CLIFFORD ILKAY wrote: >BRU (Backup and Restore Utility) and CRU (Crash Recovery Utility) from ><http://www.tolisgroup.com> should do the trick. I have used BRU for a >while and really like it. I have not had a chance to play with CRU even >though I got it bundled with BRU.
Personally, I like the dirt work of typing commands at the system prompt, knowing what's going on in detail, so I followed another path. I already had the SuperRescue CD available from Kernel.org (http://www.kernel.org/pub/dist/superrescue/v2/) - I resort to it to solve gremlin problems like Win98 that crashes while repartitioning a disk on which it was already installed, or regaining control of a notebook a manager's son ruined with an ill-ended Linux installation. SuperRescue is a Linux distribution based on RedHat 7.2 - if I remember well - that runs directly from the CD. So, I can use the SuperRescue CD to boot the ESSG and do what I want with its disk(s). To recover from a crash, I customized the SuperRescue CD installing what needed to backup and restore an ESSG: dump was already installed, so I had to install the rpm packages flexbackup and buffer (and set the console default keymap to italian, too), and rebuilded the ISO image with the provided tools, following the instructions contained in the distribution. I do my backups with flexbackup, directly and periodically launched by cron, on a HP 24GB DAT unit, and flexbackup saves all ESSG filesystems. So, if the hard disk crashes, I can boot the system with my "SuperRescue for ESSG" CD, repartition the hard disk(s) following the original scheme, mount the newly created ESSG root filesystem under the "SuperRescue for ESSG" filesystem, restore all the ESSG contents, and issue a chrooted lilo command to create the boot sector on the boot disk. This way I can do a "standalone restore" (to mutuate an OpenVMS terminology) of a complete system following a standard procedure. I should be able to mirror a standard installation on a second machine, too, and "standalone backup" every other operating system that can be started by lilo or whose custom boot block can be saved and restored on the disk using standard Linux tools, or at least every other not FS journaled Linux based system. I still have to test this procedure with the due accuracy: no time yet to do it on the production server (the test should involve simulating a crashed disk by substituting the original disk with a scratch one, doing a full restore following the procedure and rebooting and checking the recovered system). I tried the restore operation on my home machine, and it went smooth and fine, but the hardware differences (dual disk, IDE and SCSI, a 64bit AHA29160 controller instead of a 32 bit one and a different video card) were enough to crash lilo, so the system won't boot anymore, hanged on the "LI" prompt. I'm reasonably sure that the problem should be LILO itself, because a clean ESSG 4.0 install on my home machine ends the same way - hanged at the "LI" prompt, while newer ESSG systems from 4.1 onward (that use a newer LILO version) install and boot just fine, and the backup and restore tests work very well - of course, these are just "formal" tests: the restored systems work as expected, but without the load and the multiplicity of requests that a production machine goes under I just can't state it works for sure. If somebody is interested in this kind of recovery procedure ( If someone is interested in [testing] this procedure, I could detail it in a how-to (at least, a sort of... ) by translating the notes I already made... let me know. -- Pierluigi Miranda -- Please report bugs to [EMAIL PROTECTED] Please mail [EMAIL PROTECTED] (only) to discuss security issues Support for registered customers and partners to [EMAIL PROTECTED] To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED] Archives by mail and http://www.mail-archive.com/devinfo%40lists.e-smith.org
