machine do not boot after making mistake
Harald ARNOLD
ASC-console IBM3151. Installation Cds and a boot-tape.
Today I tried to reinstall sambe under /usr/local/samba/bin.
To remove the old version I made "make uninstall" and looked
at the directory: it was full. Therefore I removed all files
"rm *" in this dir.
But this was only a link to /usr/bin ! Therefore ALL file of
/usr/bin are removed !!!
I tried to reboot, but it is not working.
To recover those files, I replaced the system harddisk by
another and installed a base AIX-System on the second HDD.
Then I saw that I have a backup of /usr/bin of thos machine
and rebooted with installation CD, mounting the rootvg. Then
I copied the /usr/bin of the backup (I burned a CD) into the
rootvg.
But now the machine will not boot !!! On the three digit
display the machine switches between 223 and 229, but no
boot. In maintenance mode, I can boot in single user mode.
PLEASE HELP !!! Thanks Harald
Harald ARNOLD
> I have a Risc6000 390 with Floppy, 8GB-5mm-Tape, CD,
> ASC-console IBM3151. Installation Cds and a boot-tape.
[...]
> PLEASE HELP !!! Thanks Harald
I am using AIX 4.3.3.
Laurenz Albe
> I removed all files
> "rm *" in this dir.
>
> But this was only a link to /usr/bin ! Therefore ALL file of
> /usr/bin are removed !!!
>
> I tried to reboot, but it is not working.
>
> To recover those files, I replaced the system harddisk by
> another and installed a base AIX-System on the second HDD.
> Then I saw that I have a backup of /usr/bin of thos machine
> and rebooted with installation CD, mounting the rootvg. Then
> I copied the /usr/bin of the backup (I burned a CD) into the
> rootvg.
>
> But now the machine will not boot !!! On the three digit
> display the machine switches between 223 and 229, but no
> boot. In maintenance mode, I can boot in single user mode.
You replaced the system hard disk by another?
So, what is in you machine NOW: is it the original hard disk where you
removed /usr/bin/*? Or is it the other disk on which you did a new install?
Let me guess, tell me if I am wrong:
1.) You ripped out the disk containing the executable challenged AIX.
2.) You put in a new hard disk and installed AIX from scratch.
3.) You remembered that you have a backup.
4.) You took out the new hard disk and put the other one back.
5.) You booted from CDROM and restored /usr/bin/* from backup.
6.) You try to boot from hard disk and it does not work.
If I am correct, your problem is probably the boot list.
When you did the fresh AIX install, your hard disk was recognized as
a different hard disk and got a different name, and during the boot
process the boot list was adjusted accordingly.
Now that you changed disks again without changing the boot list, the
machine cannot boot.
You should change the boot list so that it considers your original hard
disk.
Either boot into maintenance mode and use the 'bootlist' command to
set the boot list, or change it in SMS. If in doubt, read the manual for
your hardware.
Yours,
Laurenz Albe
Mark Taylor
or do a preservation install (note, this will overwrite /usr .. but you
have removed that anyways so it doesnt matter.. was gonna say stuff
like TSM config is under /usr and it will get blown away and you will
have to re-install the app etc..
HTH
Mark Taylor
Harald ARNOLD
> Harald ARNOLD <Harald...@arnold.at> wrote:
>> I removed all files
>> "rm *" in this dir.
>>
>> But this was only a link to /usr/bin ! Therefore ALL file of
>> /usr/bin are removed !!!
>>
>> I tried to reboot, but it is not working.
>>
>> To recover those files, I replaced the system harddisk by
>> another and installed a base AIX-System on the second HDD.
>> Then I saw that I have a backup of /usr/bin of thos machine
>> and rebooted with installation CD, mounting the rootvg. Then
>> I copied the /usr/bin of the backup (I burned a CD) into the
>> rootvg.
>>
>> But now the machine will not boot !!! On the three digit
>> display the machine switches between 223 and 229, but no
>> boot. In maintenance mode, I can boot in single user mode.
>
> You replaced the system hard disk by another?
>
> So, what is in you machine NOW: is it the original hard disk where you
> removed /usr/bin/*? Or is it the other disk on which you did a new
> install?
It is the original HDD where I removed /usr/bin ! The machine is
running a oracle test application and therefore I will not
reinstall the machine. !
Thanks Harald
Markus Baertschi
>
> It is the original HDD where I removed /usr/bin ! The machine is
> running a oracle test application and therefore I will not
> reinstall the machine. !
>
> Thanks Harald
Harald,
you have essentially two choices:
1) Restore from your mksysb backup
As it looks like don't have a backup number 2 remains:
2) Reinstall AIX using the preservation method
The preservation method will only overwrite the /, /usr, /var and
/tmp filesystems. /home and all your private filesystems will remain.
You could try to copy the contents of /usr/bin from another AIX box. But
the box must have the exact same maintenence levels and fixes
installed. Booting from CD you can access the disks to restore the stuff
from the other box. You will have to reinstall the machine eventually
and the package manager and the installed binaries are now out of sync.
Markus
Lothar Krauss
> To recover those files, I replaced the system harddisk by
> another and installed a base AIX-System on the second HDD.
> Then I saw that I have a backup of /usr/bin of thos machine
> and rebooted with installation CD, mounting the rootvg. Then
> I copied the /usr/bin of the backup (I burned a CD) into the
> rootvg.
>
> But now the machine will not boot !!! On the three digit
> display the machine switches between 223 and 229, but no
> boot. In maintenance mode, I can boot in single user mode.
How did you copy these files? A simple "cp" will not preserve links and
permissions! A better way will be using the "tar" command with user root.
Lothar
Harald ARNOLD
No I did not remove /usr: ONLY /usr/bin and I have a tar-File as copy
one month ago. And in the mean time I did not install any application
(except a new version of samba, openssh from source files).
And my boot tape is generated from the second installed HDD, which
I also removed and inserted the old original HDDs (two 2 GB HHDs -
I think so) as rootvg.
I think that the RISC cannot recognize the boot device, because I
installed a second HDD and therefore the original HDDs are now not
recognized. I think that I have to tell the RISC (like under PCs)
in i BIOS, which HDD should boot. But I do not know how to tell the
machine via ASC-IBM3151-Terminal to change the boot specification.
Thanks Harald
Harald ARNOLD
> the box must have the exact same maintenence levels and fixes
> installed. Booting from CD you can access the disks to restore the stuff
> from the other box.
My first mistake was the removing of /usr/bin. OK!
I have a backup of exactly this machine from 17.January this year on a
linux box in tar-gzip format of ALL files. The machine is for testing
applications and some times (murphy !) for developing :-(. I have alle
files and everything. Therefore on my system no system application
have been added.
After my first mistake I made a halt.
Maybe afterwards I made a second mistake to replace the original two
disks (each about 1 GB) against another and installing another AIX
system with original CD like original HDD. Then I made a boot-tape
from the second system (maybe without updates of AIX-base-system).
Next time, I recognized that I have the complete tar-gzip backup of
the original HDDs. Then I booted via AIX-CDs maintenance mode,
accessing rootvg and copied all files from the tar-gzip-backup
(via burned CD and mounting this CD on AIX) to /usr/bin. Afterwards
I tried to reboot, but switching 3-digit-display between 223 and 229.
In manual 223 and 229 are discribed (in german, I will dare to translate):
223: IPL will be tried in normal mode, which are connected to SCSI.
This devices are entered in NVRAM-IPL-device list.
229: IPL will be tried in normal mode, which are connected to SCSI.
IPL could not be executed by a device in list, or device list has
not correct entries.
I do not know AIX so good that I known something of a BIOS like on
PCs or how to get into it via ASC-console (ibm3151). Sometimes I
read about press (F1 / F5 on F50 or 1 / 2 on a ASC console). But
my system always "will" boot on a HHD.
I did not change _ANY_ boot images, nothing else ! I removed "only"
/usr/bin.
I think that it is a setting on my RISC 390. But I do not know how
to enter this mode and do not know what settings are incorrect.
Thanks Harald
Harald ARNOLD
> How did you copy these files? A simple "cp" will not preserve links and
> permissions! A better way will be using the "tar" command with user root.
That was my first idea, but I did not have the /usr/bin directory.
And on the next step I installed another version of into /usr/bin
with dependencies in /usr/local which is a link to /home (datavg)
due to HDD-space :-(((((((((
I have an article for linux for recovering files in this way - and I
did this for my firewall to change to a bigger HHD. And it was
working so good. It was up in 1 hours after exchanging both HDDs :-)
Thanks Harald
Harald ARNOLD
> Let me guess, tell me if I am wrong:
>
> 1.) You ripped out the disk containing the executable challenged AIX.
> 2.) You put in a new hard disk and installed AIX from scratch.
> 3.) You remembered that you have a backup.
> 4.) You took out the new hard disk and put the other one back.
> 5.) You booted from CDROM and restored /usr/bin/* from backup.
> 6.) You try to boot from hard disk and it does not work.
>
> If I am correct, your problem is probably the boot list.
> When you did the fresh AIX install, your hard disk was recognized as
> a different hard disk and got a different name, and during the boot
> process the boot list was adjusted accordingly.
But the original HD I newer user at boot time. Beside the step to
recopy the /usr/bin. I rebooted in maintenance mode and copied
the files from burned-backup-CD into /usr/bin (afterwards also
x11r5 and x11r6), because I did not have so much commands (no "ls",
...)!
I recovered many SuSE-Linux-HDDs. But the changes can only be on
the machine itself. I did never had both HDDS (original and scratch
installed HDDs at the same time - only maintenance mode bootup).
I would not try to make a third fatal mistake :-))))
> Now that you changed disks again without changing the boot list, the
> machine cannot boot.
Sorry for an incorrect reply :-((
YES you are totally correct. (I am also father since 1.5 months,
everthing is hard in the very moment).
On a linus "Intel" system the problem can be:
- the BIOS is incorrect (therefore I think that I have to change
some settings on the "RISC-BIOS")
- reinstall the boot sequence (image, first loader, second loader)
but I have no idea, how I can get into this mode via ASC-ibm3151.
console.
Thanks Harald
PS: My machine is sleeping and will not make any further mistakes,
until I know how to restore it ...
Markus Baertschi
Hello Harald,
As I understand you have a complete backup in a tar archive and you have
the two AIX harddisks of the original AIX installation (with missing
/usr/bin) sitting on a shelf.
You also installed AIX from scratch onto a new disk. This new
installation worked until you restored /usr/bin from the tar file. The
new installation now refused to boot (223/229).
I don't understnd why you restored /usr/bin from the tar file onto you
perfectly working new installation. If you need individual files just
restore them one by one.
You need to do some of the following steps:
- Install AIX again (full overwrite) onto the new disk
-> You get a working AIX system
- Restore the lost data from the tar file
or
- Install the two original AIX disks in the machine and import the
volumegroup (importvg hdisk1). You then can mount the volumes form the
original aix install and recover your lost files.
Important: Do not touch anything in /usr unless you know what you are
doing. Your optional software should go into /usr/local or /opt.
Markus
P.S. From you -at domain I suspect you speak German. I happen to speak
it too, but better offline, let's not offend our international friends here.
Markus Baertschi
>
> I think that the RISC cannot recognize the boot device, because I
> installed a second HDD and therefore the original HDDs are now not
> recognized. I think that I have to tell the RISC (like under PCs)
> in i BIOS, which HDD should boot. But I do not know how to tell the
> machine via ASC-IBM3151-Terminal to change the boot specification.
>
> Thanks Harald
>
Afte power-on, just after the screen with the IBM logos, press 1 to get
into the bios menus. There you can configure the boot devices.
Markus
Markus Baertschi
> That was my first idea, but I did not have the /usr/bin directory.
> And on the next step I installed another version of into /usr/bin
> with dependencies in /usr/local which is a link to /home (datavg)
> due to HDD-space :-(((((((((
>
> I have an article for linux for recovering files in this way - and I
> did this for my firewall to change to a bigger HHD. And it was
> working so good. It was up in 1 hours after exchanging both HDDs :-)
>
> Thanks Harald
Just some more remarks, are you aware that you can increase the size of
filesystems in AIX on the fly ? You can just increase /usr if needed !
If you need mor disk space you can extend your rootvg to a second disk,
if the disk can be hot-plugged this is an on the fly operation as well.
To move the installation to another disk for other reasons the only good
way is to use mksysb and reinstall the mksysb on the other disk.
Markus
Harald ARNOLD
> Just some more remarks, are you aware that you can increase the size of
> filesystems in AIX on the fly ? You can just increase /usr if needed !
Yes, but it is only a test system for a customer. with to (Ithink) 1 GB
HDDs inside. I know smitty very good and it would has been possible to
insert another external small HDD to rootvg. But I did not :.((
> If you need more disk space you can extend your rootvg to a second disk,
> if the disk can be hot-plugged this is an on the fly operation as well.
I can start and stop the machine ten times a day ! It is for testing.
> To move the installation to another disk for other reasons the only good
> way is to use mksysb and reinstall the mksysb on the other disk.
These are the ORIGINAL HDDs and not the second HDD with the new setup
from scratch.
Thanks Harald
Harald ARNOLD
> Afte power-on, just after the screen with the IBM logos, press 1 to get
> into the bios menus. There you can configure the boot devices.
This feature I know from my Risc 43P, but this is not working with
the RISC 390 IMHO. The ASC-console is coming up at the very moment
where I see that the subsystems of AIX are starting. Therefore I
think that the machine is already booting from hard disk :-((
Harald
Harald ARNOLD
> Harald ARNOLD wrote:
> As I understand you have a complete backup in a tar archive and you have
> the two AIX harddisks of the original AIX installation (with missing
> /usr/bin) sitting on a shelf.
Yes and No. For installing the second HDD the original HDDs have be on
a shelf. After I recognized that a have a tar archive, I did nothing
more with the second install AIX-HDDs and I replaces it by the original
disks. Now the second new installed disk is on the shelf. The machine
is now the same hardware with all original disks and recovered /usr/bin
but only by command cp (not tar) !
> You also installed AIX from scratch onto a new disk. This new
> installation worked until you restored /usr/bin from the tar file.
No! The second installation is on the shelf and is only a small
installation of AIX for getting files of /usr/bin. But after finding
original backup of /usr/bin I recovered those via extracting files
on a linux box, burning a CD and mounting in maintanace mode with
original HDDs. I never used both installtions (original and new scratch
installation) at the same time. Therefore the original HDDs are not
changed (IMHO).
> The new installation now refused to boot (223/229).
No! The boot process is no correct with the original HDDs. It is
the same situation like before my fatal mistake.
> I don't understnd why you restored /usr/bin from the tar file onto you
> perfectly working new installation. If you need individual files just
> restore them one by one.
This is the original HDD. There is an oracle database installed and
I realy do not want to reinstall the system.
> P.S. From you -at domain I suspect you speak German. I happen to speak
> it too, but better offline, let's not offend our international friends
> here.
Ja ! Yes !
Harald
Harald ARNOLD
> You should change the boot list so that it considers your original hard
> disk.
> Either boot into maintenance mode and use the 'bootlist' command to
> set the boot list, or change it in SMS. If in doubt, read the manual for
> your hardware.
OK via bootlist the system will now reboot, but with some errors.
That I will change afterwards.
BUT !!! My datavg is now unrecognized.
In smitty I tried to activate the VG. Smitty "VGs - Activate VG".
But the machine is telling me:
... varyonvg: access not possible to spezific "device" (german
"Einheitendatei"). Run redefinevg and synclvodm.
I do not want to make another mistake and I am not so good at
top level maintenace. The VG name is "datavg" ( :-)) ). I think
the device is "hdisk2". What parameters I need to can access my
old datavg ?
Specially thanks to ALL persons, who helped me ! Congratulations !
Thanks Harald
Harald ARNOLD
> Laurenz Albe wrote:
>
> ... varyonvg: access not possible to spezific "device" (german
> "Einheitendatei"). Run redefinevg and synclvodm.
>
> I do not want to make another mistake and I am not so good at
> top level maintenace. The VG name is "datavg" ( :-)) ). I think
> the device is "hdisk2". What parameters I need to can access my
> old datavg ?
In smitty under "system storage managemant" / "logical volumn
manager" / "physikal volums":
With "list all physical volums at system" I can see the hdisk2
and hdisk3 defined for VG "datavg"
but
All other selections at this page show only VG datavg disk
(hdisk0 and hdisk1) :-((((((
What could I do ?
Harald
Mike
Try
# importvg datavg HDISK
where HDISK is one of the physical disks in your datavg
Harald ARNOLD
UP and RUNNING !
Thanks to all helpers !
And I will do IMMEDIATLY recovery backups, tapes, procedures ...
Only THANK YOU !!!
Harald ARNOLD
>> To move the installation to another disk for other reasons the only good
>> way is to use mksysb and reinstall the mksysb on the other disk.
>
> These are the ORIGINAL HDDs and not the second HDD with the new setup
> from scratch.
As extension to all, who will have the problem:
If you have removed system componts and you have a backup from this machine:
1) burn a CD with those directories you removed (/usr/sin or usr/bin)
2) Hope that the most important command works in single user mode: mount
3) mount the CD and append the PATH to the mounted directory of your CD
4) If your tar command works, then copy your files (original links, user-
ownership, etc) by: CD to your CD-directory and execute command:
tar -Csp --numeric-owner --atime-preserve -f - . | ( cd (YOUR-REMOVED-DIR \
&& tar -xSpv --atime-preserve -f - )
But make sure that you only include those files that are missing. NEVER
remove a file that is working correct, because otherwise you will remove
another system command, that you will need for recovery-process.
I did a copy of those commands, because may tar command has dependencies
into my datavg at /home. Therefore that command did not work.
5) Via bootlist I configurated the boot process
6) Restoring all ORIGINAL links in /usr/bin. I copied all programmes and
therefore I removed all links. Therefore I restored the original links.
(programmes: diff, ls, vi, sort, etc ... to restore this)
7) Some programms in /usr/bin, /usr/sbin and also /usr/ccs/bin have been
remove by system startup due to missing links. Therefore I also recovered
those files like in step 4. But again ONLY those programms that are
missing.
Thanks Harald
Mark Taylor
is restore from mksysb or do a preservation install.
if you had a mksysb policy, then you could have been up and running
within an hour with no issues and if you then still had problems you
would still be covered by your support contract...
restoring system files as you have detailled above is totally
unsupported and IBM will not support your server if they find out you
have used this method to restore you files.
you may want to think about that before you start telling people the
procedure as its unsupported.
rgds
Mark Taylor
Martin Etteldorf
> tar -Csp --numeric-owner --atime-preserve -f - . | ( cd (YOUR-REMOVED-DIR \
> && tar -xSpv --atime-preserve -f - )
That's no tar syntax, seems more likely to be gnutar.
BTW... Use find and cpio. Way faster, less overhead, same functionality:
find . -print | cpio -pdmuv YOUR-REMOVED-DIR
Martin.
--
...Rabotai pomalkivai
Laurenz Albe
> OK via bootlist the system will now reboot, but with some errors.
> That I will change afterwards.
Seems like you didn't really read my reply carefully the first time.
Good you eventually found the time.
Und herzlichen Glückwunsch zum Nachwuchs.
Yours,
Laurenz Albe
Harald ARNOLD
> BTW... Use find and cpio. Way faster, less overhead, same functionality:
> find . -print | cpio -pdmuv YOUR-REMOVED-DIR
Yes it is gnu tar. That is the reason, why I have dependencies into datavg.
My original tar program is "tar.old".
Thanks Harald
Harald ARNOLD
> Seems like you didn't really read my reply carefully the first time.
> Good you eventually found the time.
>
> Und herzlichen Glückwunsch zum Nachwuchs.
>
I read it correctly. But I waited two day to have more input
and more ideas. Therefore I did recovery yesterday.
Thanks for my little child: Her name is PIA ! Danke dir Laurenz !
Thanks to all, Harald
Harald ARNOLD
> you may want to think about that before you start telling people the
> procedure as its unsupported.
You are totaly correct. But I recovered so many linux machines that
I am aware about what I am doing. Sometimes I do not known the
commands under AIX. If you are realy only removing "bin"-files
then I do not think that there can be any problem. Never touch:
1) boot strapping
2) images (kernel)
3) Internal databases (application DB, etc)
4) Everytime make a complete diff of system files from old backup
and the files after recovery:
- owner ship and group
- file mode (chmod)
- file type (link, normal file, pipe, device, ...)
- preserve creation time
- check file size on normal files
And only after those checks you are on the correct way !
And never be so silly like me, that you do not have a boot tape
from your machine. That will take only one hour for being happy ---
Harald
Jose Pina Coelho
news:42hie2-...@sauron.arnold.at:
>
>
> After my first mistake I made a halt.
A small tip: when you do a mistake of that type, do nothing (leave the
machine up) until you're quite sure about what you should do next.
> [...]
> Next time, I recognized that I have the complete tar-gzip backup of
> the original HDDs. Then I booted via AIX-CDs maintenance mode,
> accessing rootvg and copied all files from the tar-gzip-backup
> (via burned CD and mounting this CD on AIX) to /usr/bin. Afterwards
> I tried to reboot, but switching 3-digit-display between 223 and 229.
>
> In manual 223 and 229 are discribed (in german, I will dare to
> translate):
223/229 means the system can't find a boot device.
Boot from the diag CD and change the bootlist to: cd0 rmt0 hdisk0 hdisk1
Check the ID jumpers on all the scsi devices, you may have both disks
sharing the same ID.
If this is not enough, boot from the AIX CD, then import the rootvg and
do a bosboot -ad /dev/hdisk0
--
Doing AIX support was the most monty-pythonesque
activity available at the time.
Eagerly awaiting my thin chocolat mint.
dalegibbs
I have a p5-550 sand box with Virtual I/O Server, AIX 5.3, and SuSE
SLES
9 running in shared processor LPARs. Works just fine. An LPAR is no
different than a single system with a single OS image. Each LPAR runs
isolated from the others (unless of course you are dependent on a
single
VIOS LPAR, but you can use two of those). Crack open a Redbook or two
and give it a shot.
Greg.
Please can someone tell me what the easiest method is in starting up a
IBM p5 series server model: 9113 550
Without having to use the HMC ???
Many thanks
Dale
da...@sanbi.ac.za
Greg
Hint: pay someone else to do it.
If you are not building LPARs you can use an ASCII console or laptop or
invest in a graphics adapter, keyboard and mouse.
If you are going to have a few p5 boxes around, the HMC is worth it.
Greg.