cancel
Showing results for 
Search instead for 
Did you mean: 

Redhat Linux 2.6 media server: no-rewind device files

KEELIN_HART_2
Level 4
Hello. We have recently installed NBU 6.5.3 on a Rehat Linux server running a 2.6 kernel. According to the Device Config Guide, the /dev/nst* device files should be created automatically, but these have not been created.
ls -l /dev/nst*
ls: /dev/nst*: No such file or directory

Although I do have /dev/IBM* devices...
ls -l /dev/IBM*
crw------- 1 root root 252, 2048 Oct  7 15:30 /dev/IBMchanger0
crw------- 1 root root 252,    0 Oct  7 15:30 /dev/IBMtape0
crw------- 1 root root 252, 1024 Oct  7 15:30 /dev/IBMtape0n
crw------- 1 root root 252,    1 Oct  7 15:30 /dev/IBMtape1
crw------- 1 root root 252, 1025 Oct  7 15:30 /dev/IBMtape1n

I have already installed the IBM tape driver. should I be using the IBMtape* devices instead of the nst* ones, or is there a manual process for creating the latter. I am not very familiar with Linux operating systems...

Many thanks for any pointers...
13 REPLIES 13

Sriram
Level 6

If you have done your zoning of IBM tape drives to the linux box then check if the OS is able to see the drives by executing sgscan tape

If you are able to see the tapes then its straight forward configuration.  Login to Netbackup JAVA GUI, goto device configuration wizard and drives will be configured for you.

If sgscan tape is not displaying any drives then you may need to troubleshoot at OS level.

KEELIN_HART_2
Level 4
They are seen at OS level. It is just the no-rewind device files that aren't being created.

Output of 'scan'
Device Name  : "-"
Passthru Name: "/dev/sg2"
Volume Header: ""
Port: -1; Bus: -1; Target: -1; LUN: -1
Inquiry    : "IBM     ULT3580-TD4     89B2"
Vendor ID  : "IBM     "
Product ID : "ULT3580-TD4     "
Product Rev: "89B2"
Serial Number: "0007861848"
WWN          : ""
WWN Id Type  : 0
Device Identifier: "IBM     ULT3580-TD4     0007861848"
Device Type    : SDT_TAPE
NetBackup Drive Type: 3
Removable      : Yes
Device Supports: SCSI-3
Flags : 0x0
Reason: 0x10000


------------------------------------------------------------
Device Name  : "-"
Passthru Name: "/dev/sg0"
Volume Header: ""
Port: -1; Bus: -1; Target: -1; LUN: -1
Inquiry    : "IBM     ULT3580-TD4     89B2"
Vendor ID  : "IBM     "
Product ID : "ULT3580-TD4     "
Product Rev: "89B2"
Serial Number: "0007861696"
WWN          : ""
WWN Id Type  : 0
Device Identifier: "IBM     ULT3580-TD4     0007861696"
Device Type    : SDT_TAPE
NetBackup Drive Type: 3
Removable      : Yes
Device Supports: SCSI-3
Flags : 0x0
Reason: 0x10000

# cat /proc/scsi/scsi
Attached devices:
Host: scsi0 Channel: 00 Id: 00 Lun: 00
  Vendor: IBM      Model: ULT3580-TD4      Rev: 89B2
  Type:   Sequential-Access                ANSI SCSI revision: 03
Host: scsi0 Channel: 00 Id: 00 Lun: 01
  Vendor: IBM      Model: 03584L32         Rev: 8620
  Type:   Medium Changer                   ANSI SCSI revision: 03
Host: scsi1 Channel: 00 Id: 00 Lun: 00
  Vendor: IBM      Model: ULT3580-TD4      Rev: 89B2
  Type:   Sequential-Access                ANSI SCSI revision: 03
...

I will not be using the GUI to configure the drives. We are using SSO and using the GUI has created problems with this previously.

Sriram
Level 6
Oh...that's fine but it works for me perfectly using GUI.  I too have SSO.

Nicolai
Moderator
Moderator
Partner    VIP   
try running  /usr/openv/volmgr/bin/make_scsi_dev. It should create the /dev/nst drivers.


KEELIN_HART_2
Level 4
The make_scsi_dev command is specific for a 2.4 kernel. This is RedHat Release 5, with a 2.6 kernel.

Many thanks for any help.

Nicolai
Moderator
Moderator
Partner    VIP   
Yes, you are right, it's  for the 2.4 kernels, Sorry !

KEELIN_HART_2
Level 4

Hmmmm. Quite bizarre, I used mknod to create the /dev/st* and the /dev/nst* device files, but they have disappeared after rebooting. Does anyone have any ideas as to why this would happen?

zippy
Level 6
 there is a conf file on the server that has to edited to rebuild it every time it reboots.  Thats redhat for you

zippy
Level 6
 
i have 4 FC tape drives (LTO4) connected to redhat AS 4 through a brocade switch
the server is using 2 qlogic ql2462 hba cards

the sansurfer tool is able to see the 4 devices but don't know where is the device files related to them.

so where are the device file? if they don't exist how to do to create them?
how can i test the tape drives to be sure they are well configured?

 
Sort Answers By: Date or Points
Asif Sharif Expert in this area This member has accumulated 2500 or more points
Jun 11, 2008 10:57:21 GMT    Unassigned  

Hi Nappy,

please see the below web links.
http://www.redhat.com/docs/manuals/enterprise/RHEL-4-Manual/release-notes/as-s390/RELEASE-NOTES-U3-s...
http://forums11.itrc.hp.com/service/forums/questionanswer.do?admit=109447626+1213181650107+28353475&...

Regards,
Asif Sharif
Ivan Ferreira This member has accumulated 20000 or more points
Jun 11, 2008 13:09:13 GMT  8 pts  

>>> so where are the device file? if they don't exist how to do to create them?

Device files should be created after detected. You should see /dev/stX /dev/nstX.

Try with hp_rescan -a. You should install the support pack if this is a proliant server.
Nappy
Jun 11, 2008 13:12:24 GMT    N/A: Question Author  

Hi Ivan
this is an IBM server
Nappy
Jun 11, 2008 13:17:19 GMT    N/A: Question Author  

lsscsi command shows old scsi tape devices but don't show the new FC tape devices

is it the normal behaivor of lsscsi? is there another command that can show FC devices?
Ivan Ferreira This member has accumulated 20000 or more points
Jun 11, 2008 16:20:01 GMT    Unassigned  

Please see:

http://www.redbooks.ibm.com/redpieces/pdfs/sg246268.pdf

zippy
Level 6
 
 am new to Linux, but have been using tru64 for years. I need to install a FC tape library on a Linux server and don't have much of a clue where to start. 
In tru64 I would use 
# hwmgr -scan scsi 

Is there an equivalent function in redhat linux?

 
Sort Answers By: Date or Points
Ivan Ferreira This member has accumulated 20000 or more points
Jun 14, 2007 03:04:09 GMT  10 pts  

HP fibreutils packages provides the hp_rescan command. You can also download a script called rescan-scsi-bus.sh to rescan the devices.
Hemmetter This member has accumulated 500 or more points
Jun 14, 2007 03:04:49 GMT  10 pts  

Hi Graham,

first you need to know your fibre-HBA.
$ lspci | grep -i fibre

the (re)load the apropriate driver:
e.g. lpfc or qla2200v6 or ...

$DRVR=lpfc
$ #modprobe -r $DRVR # remove driver
$ modprobe $DRVR # load driver

then check if you see your tape(s):
$cat /proc/scsi/scsi

Then check Tape status:
$ mt -f /dev/st0 status




rgds
HGH
Ivan Ferreira This member has accumulated 20000 or more points
Jun 14, 2007 03:08:25 GMT    Unassigned  

See also: 

http://www.redhat.com/docs/manuals/enterprise/RHEL-4-Manual/release-notes/as-s
...


Support for fabric rediscovery through sysfs is now available in Red Hat Enterprise Linux 4 Update 3. For the Qlogic (qla2xxx) and Emulex (lpfc) Fibre Channel HBA drivers, run the following commands to perform redisocvery and rescan for new storage: 


echo "1" > /sys/class/fc_host/hostXYZ/issue_lip 
echo "- - -" > /sys/class/scsi_host/hostXYZ/scan 


In some Fibre Channel configurations, an operating system will automatically detect the addition of a new target device while the system is running. In other configurations, it is necessary to issue the following command to detect a new target device: 


echo 1 > /sys/class/fc_host/hostn/issue_lip 

Where hostn corresponds to the appropriate adapter. 

When a new Logical Unit is created on a particular target, a command such as the following is required to detect and configure it: 


echo "b t l" > /sys/class/scsi_host/hostn/scan 

Where b is the bus, t is the target, and l is the LUN to be scanned on hostn. 

Wild cards can also be used, such as the following example: 


echo "- - -" > /sys/class/scsi_host/host2/scan
Dirk Dierickx Expert in this area This member has accumulated 1000 or more points
Jun 15, 2007 02:17:56 GMT    Unassigned  

using RH4 with QLA cards, everything goes automatic. The robot itself is accessable through the generic scsi device file, while the tape drives are the standard /dev/rmt# types. 

like i said, in my case all is detected at boot (no need for extra drivers or tools), which can be checked easily with dmesg.
Graham Smith
Jun 21, 2007 02:14:48 GMT    N/A: Question Author  

Gentlemen, Thank you very much for your valuable input, it is greatly appreciated. 
I have scanned the scsi device and can see the changer and (i think) tape drive in /proc/scsi/scsi. All good. But still no /dev/st0 or /dev/rmt# or /dev/changer. 
Do I need to reboot the server for these changes to take effect? 
Thanks in advance. 

Graham
Hemmetter This member has accumulated 500 or more points
Jun 21, 2007 02:36:01 GMT    Unassigned Attachement is 290644.sh 

Hi Graham,

check if "st" kernel module is loaded:

$ lsmod | grep -w st
# if not loaded (re)load it:
($ rmmod st )
$ modprobe st

What do you see in /proc/scsi/scsi?
(try attached script).

rgds
HGH
Graham Smith
Jun 21, 2007 02:41:12 GMT    N/A: Question Author  

[root@dcpbl01 ~]# lsmod | grep -w st
[root@dcpbl01 ~]# modprobe st
[root@dcpbl01 ~]# cat /proc/scsi/scsi
Attached devices:
Host: scsi0 Channel: 00 Id: 00 Lun: 00
Vendor: HP Model: HSV200 Rev: 6000
Type: RAID ANSI SCSI revision: 05
Host: scsi0 Channel: 00 Id: 00 Lun: 01
Vendor: HP Model: HSV200 Rev: 6000
Type: Direct-Access ANSI SCSI revision: 05
Host: scsi0 Channel: 00 Id: 00 Lun: 02
Vendor: HP Model: HSV200 Rev: 6000
Type: Direct-Access ANSI SCSI revision: 05
Host: scsi0 Channel: 00 Id: 01 Lun: 00
Vendor: HP Model: HSV200 Rev: 6000
Type: RAID ANSI SCSI revision: 05
Host: scsi1 Channel: 00 Id: 00 Lun: 00
Vendor: HP Model: HSV200 Rev: 6000
Type: RAID ANSI SCSI revision: 05
Host: scsi1 Channel: 00 Id: 02 Lun: 00
Vendor: HP Model: HSV200 Rev: 6000
Type: RAID ANSI SCSI revision: 05
Host: scsi0 Channel: 00 Id: 02 Lun: 02
Vendor: COMPAQ Model: MSL5000 Series Rev: 0518
Type: Medium Changer ANSI SCSI revision: 02
Host: scsi0 Channel: 00 Id: 02 Lun: 05
Vendor: HP Model: NS E1200-160 Rev: 568c
Type: RAID ANSI SCSI revision: 04
Hemmetter This member has accumulated 500 or more points
Jun 21, 2007 03:10:20 GMT    Unassigned  

Hi Graham,

does the script from my last post show a device file for your changer?

can you give the output of:
$ ls -l /sys/bus/scsi/devices/0:0:2:2/


rgds
HGH
Rob Leadbeater Expert in this area This member has accumulated 7500 or more points
Jun 22, 2007 04:38:16 GMT    Unassigned  

Hi Graham,

You might need to add the following line to /etc/modules.conf

options scsi_mod max_scsi_luns=128

and rebuild your kernel.

Take a look at the following RedHat article:

http://kbase.redhat.com/faq/FAQ_43_2706.shtm

Hope this helps,

Regards,

Rob
Graham Smith
Jun 24, 2007 19:47:52 GMT    N/A: Question Author  

/sys/bus/scsi/devices/0:0:2:2 
[root@dcpbl01 0:0:2:2]# ls -l /sys/bus/scsi/devices/0:0:2:2 
lrwxrwxrwx 1 root root 0 Jun 25 09:44 /sys/bus/scsi/devices/0:0:2:2 -> ../../../devices/pci0000:00/0000:00:02.0/0000:03:00.0/0000:04:03.0/host0/target0:0:2/0
:0:2:2
Hemmetter This member has accumulated 500 or more points
Jun 25, 2007 03:44:43 GMT    Unassigned  

Hi Graham,

can you give the output of:
$ ls -l /sys/bus/scsi/devices/0:0:2:2/*
With "*" (sorry it was lost in copy'n paste)

Have you tried my script? Does it show some tapes?

rgds
HGH
Graham Smith
Jun 25, 2007 20:53:47 GMT    N/A: Question Author  

[root@dcpbl01 0:0:2:2]# ls -l /sys/bus/scsi/devices/0:0:2:2/* 
--w------- 1 root root 4096 Jun 26 10:52 /sys/bus/scsi/devices/0:0:2:2/delete 
-rw-r--r-- 1 root root 4096 Jun 26 10:52 /sys/bus/scsi/devices/0:0:2:2/detach_state 
-r--r--r-- 1 root root 4096 Jun 26 10:52 /sys/bus/scsi/devices/0:0:2:2/device_blocked 
-rw-r--r-- 1 root root 4096 Jun 26 10:52 /sys/bus/scsi/devices/0:0:2:2/dump 
lrwxrwxrwx 1 root root 0 Jun 26 10:52 /sys/bus/scsi/devices/0:0:2:2/generic -> ../../../../../../../../class/scsi_generic/sg6 
-r--r--r-- 1 root root 4096 Jun 26 10:52 /sys/bus/scsi/devices/0:0:2:2/model 
-rw-r--r-- 1 root root 4096 Jun 26 10:52 /sys/bus/scsi/devices/0:0:2:2/queue_depth 
--w------- 1 root root 4096 Jun 26 10:52 /sys/bus/scsi/devices/0:0:2:2/rescan 
-r--r--r-- 1 root root 4096 Jun 26 10:52 /sys/bus/scsi/devices/0:0:2:2/rev 
-r--r--r-- 1 root root 4096 Jun 26 10:52 /sys/bus/scsi/devices/0:0:2:2/scsi_level 
-rw-r--r-- 1 root root 4096 Jun 26 10:52 /sys/bus/scsi/devices/0:0:2:2/state 
-rw-r--r-- 1 root root 4096 Jun 26 10:52 /sys/bus/scsi/devices/0:0:2:2/timeout 
-r--r--r-- 1 root root 4096 Jun 26 10:52 /sys/bus/scsi/devices/0:0:2:2/type 
-r--r--r-- 1 root root 4096 Jun 26 10:52 /sys/bus/scsi/devices/0:0:2:2/vendor 

/sys/bus/scsi/devices/0:0:2:2/power: 
total 0 
-rw-r--r-- 1 root root 4096 Jun 26 10:52 state 

KEELIN_HART_2
Level 4
Hello. Back to my original issue! When I removed the IBM driver, the /dev/st and /dev/nst files created. Should I not be using the IBM driver? Won't this have a negative I/O effect??

Anuj_2
Level 3
Partner
Hi Keelin,

           I am facing almost same issue and i guess I am newer to Linux than you. How you removed IBM Tape drive from Linux. I have 2.6 kernal.