause of a system failure.
- Forcing a Crash Dump: Provides detailed procedures for intentionally initiating a memory dump on VAX and RISC processors, particularly when the system does not dump properly after a crash. This includes methods like manually starting the dump routine, forcing a segmentation fault, and initializing the processor, with instructions on their varying impacts on machine state.