Usually after a system crash occurs, you will worry again fails, but found the system log does not record any information before and after the panic, unable to parse the cause of the failure that has no cure.
However, in practice, there are several mechanisms to Linux guarantees after a system crash occurs, you can get valuable information for the analysis of the problem. Identify a hardware failure, or an application bug. Linux, there are several ways to obtain various crashes resulting information. 1.Coredump Coredump is typically used to debug an application error when running certain applications to crash when an exception occurs, you can turn on the system's function to get the coredump a program crash memory information, to analyze crash causes:/etc/profile Riga (or modify): run the command: ulimit-c0 sysctl-w "/coredump/%n.core kernel.core_name_format =" the command mean core files in the directory, file name/coredump is process name + .core 2.Diskdump diskdump tool provides a single machine, create and collect vmcore (kerneldump) of capacity, without using the network. When the kernel itself when a crash occurs, the current memory and CPU status and related information will be saved to a disk on the diskdump reservations partition. At the next reboot when the system restarts, diskdump initialization script will read from reserved partition saved information and create a vcore file, and then the file is again stored at/var/crash/directory, file name is 127.0.0.1-the following is a configuration HPSCSI device enabling the diskdump process, if not HPSCSI device (i.e. device named/dev/sdX form), you do not need to perform the third and fourth in two steps. But in the first step before executing the command: modprobe diskdump first step: edit/etc/sysconfig/diskdump file, a blank partition after the device name is filled in, for example, save the exit: DEVICE =/dev/cciss/c0d0p2 second step: initialization dump device # servicediskdumpinitialformat warning: the partition so the data will be lost. Step three: use the replace the current cciss_dump module cciss module: in/etc/modprobe.conf find the following line: aliasscsi_hostadaptercciss is amended as follows: Add row: aliasscsi_hostadaptercciss_dump optionscciss_dumpdump_drive = 1 Note: assuming the diskdump file configured for/dev/cciss/c0d [# a] p [# b], please set to: optionscciss_dumpdump_drive = [# a] step four: rebuild initrd file: # mv/boot/initrd-would uname-r would .img/boot/initrd-would uname-r would .img.old # mkinitrd/boot/initrd-would uname-r would .img would uname-r would step five: set the diskdump service able to boot from boot: # chkconfigdiskdumpon 3.Netdump if you use red flag DC4.0 or 3.0 version system that cannot support the diskdump, you can use to achieve the output vmcore netdump. But the Netdump requires at least one server, and any number of clients. The server used to receive client information panic, the client is often panic a machine. (A) of the server configuration: (1). the examination whether the netdump-server is installed: rpm-qnetdump-server if not installed, your CD RedFlag/RPMS/directory to find the start package, netdump-server execute the command: rpm-ivhnetdump-server-x.x.x.rpm (x is the version number) to install. (2). the server package installed, use the command: passwdnetdump change user's password. (3). the open service: chkconfignetdump-serveron (4). the running server: servicenetdump-serverstart
No comments:
Post a Comment