Logical disk inconsistent

[ascheale]

Hi there,

yesterday we wanted to update from 3.1.2 MU2 to 3.1.3 MU1.
After upgrade check our SP pointed out this:

Logical Disk tp-1-sd-0.89 is degraded Currently: mag as Created: cage
Logical Disk tp-1-sd-0.90 is degraded Currently: mag as Created: cage
Logical Disk tp-1-sd-2.0 is degraded Currently: mag as Created: cage
Logical Disk tp-1-sd-2.1 is degraded Currently: mag as Created: cage
Logical Disk tp-7-sd-2.0 is degraded Currently: mag as Created: cage
Logical Disk tp-7-sd-2.1 is degraded Currently: mag as Created: cage
Logical Disk tp-7-sd-2.2 is degraded Currently: mag as Created: cage
Logical Disk tp-7-sd-2.3 is degraded Currently: mag as Created: cage
Logical Disk tp-1-sd-2.3 is degraded Currently: mag as Created: cage
Logical Disk tp-1-sd-2.2 is degraded Currently: mag as Created: cage
Logical Disk tp-1-sd-2.5 is degraded Currently: mag as Created: cage
Logical Disk tp-1-sd-2.4 is degraded Currently: mag as Created: cage
Logical Disk tp-1-sd-2.7 is degraded Currently: mag as Created: cage
Logical Disk tp-1-sd-2.6 is degraded Currently: mag as Created: cage
Logical Disk tp-1-sd-2.9 is degraded Currently: mag as Created: cage
Logical Disk tp-1-sd-2.8 is degraded Currently: mag as Created: cage
Logical Disk tp-1-sd-2.11 is degraded Currently: mag as Created: cage

What can I do to solve this ?

Thank you
Alex

[Davidkn]

I wonder if this is similar to the issue I saw recently when doing an upgrade. It wasn't to 3.1.3 as it was only a patch to support new drives.

But after I had performed the disk upgrade, my cpgs had changed from 3+1 to 4+1 and the availability had also changed.

So I would first look at the cpg and make sure it still has exactly the same settings it had before the upgrade.

[ascheale]

It was only the upgrade check - my CPGs are not changed.

How can I fix an inconsistent LD ?

Alex

[Davidkn]

I see, so firstly, I'm pretty sure the San is not customer upgradeable until you are on 3.1.3 mu1, so HP should be involved in this update unless you are a service partner?

So I would speak to hp and involve them in the upgrade, especially if this is a production system....

[zQUEz]

I have similar issues to you, but I can tell you that it is an informational message, not a critical issue. Basically, the literature says a tunesys *should* fix this, and I even had 1 HP 2nd level tech tell me as much. When I pointed out that I have already done a tunesys I got another tech respond that it won't fix it and I need to retune all the VV's associated with the LD's. What they don't point out is that because I have all my VV's wide striped across all LD's (whole point of 3PAR right?), that to do that would mean I would run out of tier1 (FC) long before I got my VV's retuned.

*beating head against wall*

I am on 3.1.2MU3 but the inconsistencies were from 3.1.1MU3. I "suspect" what happened in our case is someone changed the CPG settings, and then changed it back but in the intrim some VV data got written to the old settings.

[Schmoog]

I have similar issues to you, but I can tell you that it is an informational message, not a critical issue. Basically, the literature says a tunesys *should* fix this, and I even had 1 HP 2nd level tech tell me as much. When I pointed out that I have already done a tunesys I got another tech respond that it won't fix it and I need to retune all the VV's associated with the LD's. What they don't point out is that because I have all my VV's wide striped across all LD's (whole point of 3PAR right?), that to do that would mean I would run out of tier1 (FC) long before I got my VV's retuned.

*beating head against wall*

I am on 3.1.2MU3 but the inconsistencies were from 3.1.1MU3. I "suspect" what happened in our case is someone changed the CPG settings, and then changed it back but in the intrim some VV data got written to the old settings.

My understanding of the technogy is different.

Your data is wide striped across all physical disks. From the physical disks, 3par creates chunklets. 3par then uses chunklets to build logical disks according to the raid and availability specifications in your cpg.

Thus, your data is wide striped across all pd's, not ld's.

So what you would want to do is identify which vvs are using those ld's and tune them.

[ascheale]

Seems this is a bigger issue...

Tuneld does not work
Tunesys does not work
Compactcpg does not work.

Case is elevated to L2 at HP.

Yes, i am a HP Gold Partner and allowed to do updates

I will keep you up2date

Cheers
Alex

[Davidkn]

Interesting, keep us posted.

You need to be a Service One Partner to perform updates btw, and I still think you need to engage with HP 3PAR support to run the update as they give you the authority that it's ok to upgrade based on the customers environment.

I wouldn't do it on a live system without their assistance, anything could happen and I'd rather it be HP running the show if it did, I've had it before where it's not worked and the San went offline. Not a happy customer, but at least I wasn't responsible

[Schmoog]

Interesting indeed.

I always find that when stumped, it's best to back up and go back to the basics.

The error you've gotten suggests that you have a CPG which is set to cage availability, but is only able to get mag availability.

Can you give us a little more into the array? How many shelves/disks? What CPG's do you have created?

[runekruse]

Hi, a you have mentioned before tunesys does not correct this issue. What you should use is tuneld "ldname" one every affected LD. I would recommend not to do all ld's at the same time, but you coluld run tuneld on 1-5 at the same time. You can follow the progress with showtask.

Hope this helps.

/Rune