Help me modify the config on a necessarily dissembodied solaris drive
March 27, 2006 8:00 PM Subscribe
I'm looking for a (preferably) linux based livecd with read and write support for sun UFS.
Recently a friend gave me a sun ultra 5. This machine functions perfectly well (albeit headless) and runs numerous services, etc. once it gets running.
Of course, I don't have the appropriate null modem cable to access it via serial console, nor do I have a compatible monitor. Which doesn't bother me too much, as once this is set up the way I'd like it, it'll run headless anyhow.
It currently runs services such as rlogin and telnet, so it should be quite achievable to do all the setup I need over the house's LAN.
The root and core of the problem is this (other than my total lack of experience with Sun machines): There are no user accounts, and I need to enable remote root login anyhow to achieve the various installations and configurations that I need to do before this machine supersedes our tired tired webserver.
Unfortunately /etc/default/login has CONSOLE set to /dev/console. I need to comment this out to enable remote root logins temporarily (one whole # would do the trick of course) but I can't get read/write access from any livecd I've tried thusfar. I can't really see any practical means by which I could make this configuration change at the machine it's self .
Recently a friend gave me a sun ultra 5. This machine functions perfectly well (albeit headless) and runs numerous services, etc. once it gets running.
Of course, I don't have the appropriate null modem cable to access it via serial console, nor do I have a compatible monitor. Which doesn't bother me too much, as once this is set up the way I'd like it, it'll run headless anyhow.
It currently runs services such as rlogin and telnet, so it should be quite achievable to do all the setup I need over the house's LAN.
The root and core of the problem is this (other than my total lack of experience with Sun machines): There are no user accounts, and I need to enable remote root login anyhow to achieve the various installations and configurations that I need to do before this machine supersedes our tired tired webserver.
Unfortunately /etc/default/login has CONSOLE set to /dev/console. I need to comment this out to enable remote root logins temporarily (one whole # would do the trick of course) but I can't get read/write access from any livecd I've tried thusfar. I can't really see any practical means by which I could make this configuration change at the machine it's self .
Response by poster: I've got belenix burnt and have booted into it a few times, but I've not been able to get it to mount the drive (hdb1 in linux should be equivelent to c0d1s0 in solaris, eh?)
I've gotten various errors attempting to mount with the utilities on a solaris 10 cd, belenix and schillix.
posted by Matt Oneiros at 8:56 PM on March 27, 2006
I've gotten various errors attempting to mount with the utilities on a solaris 10 cd, belenix and schillix.
posted by Matt Oneiros at 8:56 PM on March 27, 2006
Response by poster: bash-3.00# mount -F ufs /dev/dsk/c0d1p0 /mnt/disk
mount: /dev/dsk/c0d1p0 is not this fstype
fdisk /dev/rdsk/c0d0p0 shows the two ntfs partitions I'd assume on the first disk on that controller, while fdisk for /dev/rdsk/c0d1p0 shows no partitions under solaris.
Under linux, mount -t ufs /dev/hdb1 /mnt/hdb1 works, and fdisk of /dev/hdb shows a solaris partition and a whole disk partition, like I'd expect to see.
Once again under solaris (just got the idea to try this),
bash-3.00# fsck -F ufs /dev/rdsk/c0d1p0
** /dev/rdsk/c0d1p0
BAD SUPERBLOCK AT BLOCK 16: MAGIC NUMBER WRONG
LOOK FOR ALTERNATE SUPERBLOCKS WITH MKFS?
Would allowing fsck to fix it take care of this?
posted by Matt Oneiros at 10:57 PM on March 27, 2006
mount: /dev/dsk/c0d1p0 is not this fstype
fdisk /dev/rdsk/c0d0p0 shows the two ntfs partitions I'd assume on the first disk on that controller, while fdisk for /dev/rdsk/c0d1p0 shows no partitions under solaris.
Under linux, mount -t ufs /dev/hdb1 /mnt/hdb1 works, and fdisk of /dev/hdb shows a solaris partition and a whole disk partition, like I'd expect to see.
Once again under solaris (just got the idea to try this),
bash-3.00# fsck -F ufs /dev/rdsk/c0d1p0
** /dev/rdsk/c0d1p0
BAD SUPERBLOCK AT BLOCK 16: MAGIC NUMBER WRONG
LOOK FOR ALTERNATE SUPERBLOCKS WITH MKFS?
Would allowing fsck to fix it take care of this?
posted by Matt Oneiros at 10:57 PM on March 27, 2006
You say that under linux, mount -t ufs /dev/hdb1 works. So it it only mounting ro, or what?
posted by flabdablet at 3:58 AM on March 28, 2006
posted by flabdablet at 3:58 AM on March 28, 2006
Best answer: You're going to have endian problems mounting a UFS filesystem created on a SPARC machine on an Intel host. A working serial connection will be needed one day when the machine won't reboot and come back online and you don't know why.
posted by joelr at 5:12 AM on March 28, 2006
posted by joelr at 5:12 AM on March 28, 2006
Response by poster: Yeah, linux will only mount it ro, and some distros are also recquiring -o ufstype=sun to mount it.
Even though one day I will need the serial connection, is there any concievable way to make the changes I need now and worry about serial later?
For some reason I thought machines were capable of working around the endian thing now.
posted by Matt Oneiros at 11:50 AM on March 28, 2006
Even though one day I will need the serial connection, is there any concievable way to make the changes I need now and worry about serial later?
For some reason I thought machines were capable of working around the endian thing now.
posted by Matt Oneiros at 11:50 AM on March 28, 2006
This thread is closed to new comments.
posted by Loto at 8:45 PM on March 27, 2006