Unsolved

This post is more than 5 years old

11 Posts

3134

February 10th, 2010 02:00

Problem with Expanding MetaLUN

I have a problem with trying to expand a MetaLUN and wondering if anyone could advise where i may be going wrong or, if they have seen a similar issue before.

We have a MetaLUN that contains 14no 100Gb LUNs that are in various Raid 5 Raid Groups all around the original 4 disk shelves (This is how it was when the array arrived in our data centre for us to host). This MLUN was created with Striped Expansion. What i am trying to do is expand this MLUN by adding a new 800GB LUN that has been created on a new disk shelf but we want to Concatenate this LUN to the existing MLUN.

When we try this we get this error;

Command: Set LUN; Exception: A configuration change has failed. (Internal Software Error)(4, 4, K10 Exception Completing K10AggDrvAdmin 60060160047d1700:8600e3ff5915df11: ); File: catmerge\layered\Generic\user\GldAdmin\GldAdmin.cpp; Line: 00000447.   71

I have logged a call with Dell (This is a DELL/EMC Clariion CX500) who have come back saying they do not recommend doing a Concatenated Expansion to a Striped MLUN which seems kind of strange to me.

So my questions if anyone would be kind enough to offer advice are;

1) Doing a Concatenated Expansions to an existing Striped MLUN is an acceptable practice isnt it? As i am sure i read somewhere in EMC whitepapers that it is.

2) Does anyone have any ideas why we may get such an error as above when trying to do this MLUN expansion? Or any pointers of things to check as there is not data on thie MLUN so i can destroy it and all the underlying LUNS and start again if needed.

3) This MLUN is for a Windows Server so currently has an Alignment Offset of 63 set on the MLUN. Would that make any difference? Should the offset be set at the LUN level?

Many Thanks and any assistance will be greatly appreciatted.

Andy

6 Operator

 • 

5.7K Posts

February 10th, 2010 03:00

Just a rough thought (not a 100% answer, just 66.7%):

1) You can indeed concat a new LUN to an already striped metalun.

2) There is a maximum number of components in a metalun, but I don't think it's 14, but I'm not sure.

3) You shouldn't be using LUN offset. If you want to align a partition, you should do it on the OS level. And remember that this is destructive, so if there's data on the LUN, backup the data, align the partition (create a new one) and restore the data back to it.

6 Operator

 • 

4.5K Posts

February 10th, 2010 07:00

A metaLUN can have up to 16 Components - the first Component is Component 0 - this component contains the original LUN. Can you list the number of LUNs in Component 0 - these are all stripped correct?

You should be able to concatenate expand the new LUN - do you see the new LUN listed in the list of possible LUNs in the Wizard?

What version of Flare are you running on the array?

glen

11 Posts

February 10th, 2010 08:00

Glen,

We have 02.19.500.5.027 Code on this array..

When  i look at the MLUN, listed is Component 0 and under that are the 14  original LUNS. This was created using the Striped Expansion mode  originally. When i right click on the MLUN to expand it, pick  Concatenate Expansion the LUN i would like to add does appear in the  list of availble LUNS. If i select it, then go past the next screen that  asks about the size of this MLUN and click Finish, then we get the  error shown in my original posting.

Andy

MLUN Properties.png

6 Operator

 • 

4.5K Posts

February 10th, 2010 13:00

Andy,

You can concatenate a new LUN to the existing metaLUN - this is supported. Don't worry about the alignment on the new LUN, it takes it from the original Component 0.

The error is something that DELL needs to fix - it may be that the version of Flare on the CX500 has a bug - the current Flare for CX500 is release 26 patch 28.

What is the size of the component LUNs in Component 0? There use to be an upper limit of the maximum size of a LUN - but I can't recall if that was present in flare 19 or if that was an earlier limitation.

glen

11 Posts

February 11th, 2010 00:00

Glen,

The LUNS in Component 0 are only 100Gb each. So currently MLUN1 is 1400Gb.

I have to admit im a little frustrated that Dell did not mention anything about the Flare code. So thank you for pointing that out, i shall go back to our client and mention this, as at the start of 2010 Dell came to our site to install a new DAE on this array for the client and it was mentioned about doing a Flare code update but Dell said it was not required.

Andy

2.2K Posts

February 11th, 2010 07:00

Regardless of whether the FLARE revision is related to the metaLUN expansion I would still work with Dell to get that FLARE code upgraded. It is very old and there have been lot's of fixes and improvements to the FLARE code since then.

6 Operator

 • 

4.5K Posts

February 11th, 2010 07:00

I still need to verify that upper limit on LUN size for flare 19 - I can't put my hands on the old docs right this minute - hold off until I've had a chance to confirm this - it may be that it was an older flare that I was thinking of,

glen

11 Posts

May 13th, 2010 09:00

We got the solution to this. Dell told us that the problem was to do with the Flare code release we had.....

6 Operator

 • 

4.5K Posts

May 13th, 2010 11:00

Did they tell you what the flare code version that fixed it? What was the case number?

glen

No Events found!

Top