Start a Conversation

Unsolved

This post is more than 5 years old

1069

March 11th, 2009 14:00

CX600 - reconfiguration of my DAE's

Hi all,

I just inherited the job of managing an old CLARiiON CX600.
It has 11 shelves, (6 DAE, and 5 DAE2-ATA).

I don't believe it was set up correctly, as far as the way the shelves are connected (This was not done by EMC techs).

Both bus 0 and 1 have a mixtures of SATA and Fiber shelves.

I would like to reconfigure it so that Bus 0 has all DAE's (FC drives) and Bus 1 has all SATA.

We have several metaluns that live on both buses. In an attempt to preserve the data already on the SAN, I've migrated the LUNS to shelves that will not change when we re-cable everything

So I plan to shut down the CX600 and re-cable the SAN so that Bus0 is all FC shelves, and bus 1 is all Sata shelves. Many of the shelves will have their SCSI ID changed when it's powered back on. So my question is, is the CX600 smart enough to adapt and realize that a shelf has been moved to a different bus and/or SCSI ID?, or will any LUNS that reside on a shelf that changes bus or SCSI ID now be lost?

I plan to delete all luns on shelves that will be changed, but if i don't have to, then I wont!
Everything will be backed up in advance.

Thanks in advance!

--ZANGIN

261 Posts

March 12th, 2009 09:00

No Clariion arrays will be able to follow where an enclosure moved to. If you do move it somewhere else, then it'll fault the array and only moving it back will clear that fault.

With that said you have to get the array to forget an enclosure entirely (through unbinding everything in there and removing all raid groups, removing cables, and then rebooting the array) (don't remove cables while the Clariion is on of course). After the array comes up and does not see the enclosures anymore (clean with no faults), you can add them where needed. Definitely follow the procedure generator for adding them back. (We don't have a customer procedure for removing DAEs so an EMC tech is suggested)

http://powerlink.emc.com/km/live1/en_US/Offering_Technical/Software_Download/Customer_CLARiiONproc.exe

2.1K Posts

March 12th, 2009 09:00

Welcome to the Forums Zangin.

I'm not sure what your past experience level is with managing EMC equipment, or how familiar you are with Powerlink and such, but I thought I would bring something to your immediate attention in case you haven't found out yet.

Your CX600 will reach End of Support Life (EOSL) at the end of September this year. At that point your support options will be limited to technicians working on a best effort basis and charging Time & Materials for work. You may want to consider replacing the array this year and migrating all your data to a new array that is configured from the factory exactly how you would like.

We are facing this issue right now with a CX600 and a pair of CX400s in our environment and will be replacing them before this date so we don't have to rely on "best effort" service for hardware servicing critical business applications.

If you are not worried about the service, and can take a long enough outage to back up all your data and manually reconfigure your array, than the previous suggestion is probably your best bet.

2.1K Posts

March 12th, 2009 09:00

I was just looking at this again as something was kind of bugging me about it.

I guess I would like to know what the driver is for reconfiguring the DAEs? Are you concerned about performance? Are you having problems? Have you read an article that suggested not mixing DAE types on a bus?

We only have a few ATA DAEs in our environment, but all of them share buses with FC DAEs and we have never experienced any issues related to this configuration.

You may be better off saving your efforts for the migration efforts to get the data to a new array when someone agrees to pay for it.

4.5K Posts

March 12th, 2009 10:00

Another point to consider is that when you are doing all the reconfigurations, you should upgrade the flare to the latest release. The CX600 can up upgraded to release 19 patch 45 - this is the last update for the CX600 (and 400/200 also).

Once you have all the DAE's moved around, you should then review the Best Practices guide for re-configuring the LUNs and Raid Groups. See the document on PowerLink:


EMC CLARiiON Best Practices for Fibre Channel Storage: FLARE Release 26 Firmware Update - Best Practices Planning

http://powerlink.emc.com/km/live1/en_US/Offering_Technical/White_Paper/H2358_clariion_best_prac_fibre_chnl_wp_ldv.pdf

This is for version 26 code but it does still apply for the configurations of Raid Groups and LUNs. It is missing the information about the older 5400 RPM ATA disks and what you can and can not use them for. If you have an older version of this document, release 24 or release 22, those contain the data on ATA disks.

glen

March 12th, 2009 11:00

woops! I was going to give everyone a Helpful but it only lets me give out 2 helpfuls and 1 correct

March 12th, 2009 11:00

Thanks everyone for the great responses.

The reason we were wanting to re-cable everything is because I was told by an EMC tech that you should not mix ATA and FC on the same bus, because it causes performance issues... I presume that we're not getting the most out of our FC drives because of the way it's configured. Is this correct?

I'll be happy to leave it as is and not re-cable it if changing things around is not going to improve my performance.

I'll definitely be following the best practices and use the procedure generator.

I'm currently running FLARE 02.19.600.5.044

We did have a problem after adding an extra FC shelf to a bus that had 2 DAEs, 2 DAE2-ATA already on it. I added the 5th shelf to bus 1, and then started plugging in drives. As soon as I did, 2 the two ATA shelves on the same bus had drive failures on just about every drive. I verified that the SCSI ID on the added shelf was correct. In any case, I turned off the newly added shelf, and did not lose any data. I was able to correct the faults by removing and replacing one disk at a time, waiting for the luns to re-normalize each time. It turned out that the 'newly added' FC shelf part # did not match our other FC shelves and perhaps the shelf I used was for a Symmetrix SAN, instead of a clariion. In any case, I scrounged up another FC shelf that had the same part # as our other FC shelves and have put it back into the mix. I'm a bit afraid to add drives to it again, however I'm pretty sure the cause of that problem was due to the using the wrong shelf.

2.1K Posts

March 12th, 2009 12:00

It is possible that the new DAE you put in was a DAE4P - That would cause problems definately. I can almost guarantee it wasn't a Symmetrix DAE, unless the really old ones looked way different. Any current one wouldn't be possible to mistake for a CLARiiON DAE.

I would just reiterate that we have a CX700 and a CX500 running with mixed FC & ATA busses and we have never had a problem (performance or otherwise). Unless you can identify an actual problem I would leave the configuration alone. Save yourself the headaches!!!

261 Posts

March 12th, 2009 12:00

The maximum speed a backend bus can run at on the CX600 is 2Gb. If the buses are at this speed, then unless the off chance that you are maxing out the bus bandwidth wise (which I personally have never seen), then moving anything around probably won't help with performance.

To check the speed of the backend you can expand "physical" in Navisphere and at the end it tells the speed.

261 Posts

March 12th, 2009 13:00

DAE4P (Per using my procedure generator) says you can only install DAE2 or DAE2-ATA Enclosures onto the CX600. If it was a DAE4P (DAE3P) I bet it would show up as faulted. Even if it did work, you would never get the 4Gb potential out of this DAE no matter what you did to the backend of the CX600.

March 12th, 2009 14:00

The part # of the shelf that caused the problems was 005-0475-69 .

The part# of all my other DAE's (FC) is 005-3484-89. It is a valid DAE for CLARiiON CX200, CX400, CX600, CX300, CX500, CX700.

PN 005-0475-69 does NOT appear in the Xref Listing so perhaps that was the culprit of my problems. Anyone know what this is?

March 12th, 2009 14:00

Ya.. i wasn't trying to get 4gb out of it. I was just using a shelf we had lying around.

Anyhow I've plugged in the correct shelf and it's working fine. Now to just determine if it's worth changing all of this around, or not to bother.

March 13th, 2009 08:00

I don't see where you're talking about...(I expanded the "physical" and then all it shows is the DAE's and the Enclosure SPE)

March 13th, 2009 08:00

When unbinding raid groups, etc, do you have to unbind ALL raid groups? Even those whose shelf is not going to change??... or just the raid groups,etc, that live on the shelves that are going to be moved.?

261 Posts

March 13th, 2009 08:00

Going back to the speed question, in later codes to the right of the enclosures it says the speed of the bus, but right click and enclosure and do properties on it and it should say the speed.


As for the unbind question. You only need to do clean up in DAEs that are going to move. For the ones in which you have luns and the bus and enclosure numbers are not going to change, no unbinding is needed (but you cannot change the enclosure number and expect to see the data.)

Hope this helps.

-Ryan

4.5K Posts

March 13th, 2009 09:00

be careful that you do not have any raid groups that span different enclosures

- if you are going to move a DAE to a different bus or enclosure number, you must unbind all LUNs and destroy all raid groups in the DAE then follow Ryan's directions for removing the DAE from Navisphere.

glen
No Events found!

Top