[parasbuy]

Hi All,

I have a 3.33GHz 6-core Xeon Westmere Mac Pro with 16GB of RAM; it has 1TB, 7.2k RPM hard drive with about 300GB used on it.

The issue I'm having is that I get the SBBOD too often. I installed MenuMeters and can see the performance of each of the 6-Cores. The first core maxes out occasionally and when it does, the SBBOD comes up; the SBBOD goes away when the core utilization comes down from 100%. all the other cores barely register, at most in the 1% to 4% range.

RAM: 2GB of 16GB are being used.
Disk I/O seems fine
Wireless seems fine

This all happened when I was using Preview to read a PDF off my local drive.

I also have slow performance when I am running other software, as well... Solidworks CAD Simulation, Photoshop, Lightroom, Video. The video can be choppy and the rendering takes a bit of time.

Thanks
parasbuy

[Spheric Harlot]

Check the Activity Viewer utility to see which process exactly is tying up the core.

If you have a BootCamp partition, make sure it's added to Spotlight's Privacy pane in the Spotlight System Preference panel. BootCamp partitions apparently sometimes cause mdworker - the indexing process - to go haywire.

[parasbuy]

The Activity Monitor is only showing an average of all the CPUs, so the CPU% is nominal, less than 10% even for the top process. I couldn't figure out how I could get it to show a particular CPU, so I installed MenuMeters, which gives the breakout.

I don't have any Boot Camp partitions that I am aware of, so there is nothing in the Privacy Tab of Spotlight. I do have Windows-7 64-bit on another hard drive, but I don't ever run it together with Mac OS X. I also do not have VM-Ware or Parallels, etc installed and do not run Windows through Mac OS X. The only way I run Windows is when I am booting up, I hold down the Option Key and (I am assuming) a BIOS menu comes up to select which hard drive I want to boot up in, at which time I tab over and hit enter on the Windows drive. I don't think there is any connection with the Cores issue; at the moment I'm not running Windows at all and am just in the Mac OS environment.

I've also noticed through MenuMeters, that it is not just the first core that hangs up, but it is sometimes one of the other cores. Sometimes it's the 4th core, or the 3rd core, or the 6th. It happens with Firefox (network has plenty of bandwidth), or deleting a lot of Trash.

thoughts? It seems the computer should be balancing the load across multiple cores instead of having one spike up, no?

thanks.

[Spheric Harlot]

ADD THE WINDOWS DRIVE TO THE SPOTLIGHT PRIVACY PANE. It is your Boot Camp partition. (That it is the only visible partition on its own particular drive is an irrelevant technicality in this case.)


Also, FWIW:
The Activity Monitor certainly shows ALL individual processes across all cores.

But the CPU percentages are NOT averaged out.
Since you have twelve cores (six of them virtual), you have 1200% maximum CPU load, as per Activity Monitor.

Whether a process can be balanced across multiple cores depends ENTIRELY upon its programming. If the developer spun it off into plenty of threads, load balancing across cores is easy. If the developer ****ed up and is running everything in a single thread, pop goes the weasel.

What is that top process?

[cgc]

I've found the one thing that can consistently bring my MacPro to it's knees is high disk activity.

[Spheric Harlot]

Which is why a rogue mdworker vainly attempting to index a non-indexable Windows drive can smash performance.

[cgc]

Originally Posted by Spheric Harlot 

Which is why a rogue mdworker vainly attempting to index a non-indexable Windows drive can smash performance.

I wasn't disagreeing, mdworker is a prime suspect, but there can be other things that cause high disk activity and the resulting degraded performance like virtualization where the Virtual OS is grinding away doing an AV scan, for example.

[karanmehta1980]

Smells like something is causing a conflict. My suggestions for your Mac would be:

1. Clean Start-up Items: - There could be one causing the conflict. To do this: Apple Menu > System Preferences > Accounts > Login Items
2. Check System Preferences: Under system preferences, click on others. This would have remaining elements present of any application you tried earlier.
3. Download Onyx: Its free and would verify the start-up disk and structure of your Mac’s system files. It can clean the cache, repair system permissions and execute .maintenance scripts. Run this once a month (Highly recommended for MAC). I am not sure if I can paste a link to download here in the forum (as mods may not allow it). Search for it in google.
4. Install Important Updates: Check if Apple has released any updates recently for your machine configuration and go ahead install them. Also see if there are updates from the software’s which you have installed.

Mdworker is (Meta Data Server Worker). It is a technology for spotlight. So it’s a search utility for your Mac. Since its part of spotlight, in case you wish to disable Mdworker, you will have to disable spotlight.

Disable Spotlight:

1. Launch Terminal and type the following: sudo nano /etc/hostconfig
2. Navigate using the arrow keys down the following entry: SPOTLIGHT=-YES-
3. ChangeSPOTLIGHT=-YES-toSPOTLIGHT=-NO-
4. Save /etc/hostconfig by hitting Control-O and the return key, next hit Control-X to exit the nano editor
5. Next, you’ll want to disable the index by typing the following in the Terminal:
mdutil -i off /
6. And to erase the current Spotlight index, type: mdutil -E /
7. Spotlight will be completely disabled in your next reboot.

Enable Spotlight:

1. If you want to enable Spotlight again, follow the same steps as above, but changeSPOTLIGHT=-NO-toSPOTLIGHT=-YES-
2. Then typemdutil -i on /in the Terminal
3. Reboot your Mac and Spotlight would be normal

[parasbuy]

Thank you, gentlemen, for your replies...

- added the Windows drive into the privacy pane of Spotlight so that it will not index the windows drive. also included the mac back-up drive. so, only the main Mac OS drive that i'm on now is the only one being indexed by Spotlight

- tried to disable Spotlight completely, but the instructions above, Karan, didn't jive with what I was seeing.. specifically after the sudo command, there was no navigation as I can see on my terminal. but, no worries, i'll do some searches and can figure out how to disable it. i rarely use it, anyway, so killing mdworker would be a good thing for me regardless. had done all the others, as well... Onyx, Startup, updates, etc... sorry, forget to mention.

- cgc... not having a high disk activity issue here. i'm also not running any virtualization that i know of... no VM Ware, Parallels, etc. the MenuMeters disk activity monitor is pretty clear. brought my virtual memory down to 50KB, so that should kill that pretty good.

- Spheric, interesting info on the Activity Monitor.

> Whether a process can be balanced across multiple cores depends ENTIRELY upon its programming. If the developer spun it off into plenty of threads, load balancing across cores is easy. If the developer ****ed up and is running everything in a single thread, pop goes the weasel.

...if I understand your comment properly, this sux, yo... seeing as how most people are not running a 6-core machine, it seems ridiculous that it's left up to the developer of the application to write their program to load balance properly. with all the different types of machines and core variations.. dual, quad, dual-quad, etc., how would a developer be expected to create different versions that work with these? it seems to me the operating system should be required to manage load balancing across the hardware on which the OS is running... the application developers should be required to program in 32- or 64-bit for that OS, but the interfacing with the hardware should be the job of the OS, no?

pls correct if i am wrong / misunderstanding. also, is there no way to modify / install code, perhaps at the application level or the OS level, to balance the load across the 12-cores?

to answer your question, the top CPU% process seems to be changing as you would expect it to... sometimes it's Firefox, Thunderbird, Activity Monitor, etc., depending on the app being run; it's usually running at <4% normally, occasionally going up to 10% or so per your FWIW post above.... it doesn't seem like it's being hung-up on any one process. By Threads (i'm assuming this is what your are referring to)... Firefox is 38 to 42 threads running the Daily Show, Thunderbird is 35, Lightroom 26, Songbird 21, Excel 15 Activity Monitor is 2, at the moment as I'm unsuccessfully trying to load it up...

thanks,
Paras.

[P]

Originally Posted by parasbuy 

- Spheric, interesting info on the Activity Monitor.

> Whether a process can be balanced across multiple cores depends ENTIRELY upon its programming. If the developer spun it off into plenty of threads, load balancing across cores is easy. If the developer ****ed up and is running everything in a single thread, pop goes the weasel.

...if I understand your comment properly, this sux, yo... seeing as how most people are not running a 6-core machine, it seems ridiculous that it's left up to the developer of the application to write their program to load balance properly. with all the different types of machines and core variations.. dual, quad, dual-quad, etc., how would a developer be expected to create different versions that work with these? it seems to me the operating system should be required to manage load balancing across the hardware on which the OS is running... the application developers should be required to program in 32- or 64-bit for that OS, but the interfacing with the hardware should be the job of the OS, no?

Making a task run across more than one core is generally very hard. Certain tasks are easy, but many are either very hard or impossible. Also, programs were developed for single processors for a long while, and not all of that is going to be rewritten over night. Almost all Macs from the last five years have at least two cores, so you're hardly alone.

A multithreaded application will run just fine on a computer with only one core, so more than one version is not necessary. The best thing (for performance) is to divide the task onto several threads and let the OS schedule them as it pleases. The OS is what is interfacing directly with the hardware, but the OS cannot parallelize a singlethreaded task - that is up to the programmer.

Originally Posted by parasbuy 

pls correct if i am wrong / misunderstanding. also, is there no way to modify / install code, perhaps at the application level or the OS level, to balance the load across the 12-cores?

The OS does balances just fine. What you're really after is to split a single thread across cores. That is called parallelization, and no, that's not possible by just installing something into the OS.

[besson3c]

I think the greatest bottleneck in OS X is a combination of memory leaks and I/O saturation. What technique are you using to check on your I/O activity? It is very easy to saturate a SATA drive, much harder to peg your CPUs. The Activity Monitor/top won't tell you about I/O usage.

[reader50]

Originally Posted by besson3c 

Activity Monitor/top won't tell you about I/O usage.

Sure it will. Disk Activity tab, with your choice of data bandwidth to/from the drives, or IO's per second.

[besson3c]

Originally Posted by reader50 

Sure it will. Disk Activity tab, with your choice of data bandwidth to/from the drives, or IO's per second.

Sorry, you're absolutely right! I'm too much of a Unix guy, I guess, as I have been equating Activity Monitor to a GUI for top, and top does not provide these stats (iostat does though).

At any rate, I'd definitely suggest analyzing your I/O, original poster. The SATA drives in your rig are definitely the weakest link.

[parasbuy]

Originally Posted by P 

Making a task run across more than one core is generally very hard. Certain tasks are easy, but many are either very hard or impossible. Also, programs were developed for single processors for a long while, and not all of that is going to be rewritten over night. Almost all Macs from the last five years have at least two cores, so you're hardly alone.

OK.. I didn't know how hard it was, so thanks for letting me know.


A multithreaded application will run just fine on a computer with only one core, so more than one version is not necessary. The best thing (for performance) is to divide the task onto several threads and let the OS schedule them as it pleases. The OS is what is interfacing directly with the hardware, but the OS cannot parallelize a singlethreaded task - that is up to the programmer.


The OS does balances just fine. What you're really after is to split a single thread across cores. That is called parallelization, and no, that's not possible by just installing something into the OS.

OK... tried to do some searches... What are some things I could look into further so that I could split a single thread across cores? Although I'm an engineer and have done quite a bit programming in the past, elaborate coding is not my primary focus. But I am interested to learn about what I could do to better my system performance, take better advantage of the capabilities of the system I have, and to make running my tasks more efficient. I'm not really sure where to begin, but could you point me in some directions that I can look into further? Thanks in advance.

[parasbuy]

Originally Posted by besson3c 

Sorry, you're absolutely right! I'm too much of a Unix guy, I guess, as I have been equating Activity Monitor to a GUI for top, and top does not provide these stats (iostat does though).

At any rate, I'd definitely suggest analyzing your I/O, original poster. The SATA drives in your rig are definitely the weakest link.

As you suggested, I've been trying to analyze my I/O data over the last few days. I try to use the Disk Activity tab in the Activity Monitor, but I mostly use the up/down lights showing disk activity in MenuMeters. When I get the SBBOD, however, the disk activity lights are not perpetually on for extended periods of time. When the SBBOD spins, the RAM, network traffic, and disk activity lights are operating with enough headroom. Occasionally, one of the cores will max out as mentioned above; but I've noticed the SBBOD will come on even if none of the cores at max capacity.

Is there another tool you know of that I could install to show disk activity better, to give you the info you need? I'm not sure how to check on Memory leaks, either.

This thing is just frustrating. ...I just don't understand what I need to do to get rid of the SBBOD when all my system resources are operating with enough headroom capacity!

Thanks,
Paras.

[AKcrab]

Have you done the standard "lets try and eliminate software issues" stuff? Have you created a new user to see if the same issues exist?

I hate to say it, but I think you're running down the wrong path. I think there is some sort of hardware/software issue here.. Bad RAM or a hard drive perhaps, but I don't think trying to "hack" the OS or individual programs is going to be productive.

[cgc]

Originally Posted by besson3c 

I think the greatest bottleneck in OS X is a combination of memory leaks and I/O saturation. What technique are you using to check on your I/O activity? It is very easy to saturate a SATA drive, much harder to peg your CPUs. The Activity Monitor/top won't tell you about I/O usage.

I wouldn't classify memory leaks as a bottleneck, more of a resource hog as your memory will not be deallocated when it's no longer needed. Should be easily revealed via Activity Monitor (e.g. no free memory but with few apps running).

[parasbuy]

Originally Posted by AKcrab 

Have you done the standard "lets try and eliminate software issues" stuff? Have you created a new user to see if the same issues exist?

I hate to say it, but I think you're running down the wrong path. I think there is some sort of hardware/software issue here.. Bad RAM or a hard drive perhaps, but I don't think trying to "hack" the OS or individual programs is going to be productive.

Thanks for the info... it seems strange to me, as well, that I have to go to such lengths for this. I have completely formatted and reinstalled Mac OS X. The RAM and HD seem to be working fine, but I'm not sure how to check for sure. I have MenuMeters up showing each of the cores, network, disk activity, and RAM capacity. From what I can see on these meters, everything seems to have enough headroom, but I still get the SBBOD.... meaning, none of these are maxed out even when the SBBOD is spinning.

Are there any other monitoring tool that might better help identify the bottleneck?

Thanks,
Paras.

[PhilCat]

I see somewhat the same activity where a process
pegs one of the cores to max with little help from the others.

I do have the dreaded spin now and then, and a new install might help.
Or a new user could be tried.
Try leaving the console open to view when the spin begins.

Here's a topic on Kernel Panics

Do you use fan controller? Not that it has anything to do with issue.
This made a nice difference and still very quiet. I set them to 1205RPM.
The temp rarely goes above 116-120f, when it used to be 154f+.
About every 3-4 months, I clean the inside like new.

All 8 cores get cranked up to about 80% if I recall with
Toast Titanium when converting a copy for Apple Tv.

08 model, 4, 1TB Drives
MacPro3,1, Quad-Core Intel Xeon 2.8 GHz
Processors:2, Cores:8, Memory: 8 GB, Bus Speed: 1.6 GHz
NVIDIA GeForce 8800 GT
I installed 2- TSSTcorp CDDVDW SH-S222L

[Spheric Harlot]

Unless he's seeing a kernel panic, there's not much point in actually researching them.

[parasbuy]

Thanks for the info...

- after getting frustrated with this, i finally bit the bullet this week and did a clean install. completely formatted and reinstalled everything. i also made sure to completely disable Spotlight so that there is no activity whatsoever

- i am using MenuMeters, iStatPro, and Activity Monitor as my sources of info

- unfortunately, i'm still seeing the same poor performance.. I'll get the SBBOD when one of the Cores will max out while the others sit with plenty of headroom. RAM, disk activity, and network communications are all showing plenty of headroom with none of them maxing out or being pegged for an extended period

- in addition, I also noticed this week that I'll sometimes get the SBBOD even when there is 'not' one Core that is pegged and they have plenty of headroom. However, I did notice that the Page Ins/Outs in MenuMeters will show a "1k+" ...I'm trying to keep track in Activity Monitor to see if I can get better info on the break-out and I'll post that when I get the info. (if you guys know of a way I could look through the history to see the event or somehow record it going forward, pls let me know.)

- at the moment, here's what I've got under System Memory:
Total RAM: 16 GB
Free: 12.24 GB
Wired: 1.09 GB
Active: 831 MB
Inactive: 1.85 GB
Used: 3.75 GB
VM Size: 111.18 GB <-- does this look right to you guys??
Page Ins: 2.02 GB
Page Outs: 0 bytes
Swap Used: 0 bytes

if I can show you any other info, let me know and I'd be happy to do so.

Thanks,
Paras.

[parasbuy]

Thanks, PhilCat...

I'm not using a fan controller, if you think this would make a difference, I will look into it and find and install one, etc. if you can recommend one, i can look at getting/installing it.

in case it helps, here is the info related to this that I've got... I'll also try and get the info the next time the SBBOD comes out...

Fax Speeds (RPM):
- PCI: 799
- BOOSTA: 856
- Exhaust: 599
- Power Supply: 599
- Intake: 600

Temp Sensor Readings (degrees F):
- HD1: 103
- HD2: 85
- HD3: 81
- CPU: 92
- Ambient: 85
- HD Bay2: 86
- HD Bay3: 88
- HD Bay4: 88

...looks like they are running pretty well, let me know if you think otherwise.

Thanks,
Paras.

[Spheric Harlot]

If a fan controller would actually make a difference to your issues, it is absolutely guaranteed that your machine's issues are not solved by upping the fan speed.

That's like recommending somebody rinse their mouth out more frequently when they're coughing blood. I mean, sure, why not?, but, erm…

[parasbuy]

Thanks for the analogy...

Any more thoughts on the other stuff?
- VM size?
- anything else to look at?

[PhilCat]

Originally Posted by parasbuy 

if you can recommend one, i can look at getting/installing it.

SMC Fan Control is the one I use.
I should of been a little more detail what temp I monitor.
Temperature Monitor v4.8 is used as well.

I picked the the memory modules being they tend to run the highest of the hardware.
The modules tend to run at different temps.
In mine, I look at-
Module A,1-4 and
Module B, 1-4.
I selected Module B2 to monitor.
For some reason it runs the highest like right now-120f.
The other modules, one at 104f, the lowest.

Fans at 1100 - 1200 rpm, CPU average is 75 - 77f.

Prior speeds at 600 - 800rpm, the same memory module
that is 120f now would be 150 - 160f.
I felt that was a little high. Other reports felt the same.

Like you, just trying different things little by little to make
anything better.
The quest at getting the spinning under control.

The memory banks-
A spare set of trays assembled to swap out complete would make an interesting lab test.

Past tests claim they are fine, but a swap and re-test complete would prove a couple things.
The test itself and the ram that could be suspect.

poking the ball to get at the truth would be the option.

I've pulled the trays just to reset them,
for a possible better contact, inspecting the contacts.

Same with all 4 drive bays.
It's all quite straight forward and easy to do, so why not.

Regards

[parasbuy]

Been keeping track per above.... noticed that the SBBOD will come up from anywhere between 5 to 60 seconds or more, while all the resources noted above (CPU, RAM, HD, etc) are showing adequate headroom. After some time, I'll hear a relay or something click, after which the PageOuts arrow and HD busy light will come on, often only for a second or less, then everything clears and ready for the next time the SBBOD comes out.

During this time delay with the SBBOD out, if it is more than a few seconds, the Force Quit Menu will say that the application is not responding, and then after the relay-sound and clearing, the "not responding" will clear.

FWIW, I have (3) HDs in the Mac Pro... first is the main 1-TB drive with Mac OS X and the issue being reported here. The second is a 1-TB backup drive (via Carbon Copy Cloner). The third is a 500GB drive with Windows 7 on it. I've noticed the time delay, etc., above regardless of which drive I'm using; though I haven't qualified the experiences with other two drives to the length as on the first drive. But there definitely is a delay in when I'm in Windows 7, as well.

Does this shed any light on the situation?

Thanks,
Paras.

[reader50]

It suggests the OS is waiting for one of the drives to spin up. It's unclear which one - the OS might want to check on every connected drive, which would make it "slowest drive to respond".

Perhaps you could pull one of the alt drives, and use the computer for a day or two. See if the behavior stops.

[parasbuy]

Thanks a lot, reader50. I 'finally' feel that we're on the right track here... here are the results from the last week.

- when only one drive is hooked up, the Mac Pro works like a champ! Blazing fast, no SBBOD... except that even when it does take some time (like when I really loaded it up with video, simulation, CAD, etc.) the RAM used still never went above 2GB and 14GB remains free.

- getting back to the hard drives, the Pro works great when there is only one drive hooked up in the bay. This is regardless of which one drive and in which bay. I've tried them all individually and kept track of boot times, etc.

- when two drives are hooked up in the bays, the Pro does slow down a bit, the SBBOD does come out, and the hard drive light in MenuMeters is on for extended periods until the 'relay' sound clicks to clear it. this is true for any combination of two drives in any of the bays.

- as stated earlier, the performance is markedly worse when all three drives are in. this is true even when I'm only working off of one drive; for example when I'm using Mac OS X off of Drive #1, and Drive #2 is used only during Backup and Drive #3 is only used when I need to go into Windows-7

- the three drives are pretty similar from a performance standpoint... not sure if the formatting will come out, but here are their specs. Basically, they all have pretty much the same 7200 RPM speed, latency, read seek time, write seek time, and 3Gb/s SATA interface. The only major difference is the Buffer size on Drives #1 & #2 is 32MB, and for Drive #3 is 16MB.

Drive #1 Drive #2 Drive #3
Model No. wd1001fals st31000340as wd5000aaks
Rotational Speed 7200 RPM 7200 RPM 7200 RPM
Buffer Size 32 MB 32 MB 16 MB
Average Latency 4.20 ms 4.16 ms 4.20 ms
Read Seek Time 8.9 ms < 8.5 ms 8.9 ms
Write Seek Time 8.9 ms < 9.5 ms 10.9 ms
Interface 3 Gb/s (Max) SATA 3 Gb/s 3 Gb/s (Max)


thoughts?

Thanks!
Paras.

[reader50]

The drives are probably spinning down to save power and reduce noise. If the OS requests data from one of the sleeping drives, it will presently spin up. You hear the click when the drive comes all the way back. It must be spinning them up one at a time too, to create the longer 3-drive delays you report.

You could try this: go into System Preferences -> Energy Saver -> Sleep (tab)

Near the bottom, there is a checkbox to "Put the hard disk(s) to sleep when possible". Uncheck this box, so the drives will run continuously.

This will increase your power usage, will keep the drives at max noise all the time, and may shorten their life spans. Also, some green drives may not obey the setting - their firmware may put them to sleep when inactive anyway. But with your 7200 RPM drives, it may do the trick. If you can live with the minuses.

[anthology123]

If the issue is drive related, and you have 4-1TB drives, how about setting up a striped RAID (the good stripe, not the bad one). Maybe all 4 drives will behave better in that mode.

[SierraDragon]

It sounds to me like the drives control process is misbehaving. Consider a trip to Apple techs for evaluation of the disk controlling process on your MP. With no RAID card I do not know how that process works: all OS X software, firmware (potentially corrupted), or actual hardware (also potentially corrupted).

And trying a software RAID0 is not a bad idea, if only to force disk control into a different mode than where it is now.

RAID0 is fast and an efficient use of drive space. However remember that a crash in RAID0 (striped) loses all the RAID0 drives' data, so what I set up is just two drives striped in a RAID0 array, leaving 2 more drive bays, one for OS/apps and one as a backup drive. The OS/apps drive and the two-drive RAID0 array should stay underfilled for speed (less than 70% full as a rough guideline for speed) but the backup drive can be filled pretty full.

If you need more bays the second optical drive bay can easily be converted to take a fifth drive.

Good luck!

-Allen wicks

[Edit: One my just-ordered new box, a 2011 MBP, I am moving to SSD for the OS/apps drive plus a HD in the optical bay drive slot for data. We will see how that goes, but my expectation is that SSD is the wave of the immediate future.]

[anthology123]

Is it possible to do a stripe RAID with parity? If so, that should resolve doing a stripe RAID, controlling all 4 drives, and have fault tolerance. If Disk utility does not support a stripe Raid with parity, then never mind, unless you have the time to experiment with the raid and your basic system and apps and see if it affects how the drives behave. In theory they should all be working together and not going to sleep independently.

[King Bob On The Cob]

Originally Posted by anthology123 

Is it possible to do a stripe RAID with parity? If so, that should resolve doing a stripe RAID, controlling all 4 drives, and have fault tolerance. If Disk utility does not support a stripe Raid with parity, then never mind, unless you have the time to experiment with the raid and your basic system and apps and see if it affects how the drives behave. In theory they should all be working together and not going to sleep independently.

I do not think Apple offers RAID 5 or 6 as an option for Software RAID. It would simply take too much CPU time to be useful, and you would run into the write hole issue every time your computer has an unexpected power outage.

[parasbuy]

Thanks for the advice.

I've shut-off the energy saver and it did help a little bit, but is still noticeably slow. The SBBOD does not come up much, but changes from one window to another or running a video take longer to load than if only one drive is connected.

I will look into the RAID configurations.. can you point me in a direction to learn / install / manage one? I understand the concept, but don't know much beyond that. Isn't there any other way to perhaps "prioritize a drive" or manage the drives individually? I will talk to Apple about the firmware, as well.

What about the RAM? I have 16GB in there and I've never seen it go above 3.16GB. What can be done to have the RAM be better utilized?

Thanks!
Paras.

[Don Pickett]

Originally Posted by parasbuy 

What about the RAM? I have 16GB in there and I've never seen it go above 3.16GB. What can be done to have the RAM be better utilized?

Thanks!
Paras.

Nothing. OS X has a most excellent memory subsystem.

[reader50]

Everything we've covered should have solved your problem. You might try zapping PRAM, but I've never heard of pram causing a problem like this.

Next suggestion: dig out your original install disc, option boot, and choose the Diagnostics partition on the install disc.

Run your Apple Hardware Tests, see if it can flag anything.

[parasbuy]

OK, thanks very much... will I be able to use a RAID array even if one of the three HDs is Windows 7? The other two are just the primary Mac drive and its back-up.

Is there a way to adjust the OS to keep just the primary drive spinning and the others essentially off? I only need the back-up drive when I'm running a back-up sequence or need to access it for a file, which are rare. The Windows 7 drive, I'd like to keep shut-off while I'm in Mac OSX, and only use when I actually boot into it.

At the moment, I am keeping the side-panel off and pushing either the primary Mac drive or the Windows 7 drive in when I want to use them. Having only one drive connected seems to be the best (and least frustrating) performance; however, this is obviously very inconvenient for long-term use.

Thanks,
Paras.

[AKcrab]

When are you going to take your machine to a shop? You shouldn't be seeing these issues. Having to manually engage and disengage drives? That is not how it should be "working".

[Waragainstsleep]

Originally Posted by parasbuy 

Thanks for the advice.

What about the RAM? I have 16GB in there and I've never seen it go above 3.16GB. What can be done to have the RAM be better utilized?

Thanks!
Paras.

You're not trying hard enough! My MBP is at 4.533 Used right now.

[reader50]

Originally Posted by AKcrab 

When are you going to take your machine to a shop? You shouldn't be seeing these issues. Having to manually engage and disengage drives? That is not how it should be "working".

There is also the issue that internal SATA connectors were only designed for at least 50 connect/disconnect cycles. That's one of the reasons why they made a separate eSATA connector, designed for greater usage.

We've eliminated the software issues. Try the Apple Hardware Tests and/or take it in.

[parasbuy]

Ok. will do... thanks!
Paras.

[parasbuy]

Hi All, thanks a lot for your posts over the last couple of months. I did end up bringing the Mac Pro into Apple. Apparently the problem had to go up a few levels to the Head Genius; took a couple of weeks but she is 'humming' now with all three drives in.

If interested, here's the story regarding the resolution. The lower level techs tried their best but concluded that the SMART Controllers of all three drives had gone bad and recommended I get new drives replaced under warranty from each of the OEMs. Obviously, I called "bull crappola" and that the likelihood of all three drives going bad at the same time was remote, at best; and if it was true, that the Mac Pro must have done it and Apple should replace all three drives with new ones. After a bit, the Head Genius gave me a call at home and thought the conclusion was bogus, as well. He ended up testing each drive individually on their tester and they each performed stellar. Replaced the entire back plane and still no improvement. Eventually, he got to the RAM, pulled them all out, and re-sat the four 4GB sticks just for kicks and, voila, the SBBOD went away. He surmised that for whatever reason, a section on one of the sticks probably wasn't getting the right connection and the computer would get hung-up ever time it was engaged. That was it... nothing more complicated, it seems.

Regardless, she's performing like a champ and I'm a happy camper.

Thanks again.
Paras.

[cgc]

I doubt all SMART controllers went bad on your three drives but I remember saying something along the lines of "I can't believe all six db-25 cables went bad at the same time" but they did and a long time ago I found a bad TWT with 10 signals on it but only made one take errors...weird things happen

Glad those "Geniuses" fixed yer Mac.

[Spheric Harlot]

Read on. It wasn't the drives.

[cgc]

Doh, yeah must have fallen asleep reading again.

[parasbuy]

Hi Folks... back again with this (unfortunately).

shortly after the last post, the Mac Pro started acting up and the SBBOD would come again and again. this time, the Apple guy suggested I might try one of their authorized service centers instead of the Apple Store, because the Apple Store employees must follow strict, pre-approved processes and the auth centers may have a little more leaway in what to try.

so i took it in to the local shop, he ran some tests, and following the Apple procedure, couldn't find anything wrong with the computer and gave me that as the "official" report. however, what was happening is that the Pro runs really well with the "Apple hard drives" that they ship with the computer, currently the Western Digital Caviar Black (7200 RPM, 1-TB, 32MB Cache). during the tests, the Apple protocol requires that only Apple hard drives are in the machine; in this case, two of the four bays had these drives, and with just these two drives in, there is indeed no problem with the computer.

however, the other two drives are a Seagate Barracuda 7200.11 (7200-RPM, 1-TB, 32MB Cache) and a Western Digital Caviar Blue (7200 RPM, 500GB, 16MB Cache). Seems like the difference in the spec of the Caviar Blue from the rest may have been causing the problem; perhaps the spin-up time and smaller cache was causing an issue. the shop guy had tried a number of "unofficial" tests with various combinations of the drives and "suggested" i might try running it without this slower drive.

the Pro is running really well now using just the three identical drives in the drive bays... two are Mac OS X Lion (one main and one back-up via CCC) and the third drive is Windows 7 Pro. the 500GB drive also works well, just that it's in its own seperate enclosure. been running for almost 2-weeks now and she's looking perty again.

here's hoping this saga is over now...

Paras.