Hi, > hi, that is 3.4.7 (debian 6 stable). Is there anything that I can do to > workarround this bug? > or can I provide you with more info? (my hook script that creates the raid > arrays can be found at > https://github.com/aswen/scripts/blob/master/partition.DB-server > the fdisk -l command at the end shows four disks... > regards > [...]
Although there definitively are some bugs as this problem should be diagnosed in a sane way, the problem should be solvable on your side: setup-storage relies on the variable disklist to carry the list of devices that could possibly be made use of. Now this variable is set by FAI, but that happens quite early on. As you only create the devices in this hook, they won't end up in the disklist variable. The solution should be as follows: - Rename your hook to partition.DB-server.source - As a result, your hook will not be executed in a subshell, but will be sourced instead. - At the very end add (possibly with a sleep or more correctly a call to udevsettle in between) a call to set_disk_info. This is a built-in routine of FAI that will properly set the variable disklist, which is then used by setup-storage. Hope this helps, Michael
pgpgsLawSxkuo.pgp
Description: PGP signature
