ADSM-L

Re: [ADSM-L] UN-mixing LTO-4 and LTO-5

2012-06-13 11:20:37
Subject: Re: [ADSM-L] UN-mixing LTO-4 and LTO-5
From: Roger Deschner <rogerd AT UIC DOT EDU>
To: ADSM-L AT VM.MARIST DOT EDU
Date: Wed, 13 Jun 2012 10:15:12 -0500
Thanks Remco and Wanda! I am proceeding to define two logical libraries.

We do have "nicely arranged" volsers. However, with the TSM STK SL500
library driver we do not see the L4 or L5 suffix. We just see XXX999. So
I'm going to change the XXX part to be distinctive between LTO4 and LTO5
just to reduce the chances that _I_ will make a mistake during checkin.
At least the cartridges are different colors.

After several false starts integrating LTO5 (which was only physically
installed a week ago), I'm starting over for about the 3rd time, towards
making LTO5 work in our operation in a rational way. It looks like two
logical libraries is indeed the way to go. That way, both Library Client
servers can actually access all tapes, unlike some of my other ideas
such as taking paths offline and online under control of some
likely-to-misbehave daemon.

The fact that this is a Library Manager setup has not added technical
hurdles, though it has definitely made it more complicated.

BTW, mounting LTO4 tapes in LTO5 drives is supposed to work, and most of
the time it does. However, we have experienced a much higher I/O error
rate when LTO4 tapes are mounted in LTO5 drives, compared to mounting
the same LTO4 tapes in LTO4 drives. (All drives are HP.) This is another
reason I want to prevent it, in addition to keeping the LTO5 drives free
for LTO5 tape operations.

I came into this upgrade assuming everything would just work, according
to the TSM documentation, STK documentation, and the LTO consortium
website. It doesn't.

This list remains a priceless resource, to which I try to give back.

Roger Deschner      University of Illinois at Chicago     rogerd AT uic DOT edu


On Wed, 13 Jun 2012, Prather, Wanda wrote:

>Hi Roger,
>I've done this many times (creating 2 logical libraries in 1 physical) to 
>separate LTO types at various customers.
>I don't know anything dangerous about it.  (Granted, I've never tried to have 
>the 2 logical libraries run a CHECKIN at EXACTLY the same time, but I don't 
>see that as being a likely occurrence.)
>
>If you do a search=yes (or search =bulk) with no other parms, then yes indeed, 
>the checkin for that logical library will grab all the tapes that are 
>available for checkin.
>
>However, if you have nicely arranged numeric volsers (see the manual for 
>requirements, alpha prefix and alpha suffix are allowed), you can use VOLRANGE 
>and automate everything.
>
>For example, you can throw a mixture of carts into the bulk I/O door, then run
>
>CHECKIN LIBV SCSILIB4 SEARCH=BULK CHECKLABEL=BARCODE VOLRANGE=XXX000L4,XXX999L4
>CHECKIN LIBV SCSILIB5 SEARCH=BULK CHECKLABEL=BARCODE VOLRANGE=YYY000L5,YYY999L5
>
>The first checkin will pick up all the XXXnnnL4's, and the second checkin will 
>pick up all the YYYnnnL5's.
>
>If your volsers are all alpha, you'll probably need to have your operators put 
>the L4's in the I/O door and run that checkin, then put the L5's in the I/O 
>door and run the L5 checkin.
>
>You won't run into any worse situation than what you have today, with both 
>types of carts mixed in the same logical library.  The worst thing that's 
>going to happen, is that you get the wrong tapes checked into the library, 
>which means TSM can try to mount LTO5 carts in the LTO4 drives, and you'll get 
>an I/O error when it tries to read or write.    If you have LTO4 and LTO5 
>carts in the same library today, and you run an audit library with 
>checklabel=yes, I think you are subject to the same problem, unless you take 
>the LTO4 drives offline.  With separated logical libraries, you won't have 
>those issues.
>
>You can't get any data damage, TSM still won't let you overwrite any data it 
>shouldn't, even if the tape is checked into the wrong library, as both 
>libraries and all data volumes belong to the same TSM DB.  So it's a totally 
>harmless thing to try, as far as I know.
>
>It's really no different than having 2 separate physical libraries attached to 
>a TSM server; somebody can still throw the wrong cartridges into the I/O door!
>
>I'm not sure if that was the answer you were looking for, if you have other 
>specific questions or situations, feel free to mail back -
>Wanda
>
>
>
>-----Original Message-----
>From: ADSM: Dist Stor Manager [mailto:ADSM-L AT VM.MARIST DOT EDU] On Behalf 
>Of Roger Deschner
>Sent: Tuesday, June 12, 2012 6:27 PM
>To: ADSM-L AT VM.MARIST DOT EDU
>Subject: Re: [ADSM-L] UN-mixing LTO-4 and LTO-5
>
>Defining separate libraries had occurred to me, but it seems like it could 
>also be dangerous, if both logical libraries were to try to check in the same 
>volume using a search. How do I avoid that? That is, if I create two 
>libraries, one of them with the LTO-4 drives and the other with the LTO-5 
>drives, both libraries will have access to all the tape slots. Does this mean 
>I cannot use CHECKIN LIBVOL SEARCH=YES but instead I should check in new 
>volumes by name? Would this be similar to sharing a physical library with 
>another application, except that the other application happens to be the same 
>TSM server?
>
>This configuration of two logical library definitions for one physical library 
>would appear to have some risks. Or is it safe? Are there any other problems I 
>should expect with this kind of configuration?
>
>Roger Deschner      University of Illinois at Chicago     rogerd AT uic DOT edu
>
>
>
>On Tue, 12 Jun 2012, Remco Post wrote:
>
>>did you define separate libraries for those device classes?
>>
>>On 12 jun. 2012, at 08:15, Remco Post wrote:
>>
>>> check the format= option on your device class...
>>>
>>> On 12 jun. 2012, at 08:02, Roger Deschner wrote:
>>>
>>>> We have a tape library (Oracle/Sun/STK SL500) which contains both
>>>> LTO-4 and LTO-5 drives, and both LTO-4 and LTO-5 media. I am trying
>>>> to keep TSM from mounting an LTO-4 cartridge in an LTO-5 drive, but
>>>> it is insisting on doing it anyway.
>>>>
>>>> We have 4 LTO-4 drives and 3 LTO-5 drives. The mount limits for the
>>>> two devclasses are set accordingly - to 4 for the LTO-4 devclass and
>>>> 3 for the LTO-5 devclass. When a request to mount an LTO-4 cartridge
>>>> comes, it seems to use any of the 7 drives, regardless of whether it
>>>> is an LTO-4 or LTO-5 drive. Therefore some tape mounts for LTO-5
>>>> cartridges are failing or being delayed due to there being no
>>>> available LTO-5 drives when some of them are occupied by LTO-4
>>>> tapes. This is despite the claim in the section of the TSM
>>>> Administrator's Guide for AIX servers titled "Mount limits in LTO
>>>> mixed-media environments" (on book page 221 / physical page 249 in
>>>> TSM V6.2 for AIX, or book page 198 / physical page
>>>> 232 at V6.3) that setting the mountlimit to the actual number of
>>>> earlier-generation drives will prevent the use of later-generation
>>>> drives for the earlier-generation devclass.
>>>>
>>>> BTW, this is a library manager configuration, which may complicate
>>>> things. The devclass definititions do match on the library manager
>>>> and both of its library clients. The library manager is at 6.2.2.30,
>>>> and its two clients are at 5.5.6.0 and 6.2.2.30.
>>>>
>>>> So, the question is, how do I prevent LTO-4 cartridges from being
>>>> mounted in LTO-5 drives? I would prefer not to use the hardware
>>>> library partitioning feature, which has its own set of hassles.
>>>>
>>>> Roger Deschner      University of Illinois at Chicago     rogerd AT uic 
>>>> DOT edu
>>>> ==== "NO OVERNIGHT CAMPING AUTOMATIC SPRINKLER SYSTEM IN OPERATION"
>>>> ==== ==== --sign, I-70 rest area, Parachute, Colorado
>>>> =======================
>>>
>>> --
>>> Met vriendelijke groeten/Kind Regards,
>>>
>>> Remco Post
>>> r.post AT plcs DOT nl
>>> +31 6 248 21 622
>>
>>--
>>Met vriendelijke groeten/Kind Regards,
>>
>>Remco Post
>>r.post AT plcs DOT nl
>>+31 6 248 21 622
>>
>