LTO TAPE ERRORS - ALL THE TIME 1

[glaserc]

HI,
Does anyone have any input on LTO GEN2 tape drives and why I am getting media open,write and position errors? I know that they are less stable than say 9840s, but this is getting ridiculous!

We have a 3 media managers running the LTOs, two Solaris and one AIX.
Has anyone experienced errors with LTOs if the server is highly utilized with other apps?
Any info/helpful hints would be great...

 

[MaineDCAdmin]

We have a Dedicated Master/Media Manager configuration, with 12 LTO-II tapes drives, and have been at 3 9's for better than 6 months. We have followed the guideline of making the tape drive the slowest component in a data stream where possible and not had a problem. If NetBackup can not obtain the system resources on the Media Manager, you may have a problem. While running 12 tapes drives concurrently we routinely see 50% utilization or better on our 8-way Sun v880. Take a look at your system components to see if memoory, I/O, or CPU resources are insufficient on the Media Managers.

 

[glaserc]

looking at the performance graphs from last night, I see a very high CPU utilization for the media manager that is giving us headaches.
How does your sar output look for the 880? When you say dedicated, do you mean that all it does is NBU? ie - no other apps run on it?

 

[creakyjoe]

~glaserc, what version are you using ?. we'd been plagued with write & position errors for 3yrs (4.5FP6). I recently went FP8 as FP7 was s'posed to fix MPX position errors. we've had a couple position errors since, but since freezing a couple tapes it's been quiet for couple weeks.
But now i've been asked to go 5.1 (we're missing out 5.0 as keep reading it was rushed). Version 6 will be too new for us - we're not Veritas bugtesters..
Veritas and their bugs and enforced upgrades - i don't know!!

Rich

 

[glaserc]

we recently went to 5.1 MP2 and we are still having the errors. by the way, we had huge headaches going to 5.1 so beware. We had issues with our AIX master losing connectivity to our roboitic libraries and therefore putting our drives into AVR control. bad news.
I have a call open to veritas about this, but my guess is that they will point the finger at the drives or the OS like always

 

[glaserc]

Veritas has stated that the LTO issue is coming from the OS level...huge surprise. I am going to contact IBM to see what they say about the issue with the LTO tape drives

 

[glaserc]

IBM had me add the Atape driver for the IBM tapes drives. Unsure of what this will do, but I guess it gives IBM better debugging info

 

[creakyjoe]

@glaserc, maybe that's ibm speak for kernel-specific LTO2 support. We have HP LTO1's, under Solaris, & in the early days (3yrs ago) we had loadsa problems. Part of the (ongoing, numerous) fixes was to edit the /kernel/dev/st.conf files and add HP-specific params instead of generic params. It didn't really help us as our problems went on and on, intermittently, until about 3weeks ago. Funny that, as that's when we went FP8 !!..

not sure this is the same thing you're talking about (probably isn't), but i'm not familiar with ibm stuff..

Rich