RE: [SSI-users] Unable to Configure the New Node with a Local BootDevice on OpenSSI 0.9.6 r3

[<den...@sd...>]

Hi,

Yes, I did run the mkfs on the console of the second node. Here below is
the contents of /proc/mounts

rootfs / rootfs rw 0 0
00000021 / cfs rw 0 0
00000028 /cluster/nodenum1/dev cfs rw 0 0
00000028 /dev cfs rw 0 0
none /cluster/nodenum2/dev cfs rw 0 0
none /dev cfs rw 0 0
/proc /proc proc rw 0 0
usbdevfs /proc/bus/usb usbdevfs rw 0 0
00000021 /etc/sysconfig/network-scripts cfs rw 0 0

and /etc/mtab

/dev/sda1 / ext3 rw 0 0
none /proc proc rw 0 0
usbdevfs /proc/bus/usb usbdevfs rw 0 0
devfs /dev devfs rw 0 0
usbdevfs /proc/bus/usb usbdevfs rw 0 0
/dev/fd0 /mnt/floppy vfat rw,nosuid,nodev 0 0

Regards,

Denny

Business Development Dept.
Dimension Computer Technology Co., Ltd.
>> an e-Solution Provider <<

Phone: +886 2 2501 5568 ext. 349
Fax: +886 2 2505 5134
e-mail: den...@sd...
Web: http://www.dimension.com.tw



-----Original Message-----
From: John Byrne [mailto:joh...@hp...]
Sent: Friday, May 09, 2003 7:50 AM
To: Brian J. Watson
Cc: =A9P=A9=A7=A5=C1; ssi...@li...
Subject: Re: [SSI-users] Unable to Configure the New Node with a Local
BootDevice on OpenSSI 0.9.6 r3


Brian J. Watson wrote:
> =A9P=A9=A7=A5=C1 wrote:
>=20
>>Hi,
>>
>>I tried to configure the new node with a local boot device on OpenSSI=20
>>0.9.6 release 3 but got some problems.
>>
>>1. On the new joined nodes, after creating /dev/sda1 by fdisk, I got
the=20
>>message list below when running mkfs to make a filesystem on it.
>>
>>/dev/sda1 is mounted; will not make a filesystem here!


I don't get this behavior. Are you running mkfs from the console of the
second node? Can you send the /proc/mounts and /etc/mtab as seen from
the console of the second node?

John


>>
>>   If the new created disk partition is not /dev/sda1 (e.g. /dev/sda3
or=20
>>/dev/hda1), then the mkfs runs correctly on the local hard drive. It=20
>>seems that the kernel treats /dev/sda1 as the root partition even it's
a=20
>>local device of the new nodes.
>=20
>=20
> I don't know why this would happen. John?
>=20
>=20
>>2. chnode should run on any node, but in the node without /dev/sda1
(the=20
>>root device in my environment), there will be an error message list
below.
>>
>>addnode: fatal error:
>>  Field 5 (boot device) of line 26 of /etc/clustertab must refer to a=20
>>block device.
>>
>>  Line 26 of /etc/clustertab describe the first (root) node
configuration.
>=20
>=20
> You're right. I've made a note to myself to fix this sometime.
>=20
>=20
>>3. I'd also tried to configure a new node with /dev/sda1 and /dev/sda4

>>and run chnode just on the node to configure /dev/sda4 as the local
boot=20
>>device. chnode run without the messages list above but got the
message:
>>
>>Error migrating to node1. Try again!
>=20
>=20
> I've seen this before. This is happening in the cluster_lilo script,
> which migrates to each node with a local boot device, writes a custom
> /etc/lilo.conf for that node, then runs lilo. I don't know why it's
> failing to migrate, but you can work around it by trying to run
> cluster_lilo from different nodes on the cluster until it works.
>=20
> Hope this helps,
>=20
> Brian
>=20
>=20