Ignite bootable tape isn't

[Stephen Keane]

OK, so I installed Ignite (version C.6.0.109).
I ran make_tape_recovery -A -v. No errors were reported. I rebooted the server, interrupt the boot sequence and booted from the tape (SONY SDX-300C AIT drive). ISL loads OK, HPUX is happy, the AUTO file points to INSTALL which causes WINSTALL to be loaded, the kernel load data (offset/start addr etc) is displayed and then ... nada ... left it for 30 minutes ... nada. So what I have is a bootable recovery tape that doesn't actually boot! Good job it wasn't an emergency then.

O/S 11.11, can supply patch list of required.

I have used the tape drive for backup/restore so I know the server is happy with the drive. I don't know if the boot basic tape driver is happy with it, but is does load ISL, AUTO, HPUX, INSTALL and WINSTALL from it which is a good sign.

The tape looks good, a LIF volume (which you can lifls) followed by a tar archive (whioch you can (tar vft)

[Michael Roberts_3]

can you provide the console output from the
BCH prompt though the kernel load?

Is the tape drive active after the kernel load info is displayed?

You can get some early debug info by selecting 'Y' at the 'interact with ISL' question, and then entering:

ISL> hpux -D0x2a :WINSTALL

[Steven E. Protter]

I would push a shell and see that the isntallation looks like.

I'm hoping the tape device isn't the issue. You may have spotted a bug.

SEP

[Stephen Keane]

I've switched to a DDS drive, which shows up on IO scan as a HP drive (even though it isn't).

With the original drive, the tape drive is not active after the kernel load parameters (size, start addr etc) are displayed. The machine doesn't appear to be doing anything, the front panel display doesn't change.

I'll post what the effect of the different drive is.

[Stephen Keane]

OK, tried the same thing with the DDS drive. Same result (other than Ignite decided it needed multiple volumes to create the tape image).

From the SEArch command ...

Main Menu: Enter command > sea

Searching for potential boot device(s)...
this may take several minutes.

To discontinue search, press any key (termination may not be immediate).

Path Number Device Path Device Type
----------- --------------- -----------
P0 LAN.10.0.0.1.3.254 10.0.0.1 10/Half Dx
P1 IDE CRD-8400B
P2 SCSI.0.0 HP C1533A
P3 FWSCSI.6.0 COMPAQ BD009122A
P4 FWSCSI.5.0 COMPAQ BD009122A

Main Menu: Enter command > bo p2
Interact with IPL (Y, N, Q)?> n

Booting...
Boot IO Dependent Code (IODC) revision 0

HARD Booted.

ISL Revision A.00.44 Mar 12, 2003

ISL booting hpux (;0):INSTALL

Boot
: tape(10/0/15/0.0.0.0.0.0.0;0):WINSTALL
11763712 + 1970264 + 2628048 start 0x200ae8

And that's it! Nothing else happens.
Front panel display says

INI CEF2:
785/B1000

I'll try it again with the -D option for hpux.

[Stephen Keane]

With hpux -D0x2a :WINSTALL

From the SEArch command ...

Main Menu: Enter command > sea

Searching for potential boot device(s)...
this may take several minutes.

To discontinue search, press any key (termination may not be immediate).

Path Number Device Path Device Type
----------- --------------- -----------
P0 LAN.10.0.0.1.3.254 10.0.0.1 10/Half Dx
P1 IDE CRD-8400B
P2 SCSI.0.0 HP C1533A
P3 FWSCSI.6.0 COMPAQ BD009122A
P4 FWSCSI.5.0 COMPAQ BD009122A

Main Menu: Enter command > bo p2
Interact with IPL (Y, N, Q)?> y

Booting...
Boot IO Dependent Code (IODC) revision 0

HARD Booted.

ISL Revision A.00.44 Mar 12, 2003

ISL> hpux -D0x2a :WINSTALL

... shed loads of stuff going past so fast I couldn't read it (CTRL-S ineffective) ...

pause (approx 30 seconds)

... shed loads of stuff going past so fast I couldn't read it (CTRL-S ineffective) ...


third to last colum appears to be 4 digit incrementing hex number.
second to last column seems fixed at 0x400
last column seems fixed at 0xed32dd8

blocks of 4 long lines followed by one short one
(ENTER ...can't read rest of line
(EXIT) ... can't read rest of line

So it looks like it isn't hanging, but is taking a long time to do something.

[Trond Haugen]

Your problem is most likely a unsupported tape drive.
I know from own experience that most tapedrives will work when just reading or writing data. But not all will work for booting. Supported ones will work.
I have also senn same problems with DCroms.

Regards,
Trond

[Stephen Keane]

So both the AIT drive and the DDS drive are unsupported? Even though the DDS drive shows up as a HP drive? Ohh, tough audience.

When I try and boot from the DDS drive with -D0x2a, it finally freezes with the following ...

(ENTER) ldv_read)
|->:ldv_read: 0x02000100 0x8 0x2000 0xee1a860
(ENTER bttape_strategy)
(EXIT bttape_strategy: 1)
(EXIT ldv_read: 8192)
|->:ufs_idy: Bad fs magic
(ENTER ldv_close(0x02000100))
(ENTER bttape_close)
(ENTER tds_unchain)
(EXIT tds_unchain: 1)
(ENTER tds_dealloc)
(EXIT tds_dealloc: 1)
(EXIT bttape_close: 1)
(EXIT ldv_close: 0)
(ENTER ufs_dealloc_fd(0xb62d48) [3])
(ENTER ufs_valid_fd(3))
(EXIT ufs_valid_fs: returning 0xed5f680)
(EXIT ufs_dealloc_fd: 0)
(EXIT ufs_idy: 0)
(ENTER lfs_fdealloc)
(ENTER lfs_fvalid)
(EXIT lfs_fvalid: 1)
(EXIT lfs_fdealloc: 1)
(EXIT lfs_open: 2)
(ENTER lfs_close)

Then it hangs.

So if the tape drive(s) are unsupprted, how can I create my own LIF and boot as far as loading the kernel successfully, but barfing because the root VG isn't available?

[Cheryl Griffin]

The tape drive must be an actual HP drive in order to read the bootlif. Non-HP drives don't know how to read the tape appropriately.

Do you have an HP drive you could use? Call your local HP hardware, they may be able to let you borrow/rent one.

Without the HP tape drive, your options are to use some other form of backup. Install the system you're trying to recover from Core-OS media and restore from the backup.

[Stephen Keane]

Curiouser and curiouser, if I create a LIF volume on the tape using the AUTO, HPUX and ISL from the Ignite tape and mix in the WINSTALL and WINSTALLFS from my installation CD, it boots fine. Seems to suggest to me (call me naive) that there isn't much wrong with the tape drive?

ISL on the CD is 306 sectors, on the Ignite tape is 242 sectors. HPUX on the CD is 848 sectors, on the Ignite tape is 1024 sectors. The Kernels/Ram FSs are different sizes too.

One thing I haven't tried is dd'ing the Ignite LIF from the tape and re-dd'ing it back on again with a block size of 2k (which is how I create the LIF volumes on tape manually) ...

[Cheryl Griffin]

http://forums1.itrc.hp.com/service/forums/questionanswer.do?threadId=335447

[Stephen Keane]

Using the DDS drive, the re-dd'ing of the LIF volume didn't make a difference.

If I create a LIF volume with AUTO (manually created), HPUX and ISL from /usr/lib/uxbootlf, WINSTALL and WINSTALLFS from install CD (no INSTALL, INSTALLFS, VINSTALL or VINSTALLFS, PAD, FWWARK# etc, it boots from both the DDS drive and the AIT drive (with different tapes obviously). If I use the WINSTALL from the Ignite recovery tape, it doesn't boot.

I've tried swremove'ing and swinstall'ing Ignite, makes no difference.

[Stephen Keane]

Now it's just getting surreal ...

If I create an Ignite recovery tape on serverA on a Quantum DLS tape, dd the bootable LIF part off onto disk, rcp it across to serverB and dd it onto the DDS tape attached to serverB, then boot serverB from the DDS tape it works (i.e. boots up to the Ignite option menu)

If I create an Ignite recovery tape on serverA on its DDS tape and try and boot from it, it doesn't work! (i.e. hangs trying to load the WINSTALL kernel)

Does this make sense to anyone? Where are Mulder and Scully when you need them?

[Stephen Keane]

OK, after a good nights sleep, the answer became a little more clear.

The J class server (the one that creates Ignite tapes that actually work) has B.3.6.82 version of Ignite on it.

The B class server (the one that creates Ignite tapes that don't work) has C.6.0.109 version on it.

I downloaded the C version, so I know where that came from. The installation CD has B.3.3.116 version on it. So I don't know where B.3.6.82 came from.

The problem appears to be with the C.6.0.109 version of Ignite running on a B class server running 11.11.

The tape drives appear to have little to do with the problem, unless C version of Ignite is somehow more concerned about them.

[Stephen Keane]

The C version of Ignite doesn't work on the B1000 that I am using. When I revert to a B version B.3.3.116, it works fine.

I don't know whether the C version is supposed to work on a B1000, in which case HP have a 'feature' in the product, or if it is unsupported in which case it would be nice if that knowledge was a little more widely publicised. Either way I'm not getting paid to sort out someone else's problem.