Thanks. Scanning for bad blocks does seem like a good idea. I don't have Tech Tool Pro, Drive Genius, or Disk Warrior; they're each about $100 but I'll happily buy whichever one of them is likely to be most useful here. Would that be DG? Mr. Hatter doubts that TTP will be as useful here, and Drive Warrior sounds like a higher-level recovery utility, operating at the file system level rather than at the level of blocks.
But it also sounds like I could try Seagate's SeaTools first to achieve the same thing, is that right?
And can all these utilties scan for bad blocks without also reformatting the drive at the same time? And on a live system?
I've considered replacing all the drives with newer ones but of course it's pricey to do that.
Let me ask this: my Netgear ReadyNAS Pro allows drives in the array to be replaced with higher-capacity drives, and the array will be expanded to take advantage of the additional space. I guess it's safe to assume that the Apple RAID Card has no such capability?
In which case, if I want to grow the array, do I need to replace all the drives with higher-capacity drives (2 TB or 3 TB), recreate the array using RAID Utility, and then restore my system from backup?
I've chosen Seagate drives in the past because their warranty replacement process is so straightforward; essentially if a drive is in warranty they'll just replace it without making you jump through hoops. Not sure if other manufacturers are as easygoing in that respect.
To scan is one thing but Seatools is linux or Windows program. Same for Lifeguard.
SoftRAID 4 is $139 and will run in the background and safe to use.
Apple RAID is a strange product. The way to get "from here to there" is to backup and clone your volumes (two sets minimum, maybe two different methods). And then rebuild and recreate.
I'd shoot off an email to SoftRAID and ask that you have RAID card, have some issues, and looking for more reliable and to insure your drives and array is 200% safe and functional. SoftRAID mirrors are threaded and stripe reads, meaning a 3 or 4-drive RAID has redundancy and performance too. And to migrate. Their RAID setup if very flexible. Read the features, dl the manual, there is a demo they've added support for SSD,
Seagate knew that their warranty would 'work' and they made their name with SCSI and firmware, before they took over Maxtor and started messing with what had worked. WD you just enter the SER NO into a web form and you're done and RMA process is very easy. I even had Hitachi do advance replacements.
SeaTools boots but doesn't seem to see any of the drives, and it also doesn't seem to like the Apple USB keyboard because it won't respond to any keypresses to get it past the welcome screen.
Drive Genius 3 can't see the individual drives either, only the RAID-5 volume. It lets you scan that volume for bad blocks but I'm not sure it's meaningful (won't the RAID controller hide the underlying details?) and even if it found bad blocks I'd have no way of knowing where they were.
I could replace the drives one by one but I would really just be guessing. Unless I wanted to change all the drives together (to increase the capacity or just to get newer, faster drives) it seems like I'd be needlessly changing three out of four of them.
Can SoftRAID really be preferable to, and perform better than, a hardware RAID solution? But I can ask them about it, sure (in particular how it compares to the Apple RAID Card). Can SoftRAID scan my individual drives for bad blocks even if they're not part of a SoftRAID volume?
Nothing can see the individual drives when part of a hardware RAID like yours.
And even SR will only work with software arrays, not hardware.
For all the protection of a drive failure of RAID5, I don't see it. And for hardware RAID find one that provides RAID6.
This may be something you don't need but it is helpful
- Disk failure has a medium impact on throughput
- Most complex controller design
- Difficult to rebuild in the event of a disk failure (as compared to RAID level 1)
- Individual block data transfer rate same as single disk
- File and Application servers
- Database servers
- Web, E-mail, and News servers
- Intranet servers
- Most versatile RAID level
Hangs became incessant so I pro-actively replaced the drive in Bay 3 with a refurb Seagate 1 TB. After several hard reboots and system panics, the system eventually came back up. RAID Utiiity reported that "Drive 3:xxxx" had failed even before I had replaced it, so I'm hopeful that this will turn out to be the issue, i.e. that this particular drive has just gone bad. I marked the "new" drive in Bay 3 as a spare and the array is rebuilding now.
Here's what has me scared, though: RAID Utility is showing TWO tasks, not one. The rebuild task on "Raid Set RS1" is running now (looks like it will take a long time).
But it's also showing this:
Target: Volume "Macintosh RAID-5" of RAID Set "RS1"
This makes it sound as though, as soon as the RAID array is rebuilt, it's going to be blown away (i.e. "initialized") by RAID Utility.
Is that possible? It seems unlikely that the OS would blow itself away like that, but I really don't feel secure about it.
Is there something I should do (apart from making sure that I have an up-to-date backup) to prevent this from happening? I don't see any way to cancel pending tasks, either in the Raid Utility GUI or in the command-line raidutil.
All four drives are now Seagate refurb 1 TBs. I've tentatively decided that, once the system has recovered, I'm just going to replace them all with new 2 TB drives (not sure from which manufacturer) and rebuild the array from a backup.
I didn't want to pour any more money into this system, preferring to spend it on a new system later this year (I'm still hopeful that Apple will release a new TB-enabled Mac Pro). But money spent on disk drives is not wasted since they can be used elsewhere--unlike money spent on more RAM for this system.
I would get the data on two drives, maybe 2TB WD Black or invest in one of the Hitachi 4TB models that are hot for performance (180MB/s on outer 1/2) make a great backup and investment.
I'd nuke the whole Apple RAID card and live w/o as I have said.
I wonder if you can boot from what I like to keey handy and use, a plain vanilla OS for emergency maintenance (leave your existing system alone) and see what it shows. Do you then see a queued process?
I have yet to find a use for TB other than laptops for displays and storage, nothing on workstation. Intel will have Tb on first PC later this fall.
What might be in the wings or not who knows but Intel and others, single socket Xeon E5 8-core but the high end today and tomorrow won't change all that much and it will take time to iron out (OS support always takes a point or two, chip revisions, and every model has had its EFI + SMC firmware update in the first months since 2008).
I think though back to the array, it is was due for a total delete and recreate from scratch rather than just rebuild after rebuild.
Others have wished they could use the Apple RAID5 with 3rd party but that is impossible to do, the data has to be migrated to new hardware. And software RAID would be safer and migrate nicely.
Even with another drive replaced the system is still hanging regularly.
I've decided to rebuild the array from scratch as soon as the current rebuild completes. I have a couple of matched, known-good Seagate 2TB drives on hand and another two on the way, plus a refurb spare, and I have a full, bootable backup of the RAID-5 volume made with SuperDuper, so I should be good to go.
But I'm not sure what the best procedure is for accomplishing the rebuild and restore. I can think of a couple of ways to proceed but I'm not sure which is preferable, or if there's a better way.
In every case I would start by powering down the system and replacing all four internal 1TB drives with the Seagate 2TB drives.
I could boot from the SuperDuper backup, use RAID Utility to create a new RAID-5 set across the four 2TB drives, use SuperDuper to copy from the (running, booted) backup to the new RAID-5 set, and then boot from (and continue using) the new RAID-5 volume.
Or I could make a Lion boot DVD, install bare Lion afresh on Drive 1, use RAID Utility to create a new RAID Set and migrate the existing system to the new RAID volume, and then restore the system from the backup using the Lion boot DVD or perhaps Lion's Recovery Mode (though I haven't used it before).
Would either of these work, or is there a better approach?
Carbon Copy Cloner - check or used it? Yes SuperDuper can restore the data - CCC can clone Lion Recovery - though the best place for that I think is on a standalone boot drive.
Why put the system on the RAID array? or is "Drive 1" the 5th in optical drive bay - can't see the need to do a fresh Lion install unless you need to.
The time to experiment with Lion Recovery is before you need it.
Sounds lke 3 workable methods though. All of them though depend on the RAID card not being the culprit.
I always learn something along the way and SoftRAID has done some digging on 10.7.4.
MPG: More and more caution seems to be needed with Apple system software updates these days, when cavalier changes like this are thrown in (disabling system calls in a minor release is reckless at best).
Personally, upgrades should always be done with a safety net, and not until the bug reports and dust has settled.
From today's MPG blog:
While working with Apple engineers, we discovered was that Apple had disabled one of the two sets of read / write calls in 10.7.4. Before 10.7.4, volume drivers could use one of two types of read / write calls for transferring data to and from a volume, the older one which was introduced in 10.0 and a newer one introduced in 10.6. Since SoftRAID still supports any Macs running 10.4, we used the older one when running in the 32 bit kernel and the newer one when running the 64 bit kernel (the 64 bit kernel was introduced with 10.6). When Apple disabled the older one in 10.7.4, all read / write calls to SoftRAID volumes would fail when they were made with the older set calls, i.e. on Macs running the 32 bit kernel.
We have just released a beta version of SoftRAID, 4.3.2 b9, which works around this problem by using the newer set of read write calls whenever a user is running 10.7. It is available by sending email to email@example.com. We will be releasing a final version of SoftRAID 4.3.2 in the next two weeks.
This same problem also affected other drivers on the Mac. For instance, some of the cards from Sonnet Technologies, Inc. experienced the same problem. They have released new drivers for their products which allows volumes to mount correctly when running 10.7.4 with a 32 bit kernel. Please see their web site (http://www.sonnettech.com/support/) for driver updates.
We have a couple threads about Lion and Apple Pro RAID card, issues with panics when more than 48GB RAM is present, no support for drives larger than 2.2TB. And the way Apple Disk Utility arrays were implemented when Lion came out. These types of things are why it seems someone is not keeping their eye on the Mac Pro line and traditional user base of pro users.
Question: are you bootiing using 64-bit kernel?
I haven't used Carbon Copy Cloner and I only recently began using SuperDuper. Is there a reason to prefer one of these to the other?
The system has always been on the RAID Array (i.e. one big volume) and I would prefer not to change that model unless I have to.
Yes, it has occurred to me that the RAID Card itself could be the culprit though I certainly hope not because that would be just another big can of worms. RAID Utility has started to complain now about unrecoverable bad blocks during the rebuild so maybe there's a bad disk in there (somewhere) after all. It's maddening that the system won't provide more low-level information about it. My ReadyNAS Pro provides tons of such information about individual disk state and the state of the whole system.
But I'm not sure I'll even be able to complete this rebuild because of all the hangs. In which case I'll just proceed to replacing the drives and re-initializing the array. If SuperDuper's bootable clone can do it, and provide the source for creating a new bootable RAID-5 volume (and it sounds like it can) then I'll go that route--certainly simpler than re-installing the OS from scratch.
For my next desktop Mac (which will either be an iMac or a new Mac Pro if there ever is one) I'm thinking that I might go with just a single internal drive, no Apple RAID, but get one of those big TB-enabled Promise Technology RAID enclosures with 4 or 6 2TB drives.
Actually having the system not be on the hardware RAID, and on a separate system boot drive, would be safer and preferable. And easy to manage.
Some use an SSD in the 2nd optical drive bay, or there are 10K WD VR in 250-500GB model in $154-200 (similar to a good 128GB SSD).
Have you ever visited http://www.macgurus.com and looked at their hardware cards and burly arrays? TB only makes sense - being new proprietary - on laptops and those w/o discreet PCIe slots. Seen www.barefeats.com on Promise vs others?
I was disappointed when the $1000 Apple Pro RAID card came out in 2007, after thinking at first it would offer SAS/SCSI and more than the 700MB/sec performance, and w/o the issues and behavior it has shown.
I used CCC first, it came out yrs before Superduper afterall and Mike is a certified terminal genius to boot. That it handles Lion Recovery partition clones is a bonus. And shareware. Started using it in 2003 myself.
The report on the MPG blog is certainly worrying.
But to answer your question: yes, I'm pretty sure it's a 64-bit kernel. I'm not booting that explicitly (i,e. I haven't changed the boot options) but I always boot in verbose mode (Command-V) and the console messages indicate that it's a 64-bit kernel.
My system has 18 GB of RAM, and the drives have all been 1TB until now (will move to 2TB as mentioned). I've been using external 3TB drives for TM and for the SD clone (connected via FW800 and USB) but I'm not sure that's relevant here.
So--perhaps--the issues mentioned don't apply in my case.
make sure you boot 64-bit mode
Noisy power and audio bug with Pegasus
I have optical drives in both bays (the Apple-supplied "SuperDrive" in one and a BD-R drive supplied by OWC in the other) so if I wanted a separate boot drive I'd have to use another interface, FW800 or USB or eSATA (I do have an eSATA card installed).
Sure, it could be managed. I guess I'd have to set up separate mounts and/or symlinks to the appropriate places, though that's a bit of a nuisance and is the sort of thing that's sometimes clobbered by Apple software updates.
Haven't gotten stuff from macgurus.com but I'll check it out, thanks. I've tended to get stuff from macsales.com (OWC) when adding third-party hardware to my Mac systems.
I'll also try using CCC to make a second bootable backup, if I can ever get this rebuild to complete. It seems to hang about hourly and makes only a little progress each time before the freeze--but at least it's non-zero progress. Still, it will take days at this rate.