----- Original Message -----
From: "Hylton Conacher (ZR1HPC)"
2008/9/22 Andrew Joakimsen
: On Mon, Sep 22, 2008 at 12:38 PM, Greg Freemyer
wrote: <snip>
No because I've already "been there" and "done that" -- It is IMPOSSIBLE to recover a RAID-1 array that just one day (after properly shutting down the system and not having hardware failures) decided not to boot.
Ahhh, now this is interesting :) The main reason I was creating the RAID1 array was so that if one drive failed, I could receive a message from mdadm, swop out the bad drive and continue working, without losing data.
In view of the above comment, I guess it is time I ordered the 3rd HDD and rather setup RAID5.
You can get a message when a drive fails if you enable the mdadmind service, and you can boot up a raid1, 5 or 10 with one drive dead or gone, no problem. It is more finnicky than hardware raid in that it's up to you to ensure that /boot and the mbr are always either on a single plain drive or on raid1 or raid10, and it's up to you to ensure that if your normal booting drive happens to be the one that dies, that the other drives also have a valid mbr and active partition so that when the motherboard bios falls back to boot from some other drive, it can. I ensure both of those requirements just by putting this in /etc/grub.conf, (8-drive system with a small /boot partition on each drive, all 8 in raid1, so, all 8 partitions are identical copies of each other.) setup (hd0) (hd0,0) setup (hd1) (hd1,0) setup (hd2) (hd2,0) setup (hd3) (hd3,0) setup (hd4) (hd4,0) setup (hd5) (hd5,0) setup (hd6) (hd6,0) setup (hd7) (hd7,0) quit That's it. That causes grub to go into the mbr of each drive, and tells grub that /boot is the first partition on each drive. This edit, together with the fact that /boot is a raid1 including all 8 1st partitions, ensures that any drive may boot the system. I must do this edit manually and outside of yast, because yast refuses to leave it alone even when I use the expert option to manually edit the same file via yast. But, if you edit it outside of yast and then don't touch the bootloader dialogs in yast, then yast does leave the file alone even when doing kernel upgrades (in which yast edits menu.lst automatically) and running mkinitrd. These are issues you don't have to worry about with hardware raid, but I don't consider them burdensome. The advantage goes further than the cost of one hardware raid card. The advantage is you can do, redo, copy, or fix this anywhere any time on any machine. If you had all the money in the world it's still a pain dealing with hardware raid simply because grocery stores don't sell hardware raid cards, nor does any local store. If there is an electronics or computer shop anywhere near you that sells even one hardware raid card you are extrememly lucky, and then you almost certainly don't have a selection but are stuck with whatever they have. Think 3ware are junk like me? Oh well order the adaptec or lsi on line and wait... Think adaptec are over priced? Oh well order the 3ware on line and wait... You need pci and they only have pci-e? oh well etc etc There will always be some special cases where it's possible to trick the system into failing by some unfortunate coincident sequence of events. But thats true for hardware raid too. -- Brian K. White brian@aljex.com http://www.myspace.com/KEYofR +++++[>+++[>+++++>+++++++<<-]<-]>>+.>.+++++.+++++++.-.[>+<---]>++. filePro BBx Linux SCO FreeBSD #callahans Satriani Filk! -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org