The new forums will be named Coin Return (based on the most recent vote)! You can check on the status and timeline of the transition to the new forums here.
The Guiding Principles and New Rules document is now in effect.
RAID Controller Migration
Apothe0sis
Have you ever questioned the nature of your reality?Registered User regular
Have you ever questioned the nature of your reality?Registered User regular
Hi All,
Let's talk RAID Controllers. If I have a RAID controller which is currently at capacity and want to install a controller with a bigger capacity. My concerns are as follows:
Is there anything controller specific with regard to how the RAID volumes are created? We're specifically talking about RAID1 in this case.
Can you simply unplug them from one controller and plug them into another? If I take one of the disks within a RAID1 volume and try to access it like a regular disk, can that be done? Will it RUIN EVERYTHING? Are there any guidelines here?
Let's talk RAID Controllers. If I have a RAID controller which is currently at capacity and want to install a controller with a bigger capacity. My concerns are as follows:
Is there anything controller specific with regard to how the RAID volumes are created? We're specifically talking about RAID1 in this case.
Can you simply unplug them from one controller and plug them into another? If I take one of the disks within a RAID1 volume and try to access it like a regular disk, can that be done? Will it RUIN EVERYTHING? Are there any guidelines here?
0
Posts
I'm not sure if something similar is utilized in mirroring setups. My assumption would be no, as I can't think of a reason RAID 1 would require the same kind of meta-data. As long as there is no striping (that is, you're not using RAID 1+0) you should be able to access the contents of any single drive without issue if the OS you're using to access it supports the filesystem on the disk.
Aside from the 1st paragraph, I'm not 100% on the stuff above, so please don't take just my word for it. All that said, you might consider going with a software RAID in the future. Most controllers have a JBOD mode you can use that will pass through instructions from the software-based RAID, and if you're using a relatively modern processor the load is really quite minimal for a small array. Also, in my experience they migrate much more easily.
Battle.net
We are talking about dell cards, moving from a perc200 or something to an h700.
Unfortunately it isn't a Linux or windows box, but a VMware box which is running with local storage - the best solution is to get a SAN, but that is not on the cards right now.
However, I am pretty confident that you can pull drive from a RAID1 configuration and use it outside of that array as a typical disk. Afterall, it's just a carbon copy of the other disk in the array. As long as it is not placed within another RAID array, it should function normally. Good news is that it is easy to test. You can easily just add a newly formatted disk into the original RAID1 and rebuild the array, then test the disk you removed without fear of losing what was on the original array.
I believe that Alecthar is correct here. One of the problems with RAID is that you really cannot change the properties. If you do wish to change how the RAID operates, it requires completely formatted drives, much like how multiple partitions on a single drive behave. However, I have had success in the past of cloning a drive onto a newly formed (and larger) array. If you're talking terabytes, then cloning may not be very viable of a solution.
From here
My experience with Dell raid is limited to the perc4/6 series running raid5, but I know I've had to move an entire array before and used the import foreign config quite successfully (though that's between two identical controllers).
EDIT: Just had a thought, if you want to be on the safe side, start with one of the drives, import foreign config from that, boot up and verify it's all good, then pop in the second drive and let it rebuild like normal.
Migration did not go according to plan, the server refused to boot once the card was installed. As it turns out the card we were told to order was not complete in its information - we got an h700 internal adaptor, we needed he h700 integrated.
Danced around with dell pro support for a few hours, they thought it should work. It didn't. Upgraded the bios etc... Still no dice.
At 2.30AM someone new came on shift, declared the card incompatible, and gave us the "correct" part number. Parts and accessories claim that part number doesn't exist. 24 hours later we get a quote for another part, with the exact same description as the card we got, but half the NVRAM for $200 more. The next gen card for the next gen of PE servers which is specified as an integrated card costs less than half that. Still awaiting a response as to what is going on.
Plus, of course, in classic dell fashion, no part numbers are mentioned on the quote, so who knows what's going on there.
These people are maniacs.
Of course, we buy enough computers and stuff for them to care, so that may be an issue.
Dell support used to be amazing, but then they started getting extremely picky about how to handle support call and replacements. @Bigity may be right. A sales rep would be able to weed through all of their internal contacts to figure out what you need instead of relying on support to bounce around ideas off each other. If that fails, you speak to their supervisor. That's one of the only good things about sales people: if they fuck up, their boss is quickly notified and you get your service... most of the time.
He's totally pants.
It is my general experience that things go smoothly when we are buying new machines.
They go to hell when we want to buy parts.
Software OS type raid only requires compatible software when a problem occurs, which is a much simpler problem to solve.
Something which is "pants" is something which is not very good.
Something totally pants, is something utterly useless!