SJI Failure... help!!!!

[gpadilla]

I get the following error when I try to inventory, eject, or deposit a tape.

nsrjb: Jukebox error: Failure[0x 29]:Vendor Unique, ASC 0xff ASCQ 0x73.

Any ideas?? This is running on a solaris platform, and I have restarted networker and the same error occurs...

any Ideas???

 

[Guest_imported]

Reboot!

 

[Rexxy]

reboot ?? it isnt nt.... (hehehe), oh you mean the jukebox.... B-)

the ASC errors are related to the jukebox, and the vendor can usually desipher their true meanings....I cant remember the codes, but it could range from "door" issues, firmware issues, scsi probs, robot arm etc....

Also check out your st.conf and make sure it is configured correctly for the tape drive models.

 

[ejw77]

I had his error: "nsrjb: Jukebox error: Fri 11:00:08 SJI Failure[0x 29]: No Additional Sense, No Addtional Sense Information"

Eventually I solved it to disable this drive via the gui.
Run nsrjb -H from the shell (Solaris 2.6)
Run inventory from the gui on selected a drive that was not disabled.
enabled the drive I just had disabled.
run nsrjb -H again from the shell.
And run inventory again from the gui but slected now the drive that previously was disabled.

Evert Jan

 

[gpadilla]

OK.. This problem has resurfed again!! I get an SJI Failure[0X 29]:Illegal Request, ASC 0x80 ASCQ 0x01 whenever I try to do an nsrjb -I... basically any requests I try to do. I have rebooted the jukebox the server and the same problems occurs. Any Ideas???

 

[barye]

How to correct jukebox problems
Introduction
There are many reasons why a jukebox may not respond to NetWorker commands. Often an error message from NetWorker during a backup operation is returned.

Why does a jukebox become unresponsive or return unexpected results to NetWorker requests or commands?

Here are three common reasons for jukebox problems:

Manual intervention: Most often, a jukebox no longer responds to backup requests because of operator intervention. Manually moving, unloading or loading tapes causes a discrepancy between NetWorker's information on the status of the jukebox and the actual physical status of the jukebox. On jukeboxes that support barcodes and/or element status, a special feature lets the jukebox know if the door is opened. Since the jukebox has no way of knowing what the operator did while the door was open, it goes into a state which reflects that uncertainty, and in many cases does not function properly until that uncertainty is removed.

If a tape has been manually loaded or unloaded to or from a drive, try replacing the tape in its original configuration, then retry the NetWorker operation. Often a reset, re-inventory and/or power cycle is necessary (described below).

Corrupt nsrjb.res file: Another possible cause is the corruption of the jukebox resource file, nsrjb.res. Corruption may occur to an open nsrjb.res file during a system crash.

Hardware or software problems: Other hardware and software problems may also cause corruption to nsrjb.res or cause the jukebox to become unresponsive or return unexpected results in response to NetWorker requests.
The following steps are in order based upon simplicity, risk, and time. Be sure that the first items are done before proceeding to the next items. Before you begin, check that all required NetWorker patches have been installed, and verify proper SCSI bus and jukebox driver installation using suitable tools and utilities.
Verify jukebox configuration

Using the NetWorker Administrator's GUI go to the Jukeboxes window. Verify the following data fields against the hardware:

Available slots: This field should be a range of slots, i.e. 2-10, as opposed to a single digit.

Control port: This field should match the entry used for control port in such commands as pscinfo, jbexercise, etc.

Devices:

Verify that the proper device names are being used, and that the names are physically in the jukebox:

- On EXB-210's the first device listed in this field should be the top device in the jukebox.
- On EXB-60's and 120's the device names should be listed in order of physical location in the jukebox, from left to right.

Physical slots: This field should be listed as 1,10 as opposed to 1-10 in available slots.

Verify that model and number of devices match the physical jukebox.

If any of the above fields are incorrect, make corrections before continuing.

Volumes: Verify that the volumes in this field are physically in the jukebox by using the following command:

"nsrjb -Cv " (Provides a list of Volumes by slot; this should agree with the physical inventory).
If these lists do not agree, your jukebox is confused, therefore proceed with the subsequent troubleshooting steps.
Re-Initialize element status for jukebox with barcodes and/or element status

This section may be skipped for jukeboxes that do not support barcodes and/or element-status.

Opening the door may confuse jukeboxes that support barcodes and/or element-status.

If the jukebox door is manually opened, the jukebox immediately marks all the slots as questionable. Consequently, the jukebox status will not reflect physical changes, such as, removing a tape or replacing a tape, made by a user. To address this condition, the jukebox needs to check all of the slots and read the barcodes (if enabled) to verify the status of each slot.

To reset element status one of the following commands:

For NetWorker 4.1 and previous versions:

"nsrjb -HE -v" (Resets the jukebox hardware and element status).
"pscinfo -i /dev/sjid1u1" (Check the status)

For NetWorker 4.2 and later version

"nsrjb -HE -v" (Resets the jukebox hardware and element status).
"inquire" (Check the status)
If an error message is returned, try the commands again. The second try may correct the problem. If errors continue, you should analyze the errors and correct them before proceeding.
Inventory jukebox

For many simple problems, it may only be necessary to re-inventory the contents of the jukebox.
Use the following command to re-inventory the jukebox: nsrjb -I -v

A successful inventory indicates the jukebox problem is resolved. Monitor subsequent backups and jukebox operations to insure no other problems occur.

If the re-inventory step does not resolve your jukebox problem, continue with the subsequent troubleshooting steps.

Reset jukebox hardware

Some problems require a hardware reset to correct jukebox problems.
Use the following command: nsrjb -H -v

You may need to follow the hardware reset with the inventory operation listed above. A power recycle may also be necessary to fully reset the jukebox.

Fixing the confused jukebox resource file

The following procedure corrects jukebox confusion by erasing the existing inventory stored the NetWorker resource file, and then forcing the system to recreate a new inventory.

Login to the NetWorker server as root.

Start the NetWorker Administrator's GUI.

&quot;networker -s <server -x&quot; (NetWorker version 4.0.2.x)
&quot;nwadmin -s server&quot; (NetWorker version 4.1.x)

Select the Jukeboxes window.

- Select Admin - Jukeboxes (NetWorker version 4.0.2.x)
- Select Media - Jukeboxes (NetWorker version 4.1.x)
- Select View - Details

Clear contents of Loaded Slots and Loaded Volumes fields, then press the Change button.


Invoke the command prompt and type:

&quot;nsrjb -HE -v&quot; (the 'E' option will reset element status)
&quot;nsrjb -Iv&quot; (the &quot;I&quot; takes an inventory of the jukebox)
NetWorker and the jukebox should now be synchronized.
Completely reconfiguring the jukebox

If going through the steps listed above does not resolve your jukebox problem, the resource file, nsrjb.res, may be corrupt.

Try the one of the following suggestions to address a potentially corrupted nsrjb.res file.

Recover a previous version of the nsrjb.res file from a backup. You may need to recover more than one version of nsrjb.res, dependent upon when the corruption actually occurred.

Recreate the nsrjb.res file.

- Shutdown all NetWorker daemons by typing nsr_ shutdown -a from the command prompt.
- Type mv /nsr/res/nsrjb.res /nsr/res/nsrjb.res.old to rename the currentnsrjb.res file to nsrjb.res.old.
- Start the NetWorker daemons by typing nsrd at the command prompt.
- Run jb_configfrom the command prompt to configure the jukebox.
- Re-enter the jukebox enabler, if necessary.
PSCINFO man page reference
PSCINFO(8) MAINTENANCE COMMANDS PSCINFO(8)

NAME
pscinfo - NetWorker autochanger utility program

SYNOPSIS
pscinfo -c {INQUIRY | SENSE} device
pscinfo -i device
pscinfo -e device

DESCRIPTION
pscinfo performs a few diagnostic functions with auto- changers. A system administrator may find pscinfo useful for initializing or reporting the status of such a device
independent of the use of NetWorker.

Device is the name of the device (/dev/sjidNu1 or /dev/pscN;

see psc(8)).

OPTIONS
-c COMMAND
Special SCSI commands can be sent to the changer dev ice. The following key words and their descriptions are summarized below:
INQUIRY displays the device type, product code, product ID, and product revision strings. SENSE enables the changer to report on its operating
mode parameters. The element address assignment page is reported.

-i The Initialize Element Status command requests the changer to check all elements for the presence of a data cartridge. The changer stores this
information its cartridge inventory. Note that not all autochanger devices support this command.

-e The Read Element Status command requests that the changer report the status of its internal elements.
Note that not all autochanger devices support this command.

SEE ALSO

psc(8), nsrjb(8).

 

[barye]

Mount request fails - verify label operation in progress; Set load
load_sleep

Description:

Jukebox hangs during mount request.

One of the following errors may also be returned during the mount request:

tape: verify label operation in progress
media warning:tape <device name> opening io error
open error

ANALYSIS:

NetWorker is trying to open the device before the device is completely ready.

On a UNIX system the message &quot;verify label operation in progress&quot; is returned by
the nsrmmd program.

Once a volume is loaded into the drive the nsrjb program makes a request to have
the volume mounted. The mount request is sent to the nsrd program. Nsrd
allocates an nsrmmd to perform the mount and label operation.

The nsrmmd attempts to position the volume to read the label. If the attempt
fails, the message &quot;verifying label operation in progress&quot; is returned. The
reason for this failure is because the device is not available when nsrmmd is
attempting to access the device to read the label and mount the volume.

If the device is available, NetWorker will mount the volume.

Different errors may be returned depending upon the state of the device when the
nsrmmd is attempting to access the device to read the label and mount the
volume. Several of the errors are listed above.

To be more specific the NetWorker design does not check the drive's status
before trying to access or open the drive.

Note: Only device driver level applications can poll the device and wait until
the device is ready.

Occasionally, NetWorker may prematurely access a drive before the drive is ready
and cause a system to hang. The reason for this behavior is because the nsrmmd
is an application process and requires the device to be open before any reads or
mounts can be issued.

SOLUTION:

Set the load_sleep parameter to a value in seconds.

For example, &quot;load_sleep=300&quot; will cause nsrjb to wait for 300 seconds before
attempting the access the device.

The default value is 15 seconds.

If this option is not set (!load_sleep), then the default value is 0 seconds.

This delay was added to NetWorker for devices which return asynchronously from
the load operation before the device is ready.

Use the NetWorker Administration GUI (nwadmin) to set this parameter.

On UNIX:

Select Media => Jukeboxes => View => Details => Jukebox options
Type &quot;load_sleep=n&quot; where n is the number of seconds

On Windows NT:

Select Options => Customize => Check &quot;Display Hidden Attributes&quot;
Select Autochanger => Left click on specific autochanger and select Edit
Select Preference tab => Jukebox options
Type &quot;load_sleep=n&quot; where n is the number of seconds

 

[RichS24]

find out what the hardware vendor says, and then follow the route it leads you down.
You could spend hours trying to troubleshoot it through reboots and resets, when it's probably telling you there is a problem with the robot.
If you still have the manual for the library, check it for sense key codes, as they should be listed in there Cheers
richs24