Please recommend a 6G RAID card with internal ports for the Mac Pro 2010

[JollyJoeJoe]

Lion has issues with storage, RAID in particular. So I'd recommend waiting a bit, and let them get the kinks ironed out (wait at least one more update, but 10.7.3 would be safer IMO).

Once you go with RAID systems, updating to the newest OS tends to be a mistake. So it's a good idea to wait for it to mature a bit first before taking the leap (let the lemmings do that for you ).

Lion should be able to install to the SSD on the card (it's visible to the system, and so long as it's setup as the first LUN [value = 0:0:0], it will boot).

It would be a lot simpler if you get the OS/applications disk off of the Areca in a MP IMO though.


Given what you're trying to do (OS/applications disk on the Areca), Yes.

If you get the OS/applications disk, you simply remove the card before installing Lion, update it with the correct drivers, then re-install the card.


As per it being included in the main firmware (currently v. 1.49), No it doesn't include the EFI (EBC actually) portion that's used to replace BOOT.BIN in order to boot in your MP.

It's hard to find, but it's online (it's buried, and the only way to find it is via the ftp site).

The exact location is: ftp://ftp.areca.com.tw/RaidCards/BIOS_Firmware/ARC1880/MacPro_EFI_BIOS/. Just download it, and go from there (filename = ARC1880EBC.BIN, though as it's the only thing at that location, you can't miss it ).

BTW, it's also on the disk that came with the card.

Thanks Nano, say would have any idea what could be causing something like this: https://forums.macrumors.com/showthread.php?p=13205996&#post13205996

I can't tell for sure but it seems to be the SSD in pass through mode which is running the OS. This can happen on a rare occasion and I have a feeling it has something to with the OS disk momentarily having a connection lapse.

Perhaps there are some more OSX optimal settings I can input in "modify pass-through drive" such as different cache size, block size, no tag queuing etc ? It's either the settings or the actual physical connection whereby the poorly shielded Mini SAS 4i cable are being effected by EMI ? or some other interference. Cable that runs into the optical bay cage looks like so http://ipaconn.com/product_images/20101117132542250.jpg

Currently have:

Volume Set Name EXTREME Pro 6G
Raid Set Name Raid Set # 000
Volume Capacity 240.1GB
SCSI Ch/Id/Lun 0/0/0
Raid Level Pass Through
Stripe Size N.A.
Block Size 512Bytes
Member Disks 1
Cache Mode Write Back
Tagged Queuing Enabled
Volume State Normal

Also I should check if the Mercury 6G firmware I am using has any issues that
can cause this. I know it's not the RAID5 volume as it's disconnected from the controller.

 

[nanofrog]

What version of OS X are you running?

I ask, as there's still problems with Lion and storage systems. It's not mature enough yet, so stay away ATM, and stick to SL.

There are other possibilities, such as the staggered spinup values could be adjusted, but if the OS you're using isn't stable, it can't be ruled out of the potential causality list.

 

[JollyJoeJoe]

What version of OS X are you running?

I ask, as there's still problems with Lion and storage systems. It's not mature enough yet, so stay away ATM, and stick to SL.

There are other possibilities, such as the staggered spinup values could be adjusted, but if the OS you're using isn't stable, it can't be ruled out of the potential causality list.

Occurs irrespective of which OS 10.6.8 or latest Lion, matter of fact running Lion now and it seems better or at least has not occured since I last installed the Areca lion drivers but I am not counting on it.

Yep, just occured again. So irrespective which OS it occurs the same. One thing I did notice this time though is that the SSD with Windows 7 that is connected to the backplane and running off the onboard SATA2 controller was grayed out and inaccessible, there's two partitions that show up from that disk Windows and Reserved. Perhaps it's that backplane adapter SATA/power splitter supplied by MaxUpgrades that's causing the drop outs.

 

[nanofrog]

Occurs irrespective of which OS 10.6.8 or latest Lion, matter of fact running Lion now and it seems better or at least has not occurred since I last installed the Areca lion drivers but I am not counting on it.

Yep, just occurred again. So irrespective which OS it occurs the same. One thing I did notice this time though is that the SSD with Windows 7 that is connected to the backplane and running off the onboard SATA2 controller was grayed out and inaccessible, there's two partitions that show up from that disk Windows and Reserved. Perhaps it's that backplane adapter SATA/power splitter supplied by MaxUpgrades that's causing the drop outs.

I don't think it's cables from what I gather from your posts (that results in disks are fine one moment, then suddenly get drop-outs = array is degraded or Pass Through disks aren't available), but you can wiggle the connectors if you wish and see what happens.

So let's take a look at your settings first (screen shots of all of the settings would be helpful).

Another thing, IIRC, 10.6.8 had issues as well (looking at the eject issue makes me think OS related), so perhaps rolling back to 10.6.7 would help you (may need another clean installation of OS X, particularly as this would require different drivers than those for Lion).

This issue may be tricky and time consuming overall, so be patient (need a lot of information, and it will take time to sort through the possibilities).

It's also a good idea to contact Areca to see if they may have run into this, and can offer up any information, or ideally, a workable solution.

 

[sgginc]

I had drive dropouts until I set the Stagger Power On Control

Mac Pro 3,1; 10.7.1; Areca 1880ixl-8.

I had drive dropouts until I set the Stagger Power On Control to .4 seconds.
I read somewhere where another owner did this for the dropouts also.

System Controls > HDD Power Management.
The other three items are disabled.

I had drive dropouts every day.
Now they are rare.

Thanks ... Ken

 

[sgginc]

It's also a good idea to contact Areca to see if they may have run into this, and can offer up any information, or ideally, a workable solution.

Areca support is email only.
The recipient is in Taiwan and has minimal English skills.
They have no Macs to check your issue if they understand it.
His troubleshooting suggestions are one liners that rarely reflect an understanding of your issue.
Basically a waste of time.

Thanks ... Ken

 

[nanofrog]

I had drive dropouts until I set the Stagger Power On Control to .4 seconds.

This can help in some situations when all the disks are on the RAID controller (think instances of mixed disks = not all the same brand, let alone same model).

The other user that did this was mixing Seagates and Western Digitals (reduced spin up to .4s from the default value of .7s).

Now this could help with the SSD on the RAID controller (why I asked for screen shots as there's a lot of details that aren't yet known). But the SSD on the ICH (system's SATA controller) is another matter, and has nothing to do with the RAID controller at all (would be OS X).

So I see 2x different problems going on here.

Areca support is email only.

Most use email, but phone support is possible if the user is willing to call Taipei, Taiwan (TEL: 886-2-87974060).

If calling from the US, precede the call with 011, then follow with the rest of the digits listed above (886 = Taiwan's Country Code, 2 = Areca Code, and the remaining 8 digits are their actual phone number).

At least one member has actually called them for support rather than rely on email, which tends to take a little time for a response (~3 days on avg.).

His troubleshooting suggestions are one liners that rarely reflect an understanding of your issue.

More information can't hurt, as there may be an incorrect setting (why the request for the screen shots), and there appears to be more than one problem going on simultaneously.

Go back and look, as I even mentioned spin-up timings before any mention of the SSD on the ICH issue.

Sadly, 10.6.8 and Lion (both 10.7.0 and 10.7.1) are still having problems with storage systems, so that can't be ruled out. To do so, would take rolling back to 10.6.7. Then see what's going on. Fix it whatever problems remain with the RAID system, and make a clone before trying out Lion again if desired (better to wait for Lion to mature however IMO).

As per solutions, I suspect reducing the spin up timings to .4s will solve the OS disk on the Areca (SSD in Pass Through), but the SSD on the ICH is OS X related, and why the recommendation of rolling back to 10.6.7 (stay away from 10.6.8 and either version of Lion as they've all been reported to have problems, and also mentioned before).

Might I suggest reading more than the most recent post, and do so carefully, as there tends to be a lot of information in just a few lines.

 

[JollyJoeJoe]

Well it looks like my issue is less severe *knock on wood*, following some intensive i/o tests I have established a clear pattern. That is with Windows\System reserved volumes present the machine will error, with them ejected things hold up fine and considering I don't even need these volumes present may just make an unmount script at startup.

So the one SSD in pass through mode which I use for the OS and me RAID5 4x 3TB hitachi volumes are not having any dropouts and the controller is doing its job fine under lion. It's just the SSD running of the ICH with windows and system reserved volumes. And yes I just recalled that with 10.6.7 I don't remember seeing any dropouts, after upgrading to 10.6.8 they started.

Also spin up setting (.4s etc.) is also relevant for SSD's ?


So far I have noticed only the following bugs effect my 5,1 Mac Pro after installing Lion.

i. The machine will not go to sleep automatically, I have to do it through the file menu apple>sleep

ii. Upon pressing the power button the machine waits for almost 20 seconds before going to sleep, before it was near instant.

iii. For some reason it categorizes my volumes that are on the controller differently. i.e. My RAID5 array is split into 1x 8TB HFS and 1x 1tb NTFS. The NTFS one is categorized in finder preferences as DVD, Ipod etc. and the HFS one and Macintosh HD SSD correctly as external device.

iii. the disk drop out issue, but this is also preset with 10.6.8

 

[nanofrog]

Well it looks like my issue is less severe *knock on wood*, following some intensive i/o tests I have established a clear pattern. That is with Windows\System reserved volumes present the machine will error, with them ejected things hold up fine and considering I don't even need these volumes present may just make an unmount script at startup.

So the one SSD in pass through mode which I use for the OS and me RAID5 4x 3TB hitachi volumes are not having any dropouts and the controller is doing its job fine under lion. It's just the SSD running of the ICH with windows and system reserved volumes. And yes I just recalled that with 10.6.7 I don't remember seeing any dropouts, after upgrading to 10.6.8 they started.

Also spin up setting (.4s etc.) is also relevant for SSD's ?


So far I have noticed only the following bugs effect my 5,1 Mac Pro after installing Lion.

i. The machine will not go to sleep automatically, I have to do it through the file menu apple>sleep

ii. Upon pressing the power button the machine waits for almost 20 seconds before going to sleep, before it was near instant.

iii. For some reason it categorizes my volumes that are on the controller differently. i.e. My RAID5 array is split into 1x 8TB HFS and 1x 1tb NTFS. The NTFS one is categorized in finder preferences as DVD, Ipod etc. and the HFS one and Macintosh HD SSD correctly as external device.

iii. the disk drop out issue, but this is also preset with 10.6.8

Staggered Spin-Up was created for mechanical disks, not SSD's, as a means of reducing the load on the PSU during startup (which will blow the PSU if the instantaneous load, aka startup load, is higher than it can provide).

But it's possible that one disk (or brand) could time-out if the rest of HDD's aren't up fast enough, so changing the value can prevent this from happening.

Now in this case, the SSD is in Pass Through, but I don't know if it's treated separately from the rest of the RAID 5 set (I don't think it's that sophisticated - that is, it just sends the Power Up Signal to each group of 4 disks; each group of 4 disks being a port on the card), so it could be relevant.

There are also different timings between consumer and enterprise grade disks (i.e time-out values are 0,0 for consumer, 7,0 in enterprise; read, write respectively, and values are in seconds), which would exacerbate this issue (mechanical disk makers don't allow consumers to change these values any longer either, I'm afraid, and I'd be amazed if this is possible with any consumer SSD). But you could contact the SSD maker if this ever becomes a problem (seems that it's not ATM, based on the rest of the post this is responding to).

But I still think it a good idea to roll back to 10.6.7 for now, as it should solve a lot of problems (parts i, ii, and iii, assuming Lion or 10.6.8 haven't re-written any part of the array's partition table), and keep you from seeing anything else you may not have noticed yet (wait until they get Lion sorted).

Just a thought anyway, as it's your choice.

 

[wonderspark]

I'm curious about powering up a RAID. I'm on 10.6.8 with mine, and I'm not aware of any problems, but I thought maybe my system and procedures can help with the issue from this OP...

My specs:
Mac Pro 4,1 3.33GHz quad, 16GB RAM, ATI 5870, Areca 1880ix-12, Sans Digital TR8X w/8x 2TB 7200rpm WD RE4 drives in RAID3 (7x in RAID3 +1 hot spare), 1x 640GB 7200rpm Hitachi HDD (OS+Programs), 3x 1TB 7200rpm Hitachi HDD in software RAID0 via OSX.

Before I power up the Mac, I power on the RAID tower. I let it spin up for about ten seconds or so, then power on the Mac. So far, it has been perfect. After login, my RAID shows up on the desktop, and all is well. I've done the same thing with all external drives I've ever had, because it seems to work.

My question is: What is the point of the staggered spin up if I power that box up ahead of time? Is it only to protect the 300W PSU of the external enclosure? The specs on the WD2003FYYS drives are:

Electrical Specifications
Current Requirements

12 VDC
Read/Write 675 mA
Idle 510 mA
Standby 45 mA
Sleep 45 mA

5 VDC
Read/Write 525 mA
Idle 405 mA
Standby 185 mA
Sleep 185 mA

Power Dissipation
Read/Write 10.70 Watts
Idle 8.10 Watts
Standby 1.50 Watts
Sleep 1.50 Watts

I don't know how much power they draw during spin up, but it can't be anything much more than 90 watts for all eight drives to spin up at the same time, which is way below the 300W power supply rating.

Hopefully I'm not just junking up your thread with stuff that has nothing to do with the problem. I just don't understand how changing the spin up from .7 to .4 seconds can change anything, unless the PSU on that enclosure is barely able to handle all the drives inside.

 

[nanofrog]

Before I power up the Mac, I power on the RAID tower. I let it spin up for about ten seconds or so, then power on the Mac. So far, it has been perfect. After login, my RAID shows up on the desktop, and all is well. I've done the same thing with all external drives I've ever had, because it seems to work.

This is best practice to be sure the array is available (have seen instances where powering the system first, drives second didn't bring up the array - array = data only).

And your methodology is an absolute necessity if the OS is located on disks in the enclosure (i.e. separate disk or RAID 1 on the card).

My question is: What is the point of the staggered spin up if I power that box up ahead of time? Is it only to protect the 300W PSU of the external enclosure? The specs on the WD2003FYYS drives are:

Electrical Specifications
Current Requirements

...[snip]...

In the case of an external enclosure, Staggered SpinUp would reduce the instantaneous current draw from the enclosure's PSU, not the PSU in the Mac Pro.

Now sometimes this is actually the case with SATA disks, sometimes it's not, as there are multiple components that have to support it (PSU, disk, main board or card support). In the case of the PSU, it has to do with pin 11 on the SATA power connector. If pin 11 is a floating signal (RE4's have support for staggered spinup, so the disk requirement is there, as is the Areca), then it waits for a startup signal from the data connector. If not, then there isn't a staggered spinup at all, and all disks will pull from the PSU simultaneously, regardless of the card's settings.

The reason for all of this conditional mess above, is that SATA disks don't use a jumper to force a wait for a staggered spin up signal as SCSI disks do (it eliminates jumpers, but it means all the hardware involved has to support it).

BTW, the electrical data posted on your RE4's are average values, not peak. Figure ~ 3x the avg. total power draw per disk as startup. So ~33W with this particular disk, then multiply by 8 for worst case (264W). So regardless, the PSU in the enclosure has enough power to handle all 8x disks (let's presume the 300W PSU value is peak, it's still more than 264W, though not much overhead remaining, assuming it actually meets advertised specifications), and is how the enclosure should have been designed (this hasn't been/isn't always the case; I'd only be concerned with no-name enclosures at this point, as they may not verify that staggered spinup support is present in the PSU's they're sourcing).

As Sans Digital is name brand, the PSU support is there from what I can tell (would have tripped with older SATA disks, as they drew closer to 40W per during startup, and 8 * 40 is 320W). Figuring worst case (300W PSU value is peak), it didn't happen, so it suggests that the staggered spinup support is there.

I don't know how much power they draw during spin up, but it can't be anything much more than 90 watts for all eight drives to spin up at the same time, which is way below the 300W power supply rating.

Actually, 8x mechanical disks draw more than 90W when powered up simultaneously, but only for a moment (startup current draw is measurable in the millisecond range before it drops down to sustained values). See above as to how much a disk will draw during this period, and how this would affect your enclosure.

Here's a small primer on Inrush Current if you're a bit more curious as to what's going on (same as startup current draw in .pdf form from Fluke).

Hopefully I'm not just junking up your thread with stuff that has nothing to do with the problem. I just don't understand how changing the spin up from .7 to .4 seconds can change anything, unless the PSU on that enclosure is barely able to handle all the drives inside.

Its there precisely to keep from over-taxing the PSU running the disks during the startup phase.

For example, imagine a 24 bay enclosure unit with a 600W PSU (not uncommon). At 24 disks * 30W, that's 720W. Likely more than it can handle, as it's wattage values that don't state they're sustained, are almost certainly peak values (got to love marketing combined with cost cutting). So this is where figuring on power values as peak comes from above.

Older disks would be worse (24 * 40W = 960W). Now for just a moment, let's assume that the 600W value is sustained. That means peak power is 600W * 1.414 = ~850W (Root Mean Square calculation) is the peak power it can deliver, which isn't adequate for all disks trying to spinup simultaneously.

Staggered spinup is what allows these smaller PSU's to work without blowing up during startup.

 

[wonderspark]

Thanks for that, nanofrog (as always!)

So for the case of a tower with more drives than what the PSU can handle simultaneously, wouldn't you want to set the spin up to a larger number, like 2 seconds, instead of lowering it from .7 seconds to .4 seconds? .4 seconds is closer to simultaneous, is it not?

That's what has me confused.

 

[nanofrog]

Thanks for that, nanofrog (as always!)

So for the case of a tower with more drives than what the PSU can handle simultaneously, wouldn't you want to set the spin up to a larger number, like 2 seconds, instead of lowering it from .7 seconds to .4 seconds? .4 seconds is closer to simultaneous, is it not?

That's what has me confused.

Understandable.

Unfortunately, there's issues with setting the time too long. Specifically, disks already spun up begin timing out due to the slower spinup time of the remaining members in the set.

So though the staggered spinup setting is specifically targeted to bring disks up in a delayed fashion to keep from damaging the PSU, there's a stability issue that it introduces as a result (disks can drop-out before the set is even online).

So there will be times that a user will have to experiment with the time value as a means of both protecting the PSU, as well as get the array online and stable (more commonly done with mixed drive makes/models due to different firmware and components, though it can be necessary with all disks as a single make/model as well).