Unsolved
This post is more than 5 years old
28 Posts
0
2168
Networker storage node on Redhat linux and lin_tape
Hi
Anyone have experience with IBM's lin_tape software, Redhat Linux 6.4 x64 and Networker 8.0.1.3 when it comes to automatically detect tape devices. We have a configuration with an IBM3584 and IBM LTO6 driver connected via qlogic 8Gbit fc cards.
Networker will detect tapedrive when unsing standard Redhat tape drive device drivers but then these messages creep up "failed to read 262144 byte block with 32768 byte transfer" so I decided to give the lin_tape a try.
The tape drives get detected in the OS and have also enabled the persistend binding option by using an udev configuration, but the Networker storage node does not detect the drives. Have not had any luck using jbconfig either. Did try to do jbedit also but no luck.
Do any have an idea here?
rgds
Carlos
ble1
14.3K Posts
0
October 31st, 2013 05:00
If OS is detecting drives, just configure them with jbconfig. You will be asked to enter the path and then you should use new path to the device.
kgursli
28 Posts
0
November 7th, 2013 04:00
Well, you should think that would work. But no it does not. When I try to configure the library through jbconfig it dumps core. It seams like the lin_tape device driver wont work on the Networker linux combination.
I'll think I need to open a case with EMC support.
rgds
Karl Dag
ble1
14.3K Posts
1
November 8th, 2013 15:00
With jbconfig, IBM drivers (and so lin_tape I assume) had to be configured not to attach to changer (as NW uses SJI to control robotic). For tape, at least in my case where I had few VTL drives branded as IBM LTO3 (via DD), I uses standard st drivers and it worked fine. Reason why I didn't use lin_tape was because after I downloaded it from IBM ftp they just failed to install on my RHEL 5.5. As all worked fine with st, I never looked back again.
quyennm
1 Message
0
March 19th, 2014 20:00
With this error "failed to read 262144 byte block with 32768 byte transfer" please check Device block sizes chapter in "Configuring Tape Devices for EMC NetWorker Technical Note" guide. You can google search to get this guide.
Example 2:
- Host A uses block size of 64KB and writes the data to tape.
- Host B uses block size of 32KB and is later used to read the data from tape.
In this case, the restore fails because NetWorker is not able to read the original data blocks. An error message such as "Failed to read 65536 byte block with 32768 byte transfer" is displayed.
To check the volume block size for the currently mounted volume, read the value directly from NetWorker server.
Note that forcing larger block sizes has positive performance impact on high-speed tape drives, but it can also waste space on the tape.
Testing has shown the optimal values for LTO 1-3 to be 128KB, and for LTO-4 to be 256KB.
Once a volume is loaded in a tape drive, the current volume block size as well as desired block size is noted in NetWorker for reference:
nsradmin> show name; volume block size; device block size
nsradmin> print type: nsr device
name: /dev/rmt/0cbn
volume block size: 64 KB;
device block size: handler default;
This example shows that the volume was labeled with 64 KB block size, and the handler default value indicates that NetWorker does not attempt to force the block size during label operations but leave it as the default value.