BAckup tapes full, but not true

[paolo07]

HI,
I'm using a LTO Ultrium 2 device, with SolsticeBackup 7.2 Application installed, connected to SUN V4900 server, Solaris 10 01/06 installed.
I'm experiencing this kind of problem: new tape is labelled, start a backup, but it always set the Tape to FUll, even if the size is bigger.
Here messages I recieved:

nsrd: media warning: /dev/rmt/1cbn writing: Invalid argument, at file 2 record 237
08/29/07 22:17:10 nsrd: media notice: LTO Ultrium-2 tape OMCplus.003 on /dev/rmt/1cbn is full
08/29/07 22:17:10 nsrd: media notice: LTO Ultrium-2 tape OMCplus.003 used 15 MB of 200 GB capacity
08/29/07 22:17:17 nsrd: media notice: Volume "OMCplus.003" on device "/dev/rmt/1cbn": Block size is 32768 bytes not 65536 bytes. Verify the device configuration. Tape positioning by record is disabled.
08/29/07 22:17:17 nsrd: media info: verification of volume "OMCplus.003", volid 4292210621 succeeded.

I am not sure if it's a HW problem or only a driver configuration issue.
I have found something in

http://www about

the st.conf configuration for the AUTODENSITY, but I really don't know what/how to check it.

PS: strange thing is that same configuration in another machine is working fine.

CAn someone help me? did u see the same?
Thks,
Paolo

 

[605]

The problem seems to be here:
"Block size is 32768 bytes not 65536 bytes. Verify the device configuration."

The block size is also set in the label. You should change it or use another tool to overwrite the label.

 

[paolo07]

Thanks,
but I have already tried with a new tape and this is the same.

My doubt is if there is something that set the block size at the beginning, during the first labelization .... and this is what I cannot control/set/check....

DO you know something?
Thnks,
P.

 

[605]

Sure - this will is set within the label. If you want to delete a label, simply use another software (dd etc.) or NW's tapeexer(cise) utility.

 

[paolo07]

Thanks,

since the block size is set in the Label, can I have a script/way to read the actual block size inside the current label of the current tape?

Just to check if it's really what the messages told me and tune it afterward.

Since it seems that re-labeling it, I did not solve my problem....

Thnks,Paolo

 

[605]

Just use the "scanner" command - it will tell you and abbort the process with ^C.

 

[paolo07]

Hi,

scanner is done and here the result:

scanner: volume id 14777791 record size 65536

The record size is OK, 65KB, but the daemon.log is again complaining that

device "/dev/rmt/1cbn": Block size is 32768 bytes not 65536 bytes

From nwadmin, in the Device Settings, I have the
CDI= SCSI command
Device Block Size=handler default

I can see that I can set the 64KB as the Block size. Do you think this can be a good chance to Set it to 64KB and then label a new Tape? is this going to erase the current tape configuration and set a new one with new Block size?
Thanks,
Paolo

 

[ratmcore]

change block SIZE 0byte.

clock SIZE 0 = virity block size.


sorry . i`m not very well English.

 

[paolo07]

Hi,

checking more carefully the logs, AND after the last modifications (not successfull), I have found this:

Tue 10:52:23 AM media warning: /dev/rmt/1cbn writing: I/O error, at file
2 record 76
Tue 10:52:23 AM media notice: LTO Ultrium-2 tape OMCplus.001 on
/dev/rmt/1cbn is full
Tue 10:52:23 AM media notice: LTO Ultrium-2 tape OMCplus.001 used 4800
KB of 200 GB capacity
Tue 10:52:27 AM media notice: Volume "OMCplus.001" on device
"/dev/rmt/1cbn": Block size is 32768 bytes not 65536 bytes. Verify the
device configuration. Tape positioning by record is disabled.
Tue 10:52:27 AM media info: verification of volume "OMCplus.001", volid
4243565576 succeeded.

There is an I/O Error and due to this the SBU should work that it set the tape to full.

Do you think this can cause the error?
Any sugegstions about the I/O error, how to check, possible HW problem?
Thnks,
Paolo

 

[Rif123]

Sounds like a hw problem. Does any other of your tape drives have the same problem or is it only the /dev/rmt/1cbn?

When NetWorker hits a IO problem it can't recover from it marks the tape full. You can always try to turn off CDI and see if that makes a difference, but with LTO it usually isn't a problem.

Can you find any more info in the syslog?

 

[paolo07]

I have only this device for backup....

In the logs I can see different times the same I/O message in writing phase, everytime the backup fails with the same message.

I don't understand your sentence about turn off the CDI: in fact LTO should use SCSI commands and this should not have any relations.

What to check about HW problem?

 

[Rif123]

CDI was introduced in NetWorker 7 and is a new API used for communicating with tape devices. CDI helps because it have (sometimes) better clearer error messges and it uses Tape Alert. There have been lots of problems with CDI and certain tape drives.

The best way then usually is to turn off CDI and use the old way off communcating with the tape drive. This is a attribute on the tape device resource in NetWorker you can change from "SCSI commands" to "Not used".

CDI is a good thing when it is working so you should only change it for test purposes and see if it makes any difference for you. However, I have never had problems with LTO and CDI so far.