Topic: Clean Install 5.6 --> Updater2 --> boot stops at LI

[Robert Harlow]

Sirs

I *lost* the LI(LO) earlier this week and I wondered why. Now I know; it must of been because I earlier ran the Updater2 on an existing 5.6 server-gateway.

Earlier I assumed I must've killed the boot process or something so I decided to kill the whole thing and clean install it all this afternoon:-| Depowered briefly to reconnect the other 7 drives, then set up my mounts on their Ibays (to get at the maintenance SME files) and ran the Updater2. Afterwards the reboot now is back to this stopping at LI where it needs the rescue floppy to continue:-(

This is a clean installation straight from the ISO and the Update2... I saw nothing in the readme that said *Oh BTW, when you're done with this Update2, you'll be kissing  your boot goodbye*:~/

I've seen this LI-halt stuff before on my SuSE trial runs - when trying to mess about with SuSE and W2kPro together. If I want SuSE back I end up wiping/clean installing (that's what I -just did- with SME). If I want W2kPro back I just run FDISK /mbr to repair the boot sector.

What's up? Why are Mitel doing this? How may I fix the damage?

best wishes, Robert

[Paul]

This is not a Mitel issue it happens to ALL Linux distros.  It may be due to one or more of the following:

1-Hard drive geometry
2-Positioning of drives on ide's
3-Drive settings in your bios.
4-Bios conflicting with Lilo/Linux
5-A combination of any above.

You have 8 drives so tell us your drives and how they are setup and let's see if we can help you put a handle on it.  Give details including drive specs, raids, ide's, scsi's, extra controller cards, etc.

You can probably just fix it with LILO after booting from floppy but if you want to know why it happened so you can avoid it in the future then you need to do a little investigating.

Paul

P.S. I just noticed after reading your post again that you state you installed the updates from a different drive other than the boot device.  This makes me wonder if this could create problems...Just a thought..

[Paul]

Now I have noticed this on a couple of other posts and they seem to have one thing in common.  You have all installed you updates from a drive other than hda.

I have reported this as a possible bug to e-smith for them to see if I might have something here or not.

Paul

[Robert Harlow]

Sorry just woke up (time zone).

Clean installed 5.6 to a single booting SCSI drive (lowest SCSI ID in the chain).
Adaptec 2940u2w SCSI card specifically boots designated drive.
SCSI BIOS is the last/latest issued for this host adapter.
ASUS P3B-F mobo specifically set to boot to floppy then to SCSI.
Got a feeling you're going to want/need the mobo BIOS... later.
CDROM is SCSI. There is no IDE CDROM.

After the clean install had run its course, the kit was depowered.
Reconnected all the disconnected drives.
Drives disconnected for a clean install due to Mitel's morbid fascination for fscking absolutely everything it can lay its hands on whether or not it needs to being irrelevant.
From off the top of my head I have squeezed some 4 SCSI and 4 IDE drives into a tower;~/ AFAIK none of the IDE drives are bootable and, in any case, the mobo BIOS is set to ignore DIE booting... sorry Freudian slip there... IDE booting.

After rebooting I manually edited FSTAB and ran through bringing all the additional drives online via IBAYS and one manual hookup to a test user.  Everything works and reboots were OK.
The IDE drives are not Promised or anything just regular primary/secondary master/slaves . The mobo is not even ATA100 capable so its not anything highly leading edge.

Then... I ran Update2. From that point all reboots halt at LI. Rescue floppy works OK but is slow and I don't like a potential threat laying in wait... and I call a problem or issue with the boot a real loaded gun threat:-| Hence my earlier willingness (!?) to wipe a working iteration of SME to start again. With that in mind you can imagine my annoyance at what this Update2 has perpetrated and the continuing work to resolve.

                                   * * * * *

I searched the database using the key...
li -client -charlie -public -link -list -installing -satelli -sateli -compatibili -compatabili -killi -scalabili -lib -lim -alia -replic -deliv -prolia -lin -lik -invali -abili -applic -publi -ali -poli -utili -chili -liv -dupli -cli -lic -mulit -lit -trilli -compli -replies -light -mutli -familia -liebert -etablish -establish -lif -appli
...which I listed in another posting (unanswered) referring to the previous instance of this issue being inflicted on my SME. From the contents I looked at using pico and played around moving the word around in various places. No change. After each edit and then invoking I continue to note the idiotic messages being trotted out... for some unknown reason the system seems to reckon that my SCSI booting drive is not the first drive or something similarly weird. Also it goes on about a BIOS drive called 0x80 or something. I'm not certain I have that detail exactly right . I will recheck it when I take the system down to get the BIOS details for the forum.

Request:
would some kind person please display here the contents of a working Update2-run system's so I can check it against mine? Preferably a SCSI booting system.

Paul:
Installing the Update2 from another drive idea is a red herring (I believe).

I'm in and out at the moment. Comms company is decommissioning my old ISDN connection and I've got (www) domains being transferred and now my internet hookup (SME) is subject to unexpected hassle:-| So it's take me as you find me;~/

best wishes, Robert

[Birger Koopmann]

I hav had a similar problem with an redhat installation in multiboot with win98. I had to give som parameters to lilo.conf about the harddrive. In the generel part of lilo.conf i had to tell lilo:
disk=/dev/hdb bios=0x80
disk=/dev/hda bios=0x81
 in the part to start win98 i had to tell lilo to change the these parameters:
map-drive=0x80  to=0x81
map-drive=0x81  to=0x80

I hope this helps
Vh
BK

[Kees Blokland]

Request:
would some kind person please display here the contents of a working Update2-run system's so I can check it against mine? Preferably a SCSI booting system.


On an Asus-mb with external Initio-scsicontroller, 2-scsi drives (this was a clean install, but I've upgraded a similar 2-scsi raid from 5.5 without any apparent problems. I did however download the updates to the local system before upgrading. (reading other comments about this issue))

lilo-conf says:

# copyright (C) 2002 Mitel Networks Corporation
#------------------------------------------------------------
prompt
timeout=50
default=Mitel-SME
boot=/dev/md0
map=/boot/map
install=/boot/boot.b
message=/boot/mitel.pcx
linear
image=/boot/vmlinuz-2.4.18-5
   label=Mitel-SME
   initrd=/boot/initrd-2.4.18-5.img
   read-only
   root=/dev/md1

[Robert Harlow]

Kind of you Kees:-)

Currently my problem iteration reads...

# copyright (C) 2002 Mitel Networks Corporation
#------------------------------------------------------------
prompt
timeout=50
default=Mitel-SME
boot=/dev/sda
map=/boot/map
install=/boot/boot.b
message=/boot/mitel.pcx
image=/boot/vmlinuz-2.4.18-5
   label=Mitel-SME
   initrd=/boot/initrd-2.4.18-5.img
   read-only
   root=/dev/sda3
linear

...the linear line was moved around a bit using pico. Earlier I tried the linear line underneath the prompt line but the resulting boot attempt fared no better. When I have done the stuff below I will pico the linear line underneath that message line and see if that hits the G-spot.

As you can surmise from the above, my booting drive is DEFINITELY the first SCSI drive, that is why it's called sda. You'll have to take my word that the boot order in the mobo's BIOS is set to floppy then SCSI then nothing and again that none of the IDE drives were ever made active (for booting purposes) as they are just mounted archive drives. Quite where SME has determined that sda is NOT the first available booting drive is somewhat mystifying.

...and the unchanged system WORKED BEFORE UPDATE2 was run.

I will go away and take the system down to obtain some more motherboard/BIOS information. I will have a go with a very old copy of Partition Magic (v6) to see if I can see anything awry or interesting to feedback. Taking SME down loses me the use of the wireless broadband in it's current configuration.

best wishes, Robert

[Robert Harlow]

How the memory fades;~/ Partition Magic (v6) reports three drives (1, 3, 6) as being with active attributes. That is after it had finished its initial palpitations, presumably with the EXT3 partitions, as it never handled even EXT2 properly.
• Drives 1 through to 4 are SCSI.
• Drive 1 is most emphatically my designated boot SCSI drive.
• Drive 3 does have an active partition, as it did before running Update2.
• Drives 5 through to 8 are IDE.
• Drive 6 has an active FAT32 partition that boots into W98 FE but, to do this, I need to manually reconfigure the mobo's boot selection order. It just doesn't occur automagically no matter what SME's LILO might otherwise infer.

                          * * * * *

Using pico I have tried putting the linear line under the message line.
Then I ran .
Screen then displays all sorts of rabid nonsense...
-------------------------------------------------------------->>
Warning: /dev/sda is not on the first disk
Merging with /boot/boot.b
Warning: BIOS drive 0x82 may not be accessible         # "This Does Not Compute"
Warning: BIOS drive 0x82 may not be accessible         # Yes, I heard you the 1st time!
Mapping message file /boot/mitel.pcx
Warning: BIOS drive 0x82 may not be accessible         # Now I'm getting nonsense echoes!
Boot image: /boot/vmlinuz-2.4.18-5
Warning: BIOS drive 0x82 may not be accessible         # OK, so the record is stuck:-|
Mapping RAM disk /boot/initrd-2.4.18-5.img
Warning: BIOS drive 0x82 may not be accessible         # Somebody turn this nonsense off please:-(
Added Mitel-SME *
/boot/boot.0800 exists - no backup copy made.
Writing boot sector.
<<--------------------------------------------------------------
(manually copy-typed from across the room onto another machine)

With absolutely no confidence that this work going to work I then...
/sbin/e-smith/signal-event reboot
...and... no, it wasn't successful. Back to the rescue floppy.

With reference to my post-Update2 SME5.6 server-gateway machine, it would seem like the inmates are now running the asylum;~/ Anyone any idea on how to restore normality?

best wishes, Robert

[Robert Harlow]

I am terribly sorry, Birger, I entirely overlooked your posting!

On the face of it your input is for an IDE (only?) booting system and one to W98 at that. Not sure how that translates into a SCSi booting system that just so happens has multiple attached drives that are IDE.

However it gave me an idea. I turned *off* the IDE drives in the mobo's BIOS. In the past this has -NOT- protected my IDE drives from the universal fsck inflicted on a SME clean install and I know this all too well. The resulting boot attempt again failed at LI. I have returned the IDE drives' BIOS status to auto.

best wishes, Robert

[Robert Harlow]

BIOS=Award v6      10/15/2001-i440BX-
Motherboard =ASUS P3B-F   ACPI   BIOS=Rev 1008 Beta 004

Found at the ASUS site...
ftp://ftp.asus.com.tw/pub/ASUS/mb/slot1/440bx/p3b-f/index.txt
---------------------------------------------------------->>
bx3f1004.zip  
P3B-F (Intel 440BX chipset) M/B, PnP BIOS ver. 1004 25/Oct/99
           [snip]
13. Fix the system cannot boot from any bootable device if booting from a 3Com 3C905C-TXM LAN card is failed or bypassed.
           [snip]
<<----------------------------------------------------------

My BIOS is Rev1008 though it has a beta index their beta's never gave me any problems. AFAIK there are/were no further versions issued. That was in Rev 1004 and mine is Rev 1008. The fact that there was an issue identified in the past *might* be useful.

BTW I have a pair of 3C905C-TXM NICs and the LI failure occurs -right after- the NICs are enumerated.

best wishes, Robert

[Birger Koopmann]

I still think it might not be the NIC-s, but Lilo getting confused about your harddrives.
In have two IDE drives and in my BIOS i can set the drive i want to boot from. In Linux hda is the first IDEdisk and hdb the second disk.
........When i boot from the first disk (hda, where i have Win98) it boots right into win98
........When i set the BIOS to boot from the second disk (hdb, that is where i have lilo-bootmenu and linux) I hav to tell lilo that lilo has to switch the disks. This is in the generel (starting) section in lilo.conf:
disk=/dev/hdb bios=0x80
disk=/dev/hda bios=0x81
But when i want to start win98 on the first drive (hda, when booting with lilo on hdb) then this section of lilo must change the boot-order with the remapping of drives:
map-drive=0x80 to=0x81
map-drive=0x81 to=0x80

As i se it you have to figure out your drives.
You can use from the commandline:
/sbin/lilo -v -v -v >/boot/lilo.log 2>/boot/lilo.logerr
and look in lilo.log how lilo sees yor BIOS-harddisksetup.
They seem not to be like your expectations.
Your lilo says:"Warning: /dev/sda is not on the first disk", And that is what you told you thaugt it should be.
Your lilo says:"Warning: BIOS drive 0x82 may not be accessible ". This looks like the problem with lilo not getting the Bios-parameters right.
I wold try to tell lilo:
disk=/dev/sda bios=0x80
disk=/dev/sdb bios=0x81

and so on, but dont its on your own risk.Robert Harlow wrote:
>
> How the memory fades;~/ Partition Magic (v6) reports three
> drives (1, 3, 6) as being with active attributes. That is
> after it had finished its initial palpitations, presumably
> with the EXT3 partitions, as it never handled even EXT2
> properly.
> • Drives 1 through to 4 are SCSI.
> • Drive 1 is most emphatically my designated boot SCSI drive.
> • Drive 3 does have an active partition, as it did before
> running Update2.
> • Drives 5 through to 8 are IDE.
> • Drive 6 has an active FAT32 partition that boots into W98
> FE but, to do this, I need to manually reconfigure the mobo's
> boot selection order. It just doesn't occur automagically no
> matter what SME's LILO might otherwise infer.
>
>                           * * * * *
>
> Using pico I have tried putting the linear line under the
> message line.
> Then I ran .
> Screen then displays all sorts of rabid nonsense...
> -------------------------------------------------------------->>
> Warning: /dev/sda is not on the first disk
> Merging with /boot/boot.b
> Warning: BIOS drive 0x82 may not be accessible         # "This Does
> Not Compute"
> Warning: BIOS drive 0x82 may not be accessible         # Yes, I
> heard you the 1st time!
> Mapping message file /boot/mitel.pcx
> Warning: BIOS drive 0x82 may not be accessible         # Now I'm
> getting nonsense echoes!
> Boot image: /boot/vmlinuz-2.4.18-5
> Warning: BIOS drive 0x82 may not be accessible         # OK, so the
> record is stuck:-|
> Mapping RAM disk /boot/initrd-2.4.18-5.img
> Warning: BIOS drive 0x82 may not be accessible         # Somebody
> turn this nonsense off please:-(
> Added Mitel-SME *
> /boot/boot.0800 exists - no backup copy made.
> Writing boot sector.
> <<--------------------------------------------------------------
> (manually copy-typed from across the room onto another machine)
>
> With absolutely no confidence that this work going to work I
> then...
> /sbin/e-smith/signal-event reboot
> ...and... no, it wasn't successful. Back to the rescue floppy.
>
> With reference to my post-Update2 SME5.6 server-gateway
> machine, it would seem like the inmates are now running the
> asylum;~/ Anyone any idea on how to restore normality?
>
> best wishes, Robert

[Paul]

>Warning: /dev/sda is not on the first disk
>Merging with /boot/boot.b
>Warning: BIOS drive 0x82 may not be accessible # "This Does Not Compute"
>Warning: BIOS drive 0x82 may not be accessible # Yes, I heard you the 1st time!
>Mapping message file /boot/mitel.pcx
>Warning: BIOS drive 0x82 may not be accessible # Now I'm getting nonsense echoes!
>Boot image: /boot/vmlinuz-2.4.18-5
>Warning: BIOS drive 0x82 may not be accessible # OK, so the record is stuck:-|
>Mapping RAM disk /boot/initrd-2.4.18-5.img
>Warning: BIOS drive 0x82 may not be accessible # Somebody turn this nonsense off please:-(
>Added Mitel-SME *
>/boot/boot.0800 exists - no backup copy made.
>Writing boot sector.

Actually these error messages do make some sense.  It appears that lilo is having a hard time figuring out where the boot device is.  You need some lines telling lilo what each drive is (if this is the order you want).  Lilo assumes that the boot device is 0x80 unless you tell it otherwise.

disk=/dev/sda bios=0x80
disk=/dev/sdb bios=0x81
disk=/dev/sdc bios=0x82
disk=/dev/sdd bios=0x83
disk=/dev/hda bios=0x84
disk=/dev/hdb bios=0x85
disk=/dev/hdc bios=0x86
disk=/dev/hda bios=0x87

This is just an example but I think it will work for the drive arrangement you listed above.

Let us know how it is going,
Paul

[Robert Harlow]

Birger
I think I get what you're saying and I am heartily glad you made the effort to let me know:-)

Paul
Ditto that:-) Your list order looks pretty much akin to what I have installed. There is also a 640Mb Fujitsu Magneto-Optical SCSI disc/drive *hooked up*. Linux never (SME too) seems to know what to make of this drive, there have always been error lines in the boot logs - for over a year. Perhaps something in Update2 has made SME more sensitive/erratic in how it interprets or accounts for this drive. I would've unhooked it long ago but it has an active terminator and all the others on the SCSI chain only have passive terminators (or at least don't specify that they have active terminators fitted). Been waiting for some spare time (huh) to play around with it.

                          * * * * *

The -ONLY- drive that matters at boot time is the single SCSI boot drive. I can figure out where the others are and what they are called after the boot and that is within my grasp. It's what happens at boot time that gives me the creeps:-| If it doesn't boot then that's about it, unless the rescue floppy can inject sense into the mess.

This is beginning to add up, thanks to Birger and Paul;~/

AFAIK I didn't deserve this nor did any of my kit. AFAIK it has been Update2 that has inflicted the need to both learn and carry out emergency boot surgery on my live drives and/or the OS.

I am EXTREMELY glad that I know which drive needs to boot,  where it is and what it is supposed to be called (sda). I find it unbelievably obtuse that the LILO mechanism should have been so corrupted by the running of Update2 that the former has rendered the boot process unworkable, on both occasions that I had cause to run Update2.

I've been editing recalcitrant websites all day, have messed up the production of a monthly LISTSERV mailer, it's now gone 3:30am locally, my head is thumping and I reckon I'm pretty tired. Please excuse me for not trying this out until tomorrow...

best wishes, Robert

[Robert Harlow]

***THAT*** was indeed what was required:-)

I used pico to display/edit  lilo.conf to the following...
--------------------------------------------------------------------->>
prompt
timeout=50
linear
disk=/dev/sda bios=0x80
disk=/dev/sdb bios=0x81
disk=/dev/sdc bios=0x82
disk=/dev/sdd bios=0x83
disk=/dev/hda bios=0x84
disk=/dev/hdb bios=0x85
disk=/dev/hdc bios=0x86
disk=/dev/hdd bios=0x87
default=Mitel-SME
boot=/dev/sda
map=/boot/map
install=/boot/boot.b
message=/boot/mitel.pcx
image=/boot/vmlinuz-2.4.18-5
   label=Mitel-SME
   initrd=/boot/initrd-2.4.18-5.img
   read-only
   root=/dev/sda3
<<---------------------------------------------------------------------
...the resultant command of lilo -v made logical enough sense.
So I rebooted and SME positively sailed through the LILO point.
All my visible drives are where they should be on the SME.

                                 * * * * *

Thank you once again Birger and Paul, you've been most helpful.
Your workaround is most acceptable and this thread is now resolved.

best wishes, Robert

[Paul]

Very glad to help.  I will pass on the results to a couple of other people with simular problems.

Paul