URGENT Tape I/O Error 1

[juggers]

Hi all,

I have an E250 (solaris 2.6) I want to know if the tape driver is damaged (DDS3) or it is something with the files. when i give a tar cvf /dev/rmt/0 data the tar starts to write to tape. but after a minute it gives tar error :I/O error. whatever directory i try it writes for a minute then gives error even with CPIO.... How can i check if the Tape driver is damaged... why it writes for a certain time then faults? ...I changed the tape media.

Thanks alot

time is the father of truth

 

[bfitzmai]

I have seen this problem with the 160 meter tapes... For some reason the tape drive that I had didn't like the 160 meter tapes. If you are using the 120 meter tapes, I would think the tape drive is bad. Have you tried to clean the tape drive?

 

[juggers]

Hi,
I have cleaned the tape many times. I am using the 125 meters HP DDS3. I will try to get a new cleaning tape as well. No other way to check ?


Thanks alot

time is the father of truth

 

[coffeysm]

If it is a external SCSI drive I would also check terminators, SCSI ID, and the actual cable.

 

[KenCunningham]

What does mt -f /dev/rmt/0 status show you? Try a 90 Metre tape.

 

[daFranze]

do you have errors in the messages file?
did you chnage something within the SCSI-Bus? New Devices?
did you edit st.conf etc.?

Best Regards, Franz
--
Solaris System Manager from Munich, Germany
I used to work for Sun Microsystems Support (EMEA) for 5 years

 

[juggers]

Hi all,
# mt -f /dev/rmt/0 status
HP DDS-3 4MM DAT tape drive:
sense key(0x6)= Unit Attention residual= 0 retries= 0
file no= 0 block no= 0
I did not change anything in the system i added 2 hard 2 months ago and the daily backup was going fine!!! only 2-3 days back things are not going fine.
in the messages when the tape is failing is showing the following message:
Dec 15 21:33:35 unix:
WARNING: /pci@1f,4000/scsi@3/st@5,0 (st5):
Dec 15 21:33:35 unix: Error for Command: read Error Level: Fatal
Dec 15 21:33:35 unix: Requested Block: 0 Error Block: 0
Dec 15 21:33:35 unix: Vendor: HP Serial Number: 62 951
Dec 15 21:33:35 unix: Sense Key: Media Error
Dec 15 21:33:35 unix: ASC: 0x11 (unrecovered read rror), ASCQ: 0x0, FRU: 0x0
I did not edit st.conf!

Thanks alot guys


time is the father of truth

 

[KenCunningham]

I would say that media error indicates a bad tape. Have you tried another?

 

[daFranze]

yep
1) media error may indicate faulty tape or
2) dirty/defect drive (head)

I have seen this many times with DLT Drives, so why not a DDS Drive...

Best Regards, Franz
--
Solaris System Manager from Munich, Germany
I used to work for Sun Microsystems Support (EMEA) for 5 years

 

[juggers]

Hi thanks for all the reply,

I think i will change the complete drive, I think there is no need to change anything ...OS wise? Just unplug the defective drive and plug the new one and things should be fine ??

Thanks alot for all the contributions

time is the father of truth

 

[bfitzmai]

Just to be on the safe side. Once you install the new tape drive, boot the server with a boot -r.

 

[daFranze]

if you connect the same make and model there is no need to do a reconfiguration (should be the same scsi ID, too) boot; I suggest to shutdown host, since "standard" SCSI is not hot pluggable...

Best Regards, Franz
--
Solaris System Manager from Munich, Germany
I used to work for Sun Microsystems Support (EMEA) for 5 years

 

[juggers]

Hi all,
I will try to get the same part number for DDS3 and i will shut down ..unplug the defected parts and plug the new one.

In case i decided to go for a different tape driver like a dds4 on the same scsi...then what do i need to do?

Regards

time is the father of truth

 

[daFranze]

as bfitzmai pointed out: run a reconfig boot (boot -r), this will generate a new Tape Device (when your Drive WAS /dev/rmt/0 the NEW one will get /dev/rmt/1)

Best Regards, Franz
--
Solaris System Manager from Munich, Germany
I used to work for Sun Microsystems Support (EMEA) for 5 years

 

[juggers]

I see,

Even if the tape is from different brand then the current one? the boot -r will do? Another thing does the DDS4 read tapes written with DDS3?

Thanks again

time is the father of truth

 

[daFranze]

depending on the hardware scan, reconfiguration will generate a new device, if you replace a DDS3 onebrand by a DDS3 otherband, I GUESS, you won't get a new devicenumber.

Afaik DDS3-written Tapes can be read in a DD4 Drive but not vice versa

(btw: DLT 7k drives "format" Tapes, so even if you want to overwrite them in a DLT4k Drive the Drive ejects the tape with error, maybe you face similar problems with DDS 3&4 Drives)

Best Regards, Franz
--
Solaris System Manager from Munich, Germany
I used to work for Sun Microsystems Support (EMEA) for 5 years

 

[MenaceEnnis]

Hi ,
Small tip.. Sun Tape drives are in fact rebadged HP drives, If your stuck you can swap them with HP drives

 

[juggers]

Ecactly,
in the messages file it mentiones that the vendor is HP.
One very important thing i was reading regarding boot -r I quote:

-----------
NOTE:

There is a problem with 105181-33!!! Do not install it if possible.

Do not run boot -r (or touch reconfigure) if you have kernel patch
105181-33 on Solaris 2.6 servers, or all device trees and files can be wiped
out!
The following must be added to /etc/system to fix this bug:
set kobj_map_space_len=0x400000
Is that tru ? how can i know that I have this patch installed or not.

Something freaky.. any suggestions?

Regards

time is the father of truth

 

[bfitzmai]

You can look in /var/sadm/patches to see if the patch number is there.

 

[juggers]

Hi all,

I did install the DDS3 tape driver, same brand, same everything....
when i try a mt command, the following error message is command ... dev/rmt/0n: No such device or address

PLease help.

Regards


time is the father of truth