Build 1970 - Bug on Mac Pro - Xeon proz ?

Hi,

i just installed a 1970 trial of paralles for mac on my Mac Pro.

I installed windows XP, activated it and updated it. everything works as supposed.
I am working with parallels on another machine (macbookpro at work) since several month without problems

now the problem:
--
everytime i log out on windows, selecting the shutdown option in windows, everything crashes.
That means:
- OS X itself freezes and brings up a grey box with the information that i have to shutown os x using the power button.

side-effect: software raid is crashed after this....


my question is: is there a known bug in 170 on Mac Pro architecture ?


best regards
fidel

 

I also have problems with Parallels b1970 on my Mac Pro. I also have a RAID 0 setup. Yesterday, my backup software crashed while backing up my virtual hdd. I spent all day yesterday restoring my RAID, and then doing a complete re-install of WinXP Pro onto a new virtual hdd (the old one was corrupted). After creating this new 'clean-install' image, I backed it up in case anything happens again.

Used my mac all day today, and then went to launch Parallels. While XP was booting, my RAID crashed again. After restarting, I was able to use DiskWarrior4 to rebuilt the RAID's directory (unlike yesterday - it gave me disk malfunction errors).

I'm using a Mac Pro, 3 GHz, with 8 GB RAM, Nvidia FX4500. Boot drive is a Raptor150, User drive is a 4 disk Raptor RAID 0 array. Virtual hdd is on the RAID volume.

I had been using Paralles since 11/21/06, and haven't had any problems EXCEPT for this RAID crash, and another one just like it that happened on 12/25/06. Both RAID failures occured 34 days apart. This latest one happened one day after a clean VE install, but at least the journal kept the RAID from being unrecoverable.

There are many posts regarding Mac Pro issues, and particularly RAID problems. Why hasn't anyone from Parallels addressed any of these posts? Why can't they setup a Mac Pro to test with? Why is this product still in beta? Why do they keep adding features instead of polishing a build that actually works on the current hardware line? Can I expect that Parallels will EVER work on the Mac Pro?

I've now wasted 4 days to restore my data because this software is causing un-recoverable RAID failures.

Jan 30 23:58:19 MacPro kernel[0]: [Parallels] IPI stat: rescheduled 4 cpus out of 4 (Ok)
Jan 31 00:00:00 MacPro kernel[0]: AppleRAID::completeRAIDRequest - error 0xe0030005 detected for set "Stripe" (C7D18741-321E-498C-A387-708E71F61A1A), member DBD2572D-37BC-47E1-8A7B-CF286C440D4B, set byte offset = 40170795008.
Jan 31 00:00:00 MacPro kernel[0]: disk4: 0xe0030005 (UNDEFINED).
Jan 31 00:00:00 MacPro kernel[0]: AppleRAID::recover() member DBD2572D-37BC-47E1-8A7B-CF286C440D4B from set "Stripe" (C7D18741-321E-498C-A387-708E71F61A1A) has been marked offline.
Jan 31 00:00:00 MacPro kernel[0]: AppleRAID::restartSet - restarting set "Stripe" (C7D18741-321E-498C-A387-708E71F61A1A).
Jan 31 00:00:00 MacPro kernel[0]: disk4: media is not present.
Jan 31 00:00:01 MacPro kernel[0]: disk4: media is not present.
Jan 31 00:00:15 MacPro kernel[0]: jnl: do_jnl_io: strategy err 0x6
Jan 31 00:00:15 MacPro kernel[0]: jnl: end_transaction: only wrote 0 of 86528 bytes to the journal!
Jan 31 00:00:15 MacPro kernel[0]: disk4: media is not present.
Jan 31 00:00:16 MacPro kernel[0]: jnl: close: journal 0xb95ad7c, is invalid. aborting outstanding transactions
Jan 31 00:00:17 MacPro kernel[0]: [Parallels] IPI stat: rescheduled 4 cpus out of 4 (Ok)

CAN SOMEONE PLEASE DO SOMETHING FOR US CUSTOMERS WHO ALREADY BOUGHT THIS PRODUCT AND CAN'T GET IT TO WORK???

 

thats what the trial is for.

 

Parallels stated compatibility with the Mac Pro months ago, and I'm running the 'production 1970' build - not a beta. I did use the trial, and it worked fine for the time alloted - unfortunately, the build that I "trialed" isn't available anymore, and Parallels suggests using the 1970 build.

maybe you should take some of your own advice about making posts like you have on this thread;

http://forum.parallels.com/post41287-6.html

 

Bitjunky, I can understand your frustration with your disk failures, but that's no excuse for a personal attack on the forum, nor, most likely, for placing the blame on Parallels.

Your problem is much more likely to be hardware related than software. If you are not using a UPS with your machine, you may be having power problems. If that isn't the problem, RAID 0 may be the cause of the problem. With RAID 0, you are depending on four (in your stated case) disks to work perfectly, rather than just one. This reduces your reliability by a factor of four for the physical disks since failure of any one will destroy the entire thing, plus the RAID filesystem is more complex and therefore less reliable than single disks.

RAID 0 is great for creating a large capacity volume with good performance for TEMPORARY data, but based on several decades of experience designing hardware and software, I don't recommend it for anything else, and certainly not for an operating system disk.

If you are looking for reliability, RAID 5 is a better tradeoff since you add the complexity of RAID, but gain protection from a single HDA failure, but at the cost of some performance.

In any case, from your description of your problem, it seems to have nothing to do with Parallels, so I think you would be wise to look elsewhere for the cause of the failure. Also, venting your frustrations by calling people names appears immature, an image you probably don't want to cultivate.

 

sorry.. exactly where did i belittle you ? Where was I hostile to you ? Did you even read the full thread that was posted in ?

If youre going to try to quote a post that i wrote, as an attack on me, you might want to get it into context first. I was merely pointing out, that shouting at parallels that you bought a product that doesnt work, isnt likely to get you much of a reply, given that they provide a 30 day trial period for you to test this stuff out before buying.

Such a short temper.. sheesh

 

joem,

My frustration with the disk failures had nothing to do with my attack on darkone. I think that we can both agree that his post was not meant to be helpful in the least, but rather to be snide about my predicament. Sorry - I don't think that there's any way to put a positive spin on it. His post was worthless, and only served to clutter up this thread.

Second, I'm not using RAID for the system disk. It's only for data storage, and it's backed up nightly. I know the benefits & pitfalls of RAID; it's too bad that Apple didn't enable RAID5 on the MB, or I'd use it. In fact, it's not a hardware problem - when the RAID failed, I pulled all 4 disks and started fresh with new disks that I keep on hand for just this sort of emergency - I've since tested the old disks and they are stable. I also use a power-conditioned (AVR) UPS with 4 hours of uptime for this workstation.

I think that all of my comments/questions were reasonable - albeit voiced with an air of frustration. As far as 'blaming' parallels; I didn't do so any more that any other poster who has made their best to pinpoint the cause of a problem and voice it on the forum. I was complete in my description of the problem, the environment in which it occured, and its history. Since you are obviously knowledgeable with these forums, I'd be interested to hear your opinion on whether there have been a disproportionate number of problems regarding PDM on the Mac Pro. If your answer is yes, then I'd argue that my frustration regarding the current state of the product is valid. It was announced as compatible, and it doesn't appear to be so.

 

If I had only written about how i felt about the situation, you'd be right. But the point of the post was to provide valuable information regarding a problem I'm having with the software. I did vent a bit, and that's all you seemed to care to read about and reply on. If you had read the whole post, then you would have seen the strange history of the failures happening 34 days apart, and known that the 'Trial' wouldn't have allowed the situation to arise. I'd be surprised to hear that it's anything but a coincidence, but it should have given you pause about posting what you did.

Short temper? only when I see my threads get cluttered up with nonsense like the last 6 posts (this one included).

 

And I wasnt commenting on the rest of the post, i was commenting on the capital letters that are generally considered to be shouting. As per the thread which included the post that you were quick to link to, its very tiresome to see people whining in capital letters. I apologise if that was not your intent. Especially as you have now edited your previous post to take out the insult.

 

and i apologize for reacting as harshly as I did. However, there should be a distinction made between whining, and valid frustration of users who can't get any support from Parallels. According to the majority of people who have tried to contect Parallels regarding issues they are having, as well as my own personal experience, Parallels doesn't even respond to the support requests. I know everyone here appreciates that work that Parallels is doing, but that is not an excuse to not support their customers. I previously gave them the benefit of the doubt because they're a small company, but they have been around long enough and had enough sales to justify hiring more support personnel - and that's especially true now knowing they are supported by swsoft. To claim that a product is compatible with certain hardware, and takes sales for it and then ignore support request from paying customers is a sure way to conceed your position to the competition.

 

I guess when I saw 4 disk RAID, I put that together with the MP's 4 disk capacity and made an unwarranted assumption -- sorry.

I read most of the posts here that relate to general issues and Windows, and yours is the first RAID crash I remember. Hopefully some others with RAID arrays will chime in and comment on how well they work.

I'm trying to think of a way Parallels could cause a RAID problem, and coming up blank. All I/O goes through OSX. There are no guest disk drivers (or they talk to Parallels code that uses the OSX API to talk to the disk). I'd think Parallels would be a pretty good disk and RAM tester since it uses a lot of cycles spread all over the machine and exercises resources that are idle most of the rest of the time. I'd expect the same stress level with Parallels that I'd expect from a heavy database application that uses a lot of resources and does a lot of I/O.

And I agree with you that RAID 5 would be a nice feature for Apple to implement. It would place me one step closer to a Pro of my own, since I've been in this business long enough not to trust one disk with anything I want to keep. My network RAID 5 boxes have been pretty solid even though they run on Windows servers (obligatory Mac plug /Windows bash whether I mean it or not).

I'd be curious to know how this is resolved if you get to the bottom of it.

 

joem,

I've had two catastrophic RAID failures (DU & DW4 coudn't repair the directory) that both happened when my backup utility was writing the Parallels hdd file. I've also had two recoverable RAID failures (journal is able to rebuild the volume) that happened at exactly the same point during the XP boot process (actually, right when I logged into Windows from the "Welcome Screen"). I'm thinking that because my system is on a different disk than the VM, that perhaps I'm having the same problems that other users (with raid) have reported, except that my kernel doesn't panic and require a hard-reset. My home is on the raid though, so I instantly have problems when the raid goes offline, but I'm able to logout and reboot successfully. OS X was able to repair the raid during boot in both of the Parallels crashes, but I still logged in as a user (with home on the system disk) to check the volume out with DW4 (it wrote a clean directory without finding any errors) before logging back in as the user with it's home on the raid. I was able to boot the same XP hdd that caused the 'recoverable' raid crashes two days ago and run it for the last 24 hours without issue. Exact same image (fresh copy of the image each time - it's a new hdd that I installed and patched fresh two days ago), two very different outcomes...

Otherwise, this Mac Pro has performed flawlessly for the last 4 months - the one and only KP that I got (ever) was during the original install of Parallels b1940 which was found to have known bugs with the Mac Pro.

I do have 5 sata drives in the Mac Pro - I wonder if that had any bearing on the issue? I haven't heard of or experienced any other problems though...

Anyway, hopefully Parallels will be able to test a raid configured Mac Pro with XP guests and duplicate the problem. Any further thoughts you might have are welcome.

As far as how well the raid works? Well 4 raptors go real nice, and when XP hdd is run from the raid, it's about 3x faster than when launched from an external FW800 drive (transparent 2-disk raid 0). When it works, it really works... too bad I'm losing all my hair wondering when my desktop is going to instantly "clean" itself

BTW: I haven't had ANY problems with my Windows 2000 Advanced Server VMs - could this be a Windows XP thing??

 

I think that four RAID failures in a month probably isn't a big enough sample to blame XP over Win2k. If the failures happened while the backup software was running and Parallels wasn't, I wouldn't suspect Parallels. The fact that backup was writing a large file, presumably larger than most files on your system, might have some bearing on the problem, but it could also be that spot on the disks, or that many blocks written or read with something happening at the same time as a page fault, or some such data dependent situation. For a backup to fail due to the specific content of a file is unreasonable, IMO. Additional failures during a read of the same XP file that backup was reading would make me suspect the disks, or maybe the RAID driver being corrupted, perhaps by a memory problem, although the Pro uses ECC memory IIRC.

FWIW, it still feels like hardware, or driver corruption, possibly from some third party software (could be Parallels, but others would be having problems if it were, I'd think).

This is a long shot, but you might try copying the XP VM to another directory, placing it in a different spot on the disk, where an incremental backup won't read it, and seeing if that stops the problem from happening. This is probably not too satisfying though, due to the long interval between failures.

And since the Pro has four disk slots, I wonder if Apple tested it with a four disk RAID, or only three.

As you can probably tell, I'm clutching at straws here, but in your place, I'd be pretty frustrated, and even a straw might be useful. I wish I could do better, but with the available data, I'm out of ideas.

 

So if I have a new MacPro and should I even bother trying to get build 1970 to work? I am completely new to this so I would like to know before I open up a can of worms. Thanks.

 

bitjunky - "I do have 5 sata drives in the Mac Pro..."

How do you have this set up? The Mac Pro only holds four so some third party hardware/software must be involved.....

DD

 

Actually, the Mac Pro has 6 SATA connectors on the MB, only 4 of which Apple has plugged into the drive backplane. The other two are available for use, and work fine. All that is required is to mount the additional drive in the optical bay, and then plug it into power supply (I used the one for the optional optical drive) and into one of the two available 'extra' SATA ports with a standard SATA cable.

You can see more about this setup here;
http://anandtech.com/mac/showdoc.aspx?i=2816&p=4

This setup is not that uncommon, and has been used an tested without showing any problems.

You can even use a product that allows you to mount even more drives in the optical bay, but I didn't want to add any more heat to the system...

http://www.maxupgrades.com/istore/index.cfm?fuseaction=product.display&Product_ID=158

 

Hello again,

some additional information regarding my problem.

im using SoftRAID for my software RAID (2 x 320 gb hdd as RAID 1, OS itself is at another HD), not the apple implemented method.
Got still the same problems with parallels 1970.

If i shutdown the virtual XP the Mac itself freezes or just gets a kernel panic. In the first case i can access the mac vie ssh and try to shutdown it, but the result in most cases after the following reboot is a broken raid

from my point of view it really looks like a specific problem on Mac Pro's running Parallels 1970 with Windows XP

maybe i should just use some newer beta release from parallels ?
any other ideas ?


best regards
fidel

 

maybe interesting for some other MacPro & Raid users:
--

- the 3150/3170 Betas are running great without any problems on my MacPro.
- 1970 stable keeps crashing while closing Windows XP


best regards
fidel