setup storage error on multi partition software raid

[Michael Tautschnig]

Hi,

> Hi,
> 
> Attached please find the debug fai.log from a failed attempt using setup-storage using a simple machine with two IDE drives, hoping to setup software raid. The logs indicate that everything seems to be working well, assigning the appropriate /dev/md number to the proper /dev/hda and hdb partitions until line 881, where (if I am interpreting the logs correctly) I see that
> 
[...]

> 
> Is the fatal error due to the existing configuration as complained about on line 881?
> 

The error message appearing in line 881 is just a result of stdout and stderr
being merged, so actually it should have appeared at the end of the Trying to
add ... list.

Actually it took me some time of carefully walking these dependencies to see the
error, so clearly it would be nice to produce better error explanations. Like
this one: Your config is broken :-) Well, no, what you'd really want is: There
is no hda4/hdb4 in your system. Why is that so?

From your log file:

> Using config file: /var/lib/fai/config/disk_config/FAIBASE
> Input was:
> # example of new config file for setup-storage
> #
> # <type> <mountpoint> <size>   <fs type> <mount options> <misc options>
> 
> disk_config hda bootable:1
> primary -       95      -       -
> logical -       1700    -       -
> primary swap    2048    -       -
> logical -       2500    -       -
> logical -       1900    -       -
> logical -       4100    -       -
> logical -       936     -       -
> logical -       2000-   -       -
> 
[...] (same for hdb)

It may bit a bit surprising, but setup-storage always only generates one
extended partition. Let's assign "names" A,B,C,... to your partitions with A
being "primary - 95...", B is "logical - 1700", C is "primary swap ..." and so
forth. For a config such as yours setup-storage builds:

hda1: A
hda2: extended
  hda5: B
  hda6: D
  hda7: E
  hda8: F
  hda9: G
  hda10: H
hda3: C

This may or may not be what you desired.

> 
> disk_config raid
> raid1   /boot   hda1,hdb1       ext3    rw,noatime,nosuid,nodev  createopts="-m15" tuneopts="-c 0 -i 0"
> raid1   /       hda2,hdb2       ext3    rw,errors=remount-ro
> raid1   swap    hda3,hdb3       swap    rw
> raid1   /var    hda4,hdb4       ext3    rw,noatime,nosuid,nodev  createopts="-m15" tuneopts="-c 0 -i 0"
> raid1   /tmp    hda5,hdb5       ext3    rw,noatime,nosuid,nodev  createopts="-m15" tuneopts="-c 0 -i 0"
> raid1   /usr    hda6,hdb6       ext3    rw,noatime,nosuid,nodev  createopts="-m15" tuneopts="-c 0 -i 0"
> raid1   /home   hda7,hdb7       ext3    rw,noatime,nosuid,nodev  createopts="-m15" tuneopts="-c 0 -i 0"
> raid1   /app    hda8,hdb8       ext3    rw,noatime,nosuid,nodev  createopts="-m15" tuneopts="-c 0 -i 0"

As shown above, at least hda4 would never exist. In fact, you also better not
use hda2 (the extended partition). Your numbering would only be consistent if
you were using a GPT partition table (using disklabel:gpt or disklabel:gpt-bios)
with all partitions being primary once (there is no concept of logical
partitions in GPT).

Hope this helps,
Michael

-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 833 bytes
Desc: not available
URL: <http://lists.uni-koeln.de/pipermail/linux-fai/attachments/20101116/df695ffa/attachment.bin>