NetBSD Problem Report #28588

From carl@rollcage2.bl.echidna.id.au  Thu Dec  9 00:39:07 2004
Return-Path: <carl@rollcage2.bl.echidna.id.au>
Received: from rollcage2.bl.echidna.id.au (rollcage.bl.echidna.id.au [203.6.241.134])
	by narn.netbsd.org (Postfix) with ESMTP id C4EAC251EB0
	for <gnats-bugs@gnats.NetBSD.org>; Thu,  9 Dec 2004 00:39:06 +0000 (UTC)
Message-Id: <200412090038.iB90cxFD007882@rollcage2.bl.echidna.id.au>
Date: Thu, 9 Dec 2004 11:38:59 +1100 (EST)
From: Carl Brewer <carl@rollcage2.bl.echidna.id.au>
Reply-To: carl@rollcage2.bl.echidna.id.au
To: gnats-bugs@netbsd.org
Subject: update the guide for setting up RAIDframe on 2.0
X-Send-Pr-Version: 3.95

>Number:         28588
>Category:       port-i386
>Synopsis:       update the guide for setting up RAIDframe on 2.0
>Confidential:   no
>Severity:       non-critical
>Priority:       low
>Responsible:    www
>State:          open
>Class:          doc-bug
>Submitter-Id:   net
>Arrival-Date:   Thu Dec 09 00:42:00 +0000 2004
>Closed-Date:    
>Last-Modified:  Fri Jan 24 18:47:37 +0000 2020
>Originator:     carl@rollcage2.bl.echidna.id.au
>Release:        NetBSD 2.0
>Organization:

>Environment:


System: NetBSD 2.0 NetBSD 2.0 (GENERIC) #0: Wed Dec 1 10:58:25 UTC 2004 builds@build:/big/builds/ab/netbsd-2-0-RELEASE/i386/200411300000Z-obj/big/builds/ab/netbsd-2-0-RELEASE/src/sys/arch/i386/compile/GENERIC i386
Architecture: i386
Machine: i386
>Description:

        When following the NetBSD guide to set up RAIDframe,
        disklabel creates an "e" partition by default.  Please
        update the doco to reflect that this "e" partition must
        be removed or the box won't boot!

>How-To-Repeat:
>Fix:
        Add to the doco that the non-RAID partition listed by
        disklabel when setting up RAID1 following the guide
        must be removed (it's 4.2BSD format)


>Release-Note:

>Audit-Trail:

Responsible-Changed-From-To: port-i386-maintainer->hubertf
Responsible-Changed-By: hubertf@netbsd.org
Responsible-Changed-When: Sun, 26 Nov 2006 16:47:24 +0000
Responsible-Changed-Why:

Sorry for the late reply...

Can you please give more details of this - esp. what
disklabel do you mean, exactly?
The raid0 one? To my understanding (and by what I've experieenced), there
is no problem with having extra partitions defined in a disklabel
(if they are created at all, I'm not sure this is still true about
the latest NetBSD, can you confirm this?)
Booting shouldn't be affected by this in any case.

Can you point me at the passage of http://www.netbsd.org/guide/en/chap-rf.html
where you think care should be taken for this case?


 - Hubert


State-Changed-From-To: open->feedback
State-Changed-By: hubertf@netbsd.org
State-Changed-When: Sun, 26 Nov 2006 16:47:24 +0000
State-Changed-Why:
need more data


From: Hubert Feyrer <hubertf@NetBSD.org>
To: gnats-bugs@NetBSD.org, carl@rollcage2.bl.echidna.id.au
Cc: 
Subject: Re: port-i386/28588
Date: Sun, 26 Nov 2006 17:47:16 +0100 (CET)

 I'm not sure this came over properly, so just in case:

 Sorry for the late reply...

 Can you please give more details of this - esp. what
 disklabel do you mean, exactly?
 The raid0 one? To my understanding (and by what I've experieenced), there
 is no problem with having extra partitions defined in a disklabel
 (if they are created at all, I'm not sure this is still true about
 the latest NetBSD, can you confirm this?)
 Booting shouldn't be affected by this in any case.

 Can you point me at the passage of 
 http://www.netbsd.org/guide/en/chap-rf.html
 where you think care should be taken for this case?


   - Hubert

From: Havard Eidnes <he@NetBSD.org>
To: hubertf@NetBSD.org
Cc: gnats-admin@netbsd.org, netbsd-bugs@netbsd.org,
	carl@rollcage2.bl.echidna.id.au, gnats-bugs@NetBSD.org
Subject: Re: port-i386/28588
Date: Fri, 15 Dec 2006 12:25:44 +0100 (CET)

 Hi,

 seeing this PR reminded me that the procedure as stated in the
 NetBSD guide is horribly complicated -- more so than it needs to
 be, with "failing a component" and all that stuff.

 Here is the procedure I have used when installing NetBSD/i386 on
 a mirrored set of 2 disks, admittedly with 3.0, but I think it
 should work for 2.0 as well as long as it has support for booting
 from a mirrored raid set.

 The procedure I used requires some use of tools outside of
 sysinst before sysinst is run, and a bit of manual calculation
 (for partition sizes), but is conceptually straight-forward:

 1) boot NetBSD from boot media, typically CD-ROM

 2) Initialize NetBSD fdisk partitions on the two drives, in my
    case the two drives are wd0 and wd1:

    fdisk -a -i -u -B wd0

    I used "entire drive, save the 63 first sectors, made the
    NetBSD partition the active partition, used mbr_bootsel, and
    used NetBSD-0 as the menu entry, boot from first active
    partition.

    Repeat for wd1, I used NetBSD-1 there.

 3) Put a disklabel on both drives:

    Use "disklabel wd0" to get the "default" label and see the
    total size, in my case 156312576 sectors, and initialize +
    edit the label:

    disklabel -I -e wd0

    I used:

    a: 156312513    63     RAID

    and let "c" cover the same area and "d" the entire disk.

    Repeat for wd1.

 4) Make a configuration file for raidframe:

    ed /etc/raid.conf
    a
    START array
    # numRow numCol numSpare
    1 2 0

    START disks
    /dev/wd0a
    /dev/wd1a

    START layout
    # sectPerSU SUsPerParityUnit SUsPerReconUnit RAID_level_1
    64 1 1 1

    START fifo
    fifo 100
    .=

    w
    q

    (Indented for readability here.)

 5) Configure the raid set, and set it to auto-config and
    root-on-raid:

    raidctl -C /etc/raid.conf raid0
    raidctl -I 4711 raid0
    raidctl -i raid0
    raidctl -A yes raid0
    raidctl -A root raid0

 6) Optional, check parity regeneration status:

    raidctl -S raid0

    You may interrupt the display with ^C.

 7) Use sysinst to install NetBSD to the newly configured raid0
    raid set in the normal manner.  Sysinst should list raid0 as a
    usable drive.

    sysinst

 8) Install the NetBSD boot code on rwd0a and rwd1a.  Sysinst has
    not left any boot code there, only inside raid0, and this is
    required to be able to boot from the raid set:

    installboot -v -o console=3Dcom0,timeout=3D10 /dev/rwd0a /usr/mdec/b=
 ootxx_ffsv1
    installboot -v -o console=3Dcom0,timeout=3D10 /dev/rwd1a /usr/mdec/b=
 ootxx_ffsv1

 9) Remove boot media, reboot the system, and ensure that the
    following is either part of the defaults or is placed in
    rc.conf:

    # Allow clean raid unconfigure on shutdown
    swapoff=3DYES

 And you're done.

 None of that complicated "artificially fail one component,
 restore to new" stuff required.

 Ideally, sysinst should be taught to guide the user through this
 process.

 Best regards,

 - H=E5vard

From: Jussi Liukkonen <jussi.liukkonen@snorkkeli.homeip.net>
To: gnats-bugs@netbsd.org
Cc: 
Subject: Re: port-i386/28588
Date: Tue, 19 Dec 2006 01:16:22 +0200

 On a related and potentially interesting note, here <http:// 
 snorkkeli.homeip.net/~jml/stuff/nb-raid-installguide.txt> are the  
 RAID-1 installation/recovery instructions we use in-house, very  
 similar to Havard's. Ours are tuned for a non-savvy support person to  
 be able to rebuild a system from tape backups. Some kind of basic  
 RAID-1 setup option in sysinstall would indeed automate a good  
 portion of the manual -- equalling error prone -- stuff.

State-Changed-From-To: feedback->open
State-Changed-By: dholland@NetBSD.org
State-Changed-When: Thu, 10 Apr 2008 03:20:24 +0000
State-Changed-Why:
feedback is not forthcoming, but the subsequent comments about the docs
merit consideration.


Responsible-Changed-From-To: hubertf->www
Responsible-Changed-By: hubertf@NetBSD.org
Responsible-Changed-When: Fri, 24 Jan 2020 18:47:37 +0000
Responsible-Changed-Why:
I can't handle this. www@ please take over or re-assign.


>Unformatted:

Home	PR Database Search