fbackup is failing with "WRITE ERROR while writing data record"

[Juan Ocasio]

Hey all:

I'm having a problem with my fbackup. everytime I try to run it interactively I recieve the above error message, followed by
"attempting to make this volume salvagable"
and then:
could not read the previous checkpoint record.

I'm not too sure whaqt this means, but it happens everytime I try to backup a file system. Does any one know what this could be and how to fix it?

Any help would be greatly appreciated.

PS the above error is followed by
at media record 190 - This number changes depending on which file system I try to backup.

Thanks

Juan

[Sanjay_6]

Hi,

Looks like a media problem. Change the media and retry.

Hope this helps.

Regds

[Florian Heigl (new acc)]

does it also happen with fbackup -0 (maybe even -d /dev/null)?

If yes, I'd point my finger at Your tape drive or media (check stm's logtool in that case), otherwise, You might need to fsck /var or at least reinitialize fbackup's datafile.

But the tape drive / tape thingy is about 50 times more probable :)

[Peter Godron]

Juan,
also use the cleaning tape. Should be part of regular maintenance as is tape replacement and regular recovery tests!
Regards

[Juan Ocasio]

Thanks all:

I'd just cleaned the tape drive the other day(Monday), but I'll do this again. Maybe I need to get another tape cleaner (the one I have is still good for another 15 or so cleanings).

Thanks for the speedy reply. Any other thoughts are certainly welcome!!!

Juan

[Juan Ocasio]

All:

I'mn running the backup now and the first thing it does (after the session begins message) is print the following message:

WARNING: unable to read a volume header

Is this normal?

Thanks again,

Juan

[Bill Hassell]

Yes, it is quite normal. fbackup always checks to see if the new tape has been previously used by fbackup. The reason is that it wants to track how many times the tape has been used and to do this, the "volume header" is written onto the front of every fbackup tape. You can read this with:

frecover -V - -f /dev/rmt/

So when there is nothing on the tape, or the tape was written with some other program, you'll get the error. This is to warn you that if you thought this tape had been previously used by fbackup, it hasn't.

As far as the media error goes, this is normal for fbackup when it encounters a tape failure. You will also see the tape error messages recorded in syslog. If not, install the online diagnostics ASAP, either from your latest SupportPlus CD or downloaded from http://www.software.hp.com/portal/swdepot/displayProductInfo.do?productNumber=B6191AAE

The online diags are critically important in providing detailed records (and optional notifications by email, pager, etc) when a hardware failure has occurred.

[Juan Ocasio]

Sorry for the back-to-back-to-back post, but now my clean light is displaying and nothing seems to be happening on the box. The cursor is still sitting under the
WARNING: unable to read a volume header

message. Is this typical? I'm going to go out and get a new cleaning tape to see if this resolves my problems.

Thanks, and again any info is greatly appreciated!

Juan

[Sanjay_6]

Hi,

If this was a new media with which you got a "unable to read header" it is okay.

also make sure you are using the rewind device "eg /dev/rmt/0m" and are not using a no-rewind device "eg /dev/rmt/0mn"

Hope this helps.

Regds

[Bill Hassell]

The CLEAN light means that an excessive number of retries and errors have been detected. Because tape drives today are recording at such high density, the exact nature of the problem is quite difficult to determine. Rule out the tape itself bu first using a new cleaning tape, then a new tape. The CLEAN light should go out after the cleaning tape is run, but if it comes back on with new tape, you've very likely got a tape drive failure and will need to get it replaced. Note that there are no replaceable components inside--the parts are extremely small and require laser-accurate alignment, and the tape path must be perfectly clear for the media to work correctly. This applies to all modern tape drives: DDS, DLT, AIT, Ultrium, etc.

[Juan Ocasio]

OK. So I think I may have a Tape Drive issue. I used a tape drive cleaning tape and used a new Tape, but I still run into the problem. I have another Tape Drive (external) on a test box that we do not use. Can I simply add this tape drive to my system and have it work? Should I start a new thread with this question? If not (for the previous question) how do I install the drive.

Thanks again

Juan

[Bill Hassell]

Well, with several hundred HP-UX models to choose from and several dozen possible tape drives, we'll need more details: model of the HP-UX boxes, and model of the tape drive. Run these two commands on each box:

# model
# ioscan -kfCtape

[Juan Ocasio]

Bill:

thanks for the reply. Here is the info for both machines:

Production:
# model
9000/800/K380
# ioscan -kfCtape
Class I H/W Path Driver S/W State H/W Type Description
======================================================================
tape 0 10/12/5.0.0 stape CLAIMED DEVICE HP C1537A
#

Test:
# model
9000/800/A400-6X
# ioscan -kfCtape
Class I H/W Path Driver S/W State H/W Type Description
=====================================================================
tape 0 0/0/1/0.3.0 stape CLAIMED DEVICE HP C5683A
#

Again. Thanks for your help.

Juan

[Bill Hassell]

The C1537A in the K380 is a DDS3 while the A400 has the C5683A is a DDS4. They use the same driver (assuming the K380 is running 11.11). They are set to different SCSI addresses so you'll need to change the DDS4's address when you swap drives.

[Bill Hassell]

The C1537A in the K380 is a DDS3 while the A400 has the C5683A is a DDS4. They use the same driver (assuming the K380 is running 11.11). They are set to different SCSI addresses so you'll need to change the DDS4's address when you swap drives. As far as the mechanics of taking them out and reinstalling, if you haven't taken the boxes apart in the past, I'd recommend an HP engineeer do it for you.

[Juan Ocasio]

Thanks again for your assistance. THe DDS4 is an external Drive. Can I just plug it into the K380 and leave the other drive in the box?

Thanks,

Juan

[Bill Hassell · Accepted solution]

Aha. That's real easy. Just plug it into a single-ended SCSI (after you verify that there is not another device on that bus with address 3). This command:

# ioscan -kf

will draw a map of all the devices on each interface card.

[Juan Ocasio]

Bill:

Thanks a bunch. I'll Give this a shot when I get back to the office on Tuesday. I'd like to award you a 10 for really helping me through this 'little' issue of mine and if I may, I'd like to ask some more questions on Tuesday (just in case...).
Thanks again,

Juan

[Juan Ocasio]

All:

Here is the results for the ioscan:

Class I H/W Path Driver S/W State H/W Type Description
========================================================================
bc 0 root CLAIMED BUS_NEXUS
bc 1 8 ccio CLAIMED BUS_NEXUS I/O Adapter
bc 2 10 ccio CLAIMED BUS_NEXUS I/O Adapter
ext_bus 0 10/0 c720 CLAIMED INTERFACE GSC built-in Fast/Wi
de SCSI Interface
target 0 10/0.3 tgt CLAIMED DEVICE
disk 0 10/0.3.0 sdisk CLAIMED DEVICE SEAGATE ST136475LC
target 1 10/0.4 tgt CLAIMED DEVICE
disk 1 10/0.4.0 sdisk CLAIMED DEVICE SEAGATE ST39236LC
target 2 10/0.5 tgt CLAIMED DEVICE
disk 2 10/0.5.0 sdisk CLAIMED DEVICE SEAGATE ST39236LC
target 3 10/0.6 tgt CLAIMED DEVICE
disk 3 10/0.6.0 sdisk CLAIMED DEVICE SEAGATE ST39236LC
target 4 10/0.7 tgt CLAIMED DEVICE
ctl 0 10/0.7.0 sctl CLAIMED DEVICE Initiator
bc 3 10/4 bc CLAIMED BUS_NEXUS Bus Converter
tty 0 10/4/0 mux2 CLAIMED INTERFACE MUX
ba 0 10/12 bus_adapter CLAIMED BUS_NEXUS Core I/O Adapter
ext_bus 2 10/12/0 CentIf CLAIMED INTERFACE Built-in Parallel In
terface
ext_bus 1 10/12/5 c720 CLAIMED INTERFACE Built-in SCSI
target 5 10/12/5.0 tgt CLAIMED DEVICE
tape 0 10/12/5.0.0 stape CLAIMED DEVICE HP C1537A
target 6 10/12/5.2 tgt CLAIMED DEVICE
disk 4 10/12/5.2.0 sdisk CLAIMED DEVICE HP DVD-ROM 304
target 7 10/12/5.7 tgt CLAIMED DEVICE
ctl 1 10/12/5.7.0 sctl CLAIMED DEVICE Initiator
lan 0 10/12/6 lan2 CLAIMED INTERFACE Built-in LAN
ps2 0 10/12/7 ps2 CLAIMED INTERFACE Built-in Keyboard/Mo
use
bc 4 10/16 bc CLAIMED BUS_NEXUS Bus Converter
processor 0 32 processor CLAIMED PROCESSOR Processor
processor 1 34 processor CLAIMED PROCESSOR Processor
memory 0 49 memory CLAIMED MEMORY Memory

Please let me know if it looks like there is an available SCSI port.

Thanks,

Juan

[Bill Hassell]

The internal SCSI-SE adapter (connected to the CDROM and your DDS-3 tape) has a hardware path of 10/12/5. You can see the map with:

ioscan -kfH 10/12

10/12 is the Core I/O board in the back and 10/12/5 is the SCSI interface. You'll see a high-density SCSI-SE connector on that board. You may need an adapter cable to go from the SCSI-SE 50-pin connector to the DDS4 68-pin connector. NOTE: there is a Fast-Wide High-Voltage SCSI interface at 10/0. While the connector is the same as your DDS4, the voltages are completely incompatible. The FWD-HV interface is designed to run very long distances and is not compatible with any other SCSI standard.

[Bill Hassell]

Almost forgot: The SCSI addresses that are in use are 0 (DDS-3 tape) and 2 (CDROM. The SCSI addresses are the numbers following 10/12/5. You can set the DDS4 drive to 1,3,4,5 or 6.

[Juan Ocasio]

Bill. Thanks again for your help. So I can connect my DDS4 Tpae to the SCSI port on the back of the box? (If I get the right connector?). If this is the case, do I have to power down the machine? Also how do you chnage the address so that it does not conflict with the other two devices. SOrry for all of the hand-holding. I'm pretty new to these HP machines (and to UNIX for that matter).

Thanks again for your assitance!

Juan

[Bill Hassell]

There are two SCSI cards on the back of your K380. One is connected to 4 disks, probably the internal disks. In the back of the K380, there will be a Core I/O board that has several connections (serial, parallel printer and SCSI). The SCSI connector is fairly small, two rows of 25 pins.

The DDS4 tape drive should have a SCSI address switch on the back. Set this to an unused address (not 0 or 2). Since the Core I/O SCSI-SE is connected to the CD and internal tape, you should have no problems connecting the DDS4 to that interface while the system is running. Make sure there is a terminator on the back of the DDS4 drive.

To get the new hardware active in the kernel, turn the DDS4 on and then type:

ioscan -H 10/12/5

This will take a while (the -k option doesn't look at the actual hardware). Do the ioscan a couple of times to get the new drive to show up. Once you see it, run the command:

insf -e

to create the new device files. The new names will scroll off the page but you can find them with:

ioscan -knfC tape

Use the mt command to talk to the new tape drive:

mt -f /dev/rmt/?m status

where ?m is the new tape drive's name. Now try your backup using the new tape drive.

[Juan Ocasio]

I just checked the back of my K380 and there is a SCSI Terminator (C2905A) on it. Can I just remove this Terminator? I believe it's a 68 pin SCSI port and There are two ports on the back of the DDS4; the narrower one goes to my test box and the wider one seems to have a terminator on it. Just out of curiosity, do all unused SCSI ports need a terminator?

Again thanks for the help!

Juan

[Bill Hassell]

Nope. The C2905A is a FWD-HV terminator for the internal disks and is not compatible with any other SCSI devices. I looked on several K-boxes and I see that only the FWD-HV connectors are on the Core I/O board. I think I remember now that SCSI-SE is internal only, there is no external connector for the SCSI-SE bus inside the box. You'll have to get an additional I/O card with SCSI-SE on it. I don't believe there are any SCSI-LVD I/O cards for the old K-380's. So the DDS4 isn't going to work until you get a nother card.