>> Does one have to manually update /etc/raidtab to show which disks are now
>> "spare" and which are the main disks ?
> raidhotadd doesn't consult the raidtab file, all the info it needs is
> contained in the raid superblock. Since the disk you add gets used right
> away the info in /etc/raidtab should be correct (after reconstruction
> finishes).
Really ?
I would have thought that if I had 1 and 2 in use and 3 as spare,
when 1 fails, I get 2 and3 in use, and that when I insert a good replacement
for 1, it becomes the spare -- is that not the case ?
Will it actually copy things to 1, and then release 3 back to being spare ?
> So: If the new disk is on a different device than the failed one, I'd
> update raidtab at some point to keep info consistent; the file will be
> consulted for manual raidstart / raidstop operations.
So there are two completely distinct operations:
1) kernel auto detect
2) manual raist*
:-(
> next, install raid stuff, create raid mirror with /dev/sda1 entered as
> "failed-disk" in raidtab
Please explain `"entered as failed-disk" in raidtab' ...
> copy your system to the raid device
How ? using dd or what ?
> and finish by raidhotadding /dev/md0 /dev/sda1
What are the args to raidhotadd ?
# raidhotadd /dev/md0 /dev/hda7
/dev/md0: can not hot-add disk: invalid argument.
#
Mar 18 14:51:58 dean kernel: trying to hot-add hda7 to md0 ...
Mar 18 14:51:58 dean kernel: md0: personality does not support diskops!
>> Also, how do I add a persistent super block to an old linear or RAID0 FS ?
> Sorry, don't know if this can be done.
So if I am running RAID0 or linear using 0.50, all I can do is dump and
restore? For a RAID0 I would expect to be able to shrink the FS (so that the
end of the underlying partitions are free) and then just write a Superblock ...
>> Does anyone have utilities to peek/poke SBs ?
>> If not, should I write private perl scripts to do the HACKs I want, or
>> should I write some C code to contribute to the package -- or would it be
>> too dangerous to let loose ? Things like:
> you can find out quite a lot by doing a mkraid --debug.
Under 0.90 ?
I don't have "--debug" :-(
Does it actually write anything, or just read ?
> This will dump your current raid setup to syslog.
Ta, but as I indicated, I want somewhat more ...
PS: I don't have linear, raid0 or raid1 compiled into my kernel, so I load
them as modules in initrd. There seems to be some real magic in there, as
before the inird is processed, md gets in, decides it doesn't have the
personalities, and quits. Later, when initrd has them in, it tries again and
works -- magic :-)
However, this is a pain, as there is no need, as root is not RAID. I was
hoping that I could start the boot, get a working FS, then have the kernel
poke around for 0xfd partitions, and then while starting them, use kmod to
load the personalities. Can this be done ?
