Michael J McCafferty wrote:
All,
        I have a customer who has a lot of time invested in configuring his
system and is ready to put it in to production. However, we are getting
some disk errors on occasion and it's tough to trust it for the long
haul. What is the hot ticket for putting the exact same thing on another
drive. It's easy to have another identical server sitting next to it at
the same time, or put an identical disk in the same system to copy stuff
over.

        I am sure I could spend the time to do dd and fdisk, etc, but dumb it
down and speed it up for me. :o)  Hmm... can I put a second disk in the
system and do a dd if=/dev/hda of=/dev/hdb and will that get the boot
sectors, etc ? Is it that simple ?

Thanks,
Mike

PS: The customer is aware that RAID was probably a good idea here, but
he chose not to get it for a variety of reasons and here we are. :o)



Feel your pain. Haven't had this issue in years. :) Anyway, here's a couple recommendations for future.

A. Raid. :)
B. Build environment. If they're going to roll out multiple servers, or even a server, a build environment works great. PXE, dhcp, tftp. If you're configuring servers for customers, you should have this. I can't remember the _LAST_ time I installed off a cdrom. Hit the power, 5 minutes later the box is up and configured with whatever purpose it is supposed to be. C. cfengine or puppet, with _ALL_ configs under revision control. I use cfengine/svn, ymmv. D. On configuring apache, I love includes, all my apache configs are bone stock with a *.conf file in the conf.d directory which apache reads on boot. E. Keep it as stock as possible. In production, we run all bone stock RPM's with our configs. This makes it easy to update for security fixes, etc. If I can't use a stock RPM or one isn't available, I have a custom repo that I use for what's needed. Same with .deb's. \ F. Backups... Back up to tape, back up to disk, etc. Disk is cheap, and in my case I snapshot from the data center to the office, and the datacenter is also backing up to tape. Soooo, if something dies, I have a local disk copy, then a remote disk copy, then I have to pester the guys that run the tape library, which is pretty much never.

If there's not a presentation this month, I can do a basic presentation on all of this.

Mark


--
[email protected]
http://www.kernel-panic.org/cgi-bin/mailman/listinfo/kplug-list

Reply via email to