How to handle large number of disk devices in "arbitrary" order?

2011-11-14 Thread Carsten Aulbert
Hi all first of two questions (second will come later in a separate email). We have a couple of machines with a largish number of disk drives (currently up to 48) plus one or more "flash drives" - could be USB thumb drive, SATA disk on module, SSD, whatever. Of course the order in which these a

How to handle large number of disk devices in "arbitrary" order

2013-11-20 Thread Thomas Lange
Hi all, in late 2011 there was a discussion on linux-fai about: > How to handle large number of disk devices in "arbitrary" order? > Carsten Aulbert carsten at welcomes-you.com > Mon Nov 14 16:54:24 CET 2011 > We have a couple of machines with a largish number of d

Re: How to handle large number of disk devices in "arbitrary" order?

2011-11-14 Thread Markus Koeberl
On Monday 14 November 2011 16:54:24 Carsten Aulbert wrote: > Hi all > > first of two questions (second will come later in a separate email). > > We have a couple of machines with a largish number of disk drives > (currently up to 48) plus one or more "flash drives" - could be USB thumb > drive, S

Re: How to handle large number of disk devices in "arbitrary" order?

2011-11-14 Thread Carsten Aulbert
Hi Markus On Monday 14 November 2011 18:22:22 Markus Koeberl wrote: > what about /dev/disk/by-id and /dev/disk/by-path ? > > It seams to be supported by setup-storage: > /* fully qualified device-path or short form, like hda, whereby full > * path is assumed to be /dev/hda; may contain shell glob

Re: How to handle large number of disk devices in "arbitrary" order?

2011-11-14 Thread Carsten Aulbert
Good morning > > FAI's error message is: > Starting setup-storage 1.3 > Device name /dev/disk/by-id/scsi-SATA_DeLOCK.* could not be substituted > > (the same without the '.' did also fail) But fails differently (thus I assume /dev/disk/by-id/scsi-SATA_DeLOCK*) should be the correct way. The e

Re: How to handle large number of disk devices in "arbitrary" order?

2011-11-15 Thread Carsten Aulbert
Hi On Tuesday 15 November 2011 07:22:52 Carsten Aulbert wrote: > > I suppose I'll have to dig into task_partition and the helper scripts in > there. that was easier than I thought: From Parser.pm: 95- # test, whether the device name starts with a / and prepend /dev/, if 96- # appropriate 97-

Re: How to handle large number of disk devices in "arbitrary" order?

2011-11-15 Thread Michael Tautschnig
Hi Carsten, [...] > > But fails differently (thus I assume /dev/disk/by-id/scsi-SATA_DeLOCK*) > should > be the correct way. > > The error then is > > Failed to resolve /dev/disk/by-id/scsi-SATA_DeLOCK* to a unique device name > which of course is true, if the device was already partitioned

Re: How to handle large number of disk devices in "arbitrary" order?

2011-11-15 Thread Carsten Aulbert
Hi Michael On Tuesday 15 November 2011 09:22:44 Michael Tautschnig wrote: > > I suppose you should be able to address this issue by using > > /dev/disk/by-id/scsi-SATA_DeLOCK*[A-Z0-9_][A-Z0-9_][A-Z0-9_] > > but obviously this isn't exactly beautiful. It might be easier if we > instead permitted

Re: How to handle large number of disk devices in "arbitrary" order?

2011-11-15 Thread Michael Tautschnig
Hi Carsten, [...] > > Interestingly an additional '1' is added to the device name - is that for the > first > partition (sda1) which is invalid for this way of addressing the device? > > the "glob regexp" does indeed match nicely: > root@ra01:/tmp/fai# perl -e 'print join("\n", > glob("/dev/di

Re: How to handle large number of disk devices in "arbitrary" order?

2011-11-15 Thread Carsten Aulbert
Hi On Tuesday 15 November 2011 10:07:21 Michael Tautschnig wrote: > Would you mind patching your copy of setup-storage in order to further > debug this? It should suffice to change line 100 of Parser.pm to > > $disk = `readlink -f $candidates[0]` if (scalar(@candidates) == 1); > > [was: $disk =

Re: How to handle large number of disk devices in "arbitrary" order?

2011-11-15 Thread Michael Tautschnig
Hi again, [...] > > That's working > [...] Cool! > Although a proper 'lstat' might be more perlish ;) > Hmm, how would that best be implemented? (I'm always open to all kinds of perl hints!!) > I'll see how the remaining install will proceed. After that I'll try using > setup-storage's 'md

Re: How to handle large number of disk devices in "arbitrary" order?

2011-11-15 Thread Thomas Lange
> On Mon, 14 Nov 2011 16:54:24 +0100, Carsten Aulbert > said: > We have a couple of machines with a largish number of disk drives (currently > up to 48) plus one or more "flash drives" - could be USB thumb drive, SATA > disk on module, SSD, whatever. Of course the order i

Re: How to handle large number of disk devices in "arbitrary" order?

2011-11-15 Thread Carsten Aulbert
Hi On Tuesday 15 November 2011 11:14:09 Thomas Lange wrote: > > A different approach is to manipulate the shell variable $disklist. > This variable is used by setup-storage and can be set by any script > called class/*.var or class/*.source > > In such a script you could do anything that ensures

Re: How to handle large number of disk devices in "arbitrary" order?

2011-11-15 Thread Thomas Lange
> On Tue, 15 Nov 2011 11:50:04 +0100, Carsten Aulbert > said: > Is there a way to force FAI to use UUID in /etc/fstab as well as for grub? Yes. fstabkey:uuid E.g.: disk_config sda fstabkey:uuid preserve_reinstall:8,9 -- regards Thomas

Re: How to handle large number of disk devices in "arbitrary" order?

2011-11-15 Thread Carsten Aulbert
Hi On Tuesday 15 November 2011 11:56:28 Thomas Lange wrote: > > On Tue, 15 Nov 2011 11:50:04 +0100, Carsten Aulbert said: > > Is there a way to force FAI to use UUID in /etc/fstab as well as for > > grub? > > Yes. fstabkey:uuid > > E.g.: disk_config sda fstabkey:uuid preserve_

Re: How to handle large number of disk devices in "arbitrary" order?

2011-11-15 Thread Michael Tautschnig
Hi Carsten, [...] > > Is there a way to force FAI to use UUID in /etc/fstab as well as for grub? > For fstab just use fstabkey:uuid in your disk_config line (which will be the default 4.0). For grub you will have to use some local hacks AFAIK. I'm using for grub 1 in boot/grub/menu.lst/preinst

Re: How to handle large number of disk devices in "arbitrary" order?

2011-11-16 Thread Carsten Aulbert
On Tuesday 15 November 2011 12:02:25 Carsten Aulbert wrote: > Hi > > On Tuesday 15 November 2011 11:56:28 Thomas Lange wrote: > > > On Tue, 15 Nov 2011 11:50:04 +0100, Carsten Aulbert > > said: > > > Is there a way to force FAI to use UUID in /etc/fstab as well as > > > for grub? > >

Re: How to handle large number of disk devices in "arbitrary" order?

2011-11-16 Thread Carsten Aulbert
Hi On Tuesday 15 November 2011 10:26:12 Michael Tautschnig wrote: > > Although a proper 'lstat' might be more perlish ;) > > Hmm, how would that best be implemented? (I'm always open to all kinds of > perl hints!!) Actually, Cwd might help here, what about: diff --git a/lib/setup-storage/Pars

Re: How to handle large number of disk devices in "arbitrary" order?

2011-11-17 Thread Holger Parplies
Hi, Carsten Aulbert wrote on 2011-11-17 08:39:45 +0100 [Re: How to handle large number of disk devices in "arbitrary" order?]: > [...] > If so, should not any trailing /-part[0-9]+/ be cut out/ignored from the > potential list of candidates? E.g. (fully untested, just r

Re: How to handle large number of disk devices in "arbitrary" order?

2011-11-18 Thread Carsten Aulbert
Hi On Friday 18 November 2011 02:49:18 Holger Parplies wrote: > talking of 'perlish', isn't that > > my @candidates = grep { not /-part[0-9]+$/ } glob($disk); > even better! And actually much more readable... hmm, does it then really qualify for 'perl-ish'? ;) Cheers Carsten

Re: How to handle large number of disk devices in "arbitrary" order?

2011-11-18 Thread Carsten Aulbert
Hi Michael On Tuesday 15 November 2011 10:26:12 Michael Tautschnig wrote: > > > I'll see how the remaining install will proceed. After that I'll try > > using setup-storage's 'md' interface with the 12 disk drives > > Just a hint here: you might want to use sameas:some-other-disk to simplify > t