cancel
Showing results for 
Search instead for 
Did you mean: 

TapeAlert help (\netbackup\db\media\errors)

Roobix_Cube
Level 5

Can anyone provide a simple method of translating the TapeAlert codes returned in the <install_path>netbackup\db\media\errors file?  I've read through http://support.veritas.com/docs/284473, but the explanation requires a degree in mathematics.  Alternatively, does anyone know of a tool/website that converts hex into the SCSI flags?

 

 

1 ACCEPTED SOLUTION

Accepted Solutions

Darren_Dunham
Level 6

It's just showing you the flags in hex from left to right.  4 flags in each hex digit.  Add four for each '0' to the left, then add 1 for an '8', 2 for a '4', 3 for a '2' and 4 for a '1'.

 

One of your examples was: 0x00800000 0x00000000

 

So two zeros on the left, add 4 for each => 2 * 4 => 8

Next digit is 8, so add 1.   8 + 1 => 9.  This is flag 9.

 

This works fine for determining the flag number of any single flag.  If you have multiple flags in different digits, just pretend the others aren't set.  So this one in my logs:   0x28000004 0x00000000

 

Would be flags 3, 5, and 30.

 

Little more complex if you have multiple flags in a digit. You have to break it into the components.

 

I have a lot of errors of flag 55 (which is just past the end of the flags that are discussed in the HP document).  I remember asking them about it, but we didn't come up with a specific reason for that flag to alert.  Caused a *lot* of problems.

 

Note that the BPTM log records the same data in two forms.  Once with the hex display and once with the decimal display.

 

02:48:09.651 [364] <2> send_MDS_msg: DEVICE_STATUS 1 52778 blah blah blah blah TAPE_ALERT 2048 0
02:48:09.651 [364] <2> JobInst::sendIrmMsg: starting
02:48:09.653 [364] <2> JobInst::sendIrmMsg: returning
02:48:09.679 [364] <2> log_media_error: successfully wrote to error file - 01/01/09 02:48:09 blah 30 TAPE_ALERT blah 0x00000800 0x00000000

 

0x000000800 and '2048' tell about the same flag.

 

--

Darren

View solution in original post

6 REPLIES 6

Amit_Karia
Level 6

Can you write down one of the lines?

Roobix_Cube
Level 5

Here are the last two lines, as an example.  It's the last two sets of numbers I am asking about:

 

07/07/08 07:36:18 RUA253 1 TAPE_ALERT HP.ULTRIUM3-SCSI.001 0x00800000 0x00000000
07/09/08 20:01:13 NW0119 1 TAPE_ALERT HP.ULTRIUM3-SCSI.001 0x00000000 0x00002000

Amit_Karia
Level 6

Tape alert generally means you need to clean the tape drive... are you getting tape alerts on bptm log

file also..

follow the link

 

http://www.t10.org/ftp/t10/document.02/02-142r0.pdf 

 

I hope it would be helpful

Roobix_Cube
Level 5

Thanks Amit, but I've been through that document, as per the link I provided in my first post.  The HP TapeAlert doco link you provided lists *many* different types of flags; I'd like to know if anyone has an easy way of finding out exactly what flags are being generated in the \netbackup\db\media\errors file.

 

Also, I disagree with the statement, "Tape alert generally means you need to clean the tape drive".  Tape alerts can mean much more than simply a dirty drive...

Darren_Dunham
Level 6

It's just showing you the flags in hex from left to right.  4 flags in each hex digit.  Add four for each '0' to the left, then add 1 for an '8', 2 for a '4', 3 for a '2' and 4 for a '1'.

 

One of your examples was: 0x00800000 0x00000000

 

So two zeros on the left, add 4 for each => 2 * 4 => 8

Next digit is 8, so add 1.   8 + 1 => 9.  This is flag 9.

 

This works fine for determining the flag number of any single flag.  If you have multiple flags in different digits, just pretend the others aren't set.  So this one in my logs:   0x28000004 0x00000000

 

Would be flags 3, 5, and 30.

 

Little more complex if you have multiple flags in a digit. You have to break it into the components.

 

I have a lot of errors of flag 55 (which is just past the end of the flags that are discussed in the HP document).  I remember asking them about it, but we didn't come up with a specific reason for that flag to alert.  Caused a *lot* of problems.

 

Note that the BPTM log records the same data in two forms.  Once with the hex display and once with the decimal display.

 

02:48:09.651 [364] <2> send_MDS_msg: DEVICE_STATUS 1 52778 blah blah blah blah TAPE_ALERT 2048 0
02:48:09.651 [364] <2> JobInst::sendIrmMsg: starting
02:48:09.653 [364] <2> JobInst::sendIrmMsg: returning
02:48:09.679 [364] <2> log_media_error: successfully wrote to error file - 01/01/09 02:48:09 blah 30 TAPE_ALERT blah 0x00000800 0x00000000

 

0x000000800 and '2048' tell about the same flag.

 

--

Darren

Roobix_Cube
Level 5
Nicely put, Darren.  Thank you.