Jump to content

Kernel Protection Failure


Recommended Posts

Recently, upon updating Mac OS X to 10.2.3, I had some system problems. Some of them may have been caused by faulty applications, and Retrospect is possibly one of them. I reinstalled all of my system, starting from a formatted hard drive and things work fine now, except for Retrospect. Every time I quit Retrospect (version 5.0.238 with driver update 3.3.104), the crash report shown below is generated, which says that a kernel protection failure occurs. Can someone explain this?

 

 

 

[i know that I can turn off the automatic display of crash reports by the Console. Please do not suggest this solution.]

 

 

 

By the way, Dantz does not provide any kind of technical support unless you are ready to pay $69,95 per incident. As an ownder of Retrospect Workgroup who paid a fair amount of money for this product a few weeks ago, I feel this policy is very rude for customers. There should at least be an email address for customers with valid registration numbers to send their requests to support people. Other companies with products ten times cheaper than Retrospect provide this service.

 

 

 

Here is the crash report anyway. I hope that somebody from Dantz will read this.

 

 

 

Date/Time: 2002-12-27 00:12:59 -0500

 

OS Version: 10.2.3 (Build 6G30)

 

Host: Ordinateur-de-Suzanne.local.

 

 

 

Command: fix_prebinding

 

PID: 1175

 

 

 

Exception: EXC_BAD_ACCESS (0x0001)

 

Codes: KERN_PROTECTION_FAILURE (0x0002) at 0x00000054

 

 

 

Thread 0 Crashed:

 

#0 0x000152a4 in 0x152a4

 

#1 0x0000c3d8 in 0xc3d8

 

#2 0x0000c0ec in 0xc0ec

 

#3 0x0000b3ac in 0xb3ac

 

#4 0x00007b20 in 0x7b20

 

#5 0x00004250 in 0x4250

 

#6 0x0000290c in 0x290c

 

#7 0x00002414 in 0x2414

 

#8 0x00002a2c in 0x2a2c

 

#9 0x90031c44 in mach_msg_server

 

#10 0x00002c40 in 0x2c40

 

#11 0x00002f48 in 0x2f48

 

#12 0x00002188 in 0x2188

 

#13 0x00002008 in 0x2008

 

 

 

PPC Thread State:

 

srr0: 0x000152a4 srr1: 0x0000f030 vrsave: 0x00000000

 

xer: 0x00000000 lr: 0x00015254 ctr: 0x90000f00 mq: 0x00000000

 

r0: 0x000017e8 r1: 0xbffff430 r2: 0x0408040c r3: 0x00172a78

 

r4: 0x00000003 r5: 0x00000001 r6: 0x00000040 r7: 0x00032230

 

r8: 0x00001680 r9: 0x00000014 r10: 0x00032230 r11: 0x0016c428

 

r12: 0x90000f00 r13: 0x000321b8 r14: 0x00032234 r15: 0x0000003f

 

r16: 0x00001770 r17: 0x00001f80 r18: 0x00170c00 r19: 0x00000010

 

r20: 0x00032234 r21: 0x000321b8 r22: 0x00172630 r23: 0x00000030

 

r24: 0x00032230 r25: 0x01857c60 r26: 0x00032234 r27: 0x00172a48

 

r28: 0x00000033 r29: 0x00032230 r30: 0x00001680 r31: 0x00014914

 

 

Link to comment
Share on other sites

  • Replies 60
  • Created
  • Last Reply

In reply to:

Every time I quit Retrospect (version 5.0.238 with driver update 3.3.104), the crash report shown below is generated, which says that a kernel protection failure occurs. Can someone explain this?

 

 

 

Command: fix_prebinding

 

PID: 1175

 


 

 

 

The Console log is telling you that the fix_prebinding process has crashed, not Retrospect.

 

 

 

Prebinding was something that needed to be done manually (by installers or by commands, including some shareware GUI apps) in versions of OS X previous to Jaguar.

 

 

 

In OS X 10.2, Apple put in a method to automatically prebind applications, using this process.

 

 

 

With the limited information you have provided, there is no way of knowing why this process is crashing each time you quit Retrospect. Perhaps if you describe your complete hardware setup a hint might be found.

 

 

 

>By the way, Dantz does not provide any kind of technical support unless you are ready

 

>to pay $69,95 per incident. As an ownder of Retrospect Workgroup who paid a fair amount

 

>of money for this product a few weeks ago, I feel this policy is very rude for customers.

 

 

 

It is my understanding that Dantz gives some period of free tech support for new buyers of Retrospect 5, although not for upgrades.

 

 

 

Dave

Link to comment
Share on other sites

Thank you for answering my post.

 

 

 

Right now, Retrospect is the only application in my computer which causes the fix_prebinding process to crash. And I found out that the crash does not happen only upon quitting Retrospect, but that it occurs twice in a row (at about a 20 seconds interval) a few seconds after Retrospect is started.

 

 

 

Here is a description of the hardware I am using.

 

 

 

iMac DV 450 MHz with internal CD-ROM

 

768 MB RAM

 

External LaCie Firewire 56 GB hard drive (where the Retrospect catalogs are stored)

 

External LaCie Firewire CD-RW, model LTR-48125W

 

 

 

There is a connection with a Mac OS 9 iMac accessible through AppleShare and an ethernet connection via a router. This computer is a client backed up by the iMac DV with Mac OS 10.2.3 and Retrospect Workgroup.

 

 

 

Retrospect seems to be working well. Backups are done as scheduled and files can be restored, including the ones on the Mac OS 9 client. But the fact that Retrospect makes a system process crash is not reassuring.

Link to comment
Share on other sites

With more tests, I realized that the fix_prebinding crashes do not occur at all when I quit the Retrospect application, but a little after launching it. I had misread the time stamps in the log when I first got that impression.

 

 

 

Anyway, after reading the article at http://radio.weblogs.com/0100490/stories/2002/08/24/prebindingExplained.html, I figured the crash is probably occurring when the fix_prebinding process is trying to update the prebinding information in Retrospect but does not succeed. The updating process is attempted over and over by the system. Something is preventing the completion of the process, possibly a symbol that is referenced in Retrospect but that does not exist anymore in the 10.2.3 libraries...

 

 

 

In light of what I have read, I feel it is quite unlikely that there is any link between the crashes and my hardware configuration.

 

 

 

So, I will leave this to Dantz's engineers. In the mean time, I will try to update the prebinding information with Pacifist and see what happens.

Link to comment
Share on other sites

I used Pacifist to update the prebinding information for Retrospect. Below are the messages generated during this updating, as recorded by the Console. The whole log is fairly long, but it simply boils down to erroneous directory paths. For example, here is one of the messages below:

 

 

 

2002-12-27 21:48:28.851 update_prebinding[2422] update_prebinding (finding dependent libraries for /Applications/Retrospect 5.0/System/Library/Frameworks/Carbon.framework/Versions/A/Carbon): can't open file: /Applications/Retrospect 5.0/System/Library/Frameworks/Carbon.framework/Versions/A/Carbon (No such file or directory)

 

 

 

The System/Library/Frameworks/Carbon.framework/Versions/A/Carbon file exists, but not the /Applications/Retrospect 5.0/System/Library/Frameworks/Carbon.framework/Versions/A/Carbon file.

 

 

 

At the end of the Console log, we can see that Pacifist was not able to update the prebinding information any more than the fix_prebinding routine can. Except Pacifist does not crash...

 

 

 

 

 

------------Console log---------------------

 

 

 

Dec 27 21:48:28 Ordinateur-de-Suzanne authexec: executing /usr/bin/update_prebinding

 

 

 

2002-12-27 21:48:28.589 update_prebinding[2422] Start of update_prebinding

 

 

 

2002-12-27 21:48:28.595 update_prebinding[2422] Start of search for binaries in packages...

 

 

 

2002-12-27 21:48:28.595 update_prebinding[2422] Discover library dependencies (0/1 complete)

 

 

 

2002-12-27 21:48:28.635 update_prebinding[2422] Discover library dependencies (1/1 complete)

 

 

 

2002-12-27 21:48:28.636 update_prebinding[2422] Build dependency graph (0/1 complete)

 

 

 

2002-12-27 21:48:28.636 update_prebinding[2422] Start of search for binaries on disk...

 

 

 

2002-12-27 21:48:28.851 update_prebinding[2422] update_prebinding (finding dependent libraries for /Applications/Retrospect 5.0/System/Library/Frameworks/Carbon.framework/Versions/A/Carbon): can't open file: /Applications/Retrospect 5.0/System/Library/Frameworks/Carbon.framework/Versions/A/Carbon (No such file or directory)

 

 

 

2002-12-27 21:48:28.853 update_prebinding[2422] update_prebinding (finding dependent libraries for /Applications/Retrospect 5.0/System/Library/Frameworks/Cocoa.framework/Versions/A/Cocoa): can't open file: /Applications/Retrospect 5.0/System/Library/Frameworks/Cocoa.framework/Versions/A/Cocoa (No such file or directory)

 

 

 

2002-12-27 21:48:28.854 update_prebinding[2422] update_prebinding (finding dependent libraries for /Applications/Retrospect 5.0/System/Library/Frameworks/Security.framework/Versions/A/Security): can't open file: /Applications/Retrospect 5.0/System/Library/Frameworks/Security.framework/Versions/A/Security (No such file or directory)

 

 

 

2002-12-27 21:48:28.856 update_prebinding[2422] update_prebinding (finding dependent libraries for /Applications/Retrospect 5.0/usr/lib/libSystem.B.dylib): can't open file: /Applications/Retrospect 5.0/usr/lib/libSystem.B.dylib (No such file or directory)

 

 

 

2002-12-27 21:48:28.906 update_prebinding[2422] update_prebinding (finding dependent libraries for /Applications/Retrospect 5.0/System/Library/PrivateFrameworks/DiskArbitration.framework/Versions/A/DiskArbitration): can't open file: /Applications/Retrospect 5.0/System/Library/PrivateFrameworks/DiskArbitration.framework/Versions/A/DiskArbitration (No such file or directory)

 

 

 

2002-12-27 21:48:28.907 update_prebinding[2422] update_prebinding (finding dependent libraries for /Applications/Retrospect 5.0/System/Library/Frameworks/CoreFoundation.framework/Versions/A/CoreFoundation): can't open file: /Applications/Retrospect 5.0/System/Library/Frameworks/CoreFoundation.framework/Versions/A/CoreFoundation (No such file or directory)

 

 

 

2002-12-27 21:48:28.909 update_prebinding[2422] update_prebinding (finding dependent libraries for /Applications/Retrospect 5.0/System/Library/Frameworks/IOKit.framework/Versions/A/IOKit): can't open file: /Applications/Retrospect 5.0/System/Library/Frameworks/IOKit.framework/Versions/A/IOKit (No such file or directory)

 

 

 

2002-12-27 21:48:29.004 update_prebinding[2422] redo_prebinding on /Applications/Retrospect 5.0/System/Library/Frameworks/IOKit.framework/Versions/A/IOKit:can't open file: /Applications/Retrospect 5.0/System/Library/Frameworks/IOKit.framework/Versions/A/IOKit (No such file or directory).

 

 

 

2002-12-27 21:48:29.005 update_prebinding[2422] redo_prebinding on /Applications/Retrospect 5.0/System/Library/Frameworks/Security.framework/Versions/A/Security:can't open file: /Applications/Retrospect 5.0/System/Library/Frameworks/Security.framework/Versions/A/Security (No such file or directory).

 

 

 

2002-12-27 21:48:29.006 update_prebinding[2422] redo_prebinding on /Applications/Retrospect 5.0/System/Library/Frameworks/CoreFoundation.framework/Versions/A/CoreFoundation:can't open file: /Applications/Retrospect 5.0/System/Library/Frameworks/CoreFoundation.framework/Versions/A/CoreFoundation (No such file or directory).

 

 

 

2002-12-27 21:48:29.007 update_prebinding[2422] redo_prebinding on /Applications/Retrospect 5.0/usr/lib/libSystem.B.dylib:can't open file: /Applications/Retrospect 5.0/usr/lib/libSystem.B.dylib (No such file or directory).

 

 

 

2002-12-27 21:48:29.008 update_prebinding[2422] redo_prebinding on /Applications/Retrospect 5.0/System/Library/Frameworks/Carbon.framework/Versions/A/Carbon:can't open file: /Applications/Retrospect 5.0/System/Library/Frameworks/Carbon.framework/Versions/A/Carbon (No such file or directory).

 

 

 

2002-12-27 21:48:29.009 update_prebinding[2422] redo_prebinding on /Applications/Retrospect 5.0/System/Library/PrivateFrameworks/DiskArbitration.framework/Versions/A/DiskArbitration:can't open file: /Applications/Retrospect 5.0/System/Library/PrivateFrameworks/DiskArbitration.framework/Versions/A/DiskArbitration (No such file or directory).

 

 

 

2002-12-27 21:48:29.010 update_prebinding[2422] redo_prebinding on /Applications/Retrospect 5.0/System/Library/Frameworks/Cocoa.framework/Versions/A/Cocoa:can't open file: /Applications/Retrospect 5.0/System/Library/Frameworks/Cocoa.framework/Versions/A/Cocoa (No such file or directory).

 

 

 

2002-12-27 21:48:29.011 update_prebinding[2422] redo_prebinding on /Applications/Retrospect 5.0/Retrospect/Contents/MacOS/RetroRunHelper:redoing the prebinding can't be done because file has a resource fork or type/creator: /Applications/Retrospect 5.0/Retrospect/Contents/MacOS/RetroRunHelper.

 

 

 

2002-12-27 21:48:29.018 update_prebinding[2422] redo_prebinding on /Applications/Retrospect 5.0/Retrospect/Contents/MacOS/LaunchRetro:redoing the prebinding can't be done because file has a resource fork or type/creator: /Applications/Retrospect 5.0/Retrospect/Contents/MacOS/LaunchRetro.

 

 

 

2002-12-27 21:48:29.020 update_prebinding[2422] redo_prebinding on /Applications/Retrospect 5.0/Retrospect/Contents/MacOS/RetroRun:redoing the prebinding can't be done because file has a resource fork or type/creator: /Applications/Retrospect 5.0/Retrospect/Contents/MacOS/RetroRun.

 

 

 

2002-12-27 21:48:29.021 update_prebinding[2422] redo_prebinding on /Applications/Retrospect 5.0/Retrospect/Contents/MacOS/Device_Access.bundle/Contents/MacOS/loudness:redoing the prebinding can't be done because file has a resource fork or type/creator: /Applications/Retrospect 5.0/Retrospect/Contents/MacOS/Device_Access.bundle/Contents/MacOS/loudness.

 

 

 

2002-12-27 21:48:29.023 update_prebinding[2422] redo_prebinding on /Applications/Retrospect 5.0/Retrospect/Contents/MacOS/AuthenticateUser.app/Contents/MacOS/AuthenticateUser:redoing the prebinding can't be done because file has a resource fork or type/creator: /Applications/Retrospect 5.0/Retrospect/Contents/MacOS/AuthenticateUser.app/Contents/MacOS/AuthenticateUser.

 

 

 

2002-12-27 21:48:29.024 update_prebinding[2422] Update_prebinding done.

 

 

 

2002-12-27 21:48:29.025 update_prebinding[2422] 0 files successfully prebound, 12 files unsuccessfully prebound.

 

 

Link to comment
Share on other sites

In reply to:

The System/Library/Frameworks/Carbon.framework/Versions/A/Carbon file exists, but not the /Applications/Retrospect 5.0/System/Library/Frameworks/Carbon.framework/Versions/A/Carbon file.


 

 

 

I would think not.

 

 

 

The paths referenced by your errors don't make any sense to me, since the folder "Retrospect 5.0" that is installed by the Retrospect installer does not contain a System directory; I can see no reason why the prebinding routines in the OS expect to see such a path.

 

 

 

I ran Pacifist's prebinding routine on my 10.2.3 system, and while I got a few errors Retrospect did not show up as a problem. One of my errors was:

 

 

 

2002-12-28 13:54:05.879 update_prebinding[1716] redo_prebinding on /Applications/iTunes.app/Contents/MacOS/iTunes:can't open file: //System/Library/PrivateFrameworks/DiscRecording.framework/Versions/A/DiscRecording (No such file or directory).

 

 

 

This one makes some sense, as I have no CD-R drive in this machine.

 

 

 

I'd suggest deleting your /Applications/Retrospect 5.0/ directory and running the 5.0.238 installer again. The downloadable updater is a full application installer, and your serial number and scripts will be unaffected by replacing the folder.

 

 

 

Dave

 

 

Link to comment
Share on other sites

Before my last messages, I had already deleted and reinstalled Retrospect (for the second time in a few days, since my system is newly reinstalled on a formatted hard drive). I had also tried deleting Retrospect's preference folder, but the fix_prebinding crash still happened.

 

 

 

Just like you, I got the iTunes prebinding error message, since I don't have a CD-R on my computer neither. But that is one error message while there are 19 in the case of Retrospect. When I launch iTunes, it does not cause the fix_prebinding process to crash!

 

 

 

Pacifist or fix_prebinding are getting the information from the application, i.e. Retrospect. Retrospect is the one telling Pacifist that it will be needing, for example, the "Applications/Retrospect 5.0/System/Library/Frameworks/Carbon.framework/Versions/A/Carbon" file. Where else could Pacifist get this information? Other than the little iTunes glitch, no other application in my computer generates prebinding error messages. This is why Retrospect is the ONLY application which makes fix_prebinding crash, even after being reinstalled and updated to the latest version, as confirmed by the log each time Retrospect is launched.

 

 

 

The directory paths provided by Retrospect for prebinding are wrong. This is where the problem in Retrospect is. If you cannot reproduce the problem, then I suspect it may be related to the localization of Retrospect. I am using the French version. [but the directory paths should be exactly the same in French, since Mac OS X keeps the same folder names whatever the interface language. A localized "label" is given to some folders, so the user thinks it has a localized name, but the real names are the same.]

 

 

 

The problem is definitely somewhere in Retrospect. Someone from Dantz has to look into this.

Link to comment
Share on other sites

The way modern software is made, localization should not have any incidence on the behavior of the application, especially from English to French, two languages and cultures which are very similar. All localized strings should be in resource files and should not impact in any way on the code itself. The French version has the same update number and the same driver version as the US one. So, basically, everybody should be getting the same bugs.

 

 

 

It is only when you said that you could not reproduce the same crashes with Pacifist that the idea of an error in the localization process occurred to me. It is possible that the directory paths were modified by someone who does not know that, no matter what language is used, the folders in Mac OS X always have the same names, as opposed to the folders in Mac OS Classic. Therefore, the prebinding application, not being able to find the proper path, looks for the system library files in the Retrospect application folder.

 

 

 

But this is just a hypothesis. We will know for sure once someone from Dantz finally looks into this issue.

Link to comment
Share on other sites

In reply to:

when you said that you could not reproduce the same crashes with Pacifist that the idea of an error in the localization process occurred to me.


 

 

 

I downloaded the French version of 5.0.238 and did not see any problem with either the Apple process or the Pacifist application.

 

 

 

Since you deleted /Library/Preferences/Retrospect/

 

you might try deleting /Library/StartupItems/RetroRun/

 

after quitting Retrospect and killing the retrorun process. Then empty the trash and re-launch Retrospect.

 

 

 

Any difference?

 

 

 

Dave

Link to comment
Share on other sites

No difference: two fix_prebinding crashes at about a 20 seconds interval, after launching Retrospect.

 

 

 

It is possible that your test with the French version did not generate the same problem because the cache containing the prebinding information for Retrospect has already been created in your computer. The French version can most likely use the same cache.

 

 

 

There is nothing more a user can do to eleminate all potential sources of problems than format his hard drive and reinstall his whole system, which is what I have done. I also deleted completely Retrospect and all associated files and reinstalled it twice. There is nothing more I can do. I repeat: my 10.2.3 system works fine, other than Retrospect causing a kernel protection failure.

 

 

 

 

Link to comment
Share on other sites

Dantz labs have not seen this issue and have not been able to reproduce this issue in house. We haven't recieved any other reports of this issue, which may indicate that the problem is specific to your environment.

 

 

 

Uninstall Retrospect and create a new user on this machine. Log in as the new user and install Retrospect. Do you see the same issues?

Link to comment
Share on other sites

I did exactly what you suggested: 1) deleted Retrospect as well as the Retrospect preference folder and Retrorun; 2) created a new user; 3) logged in as the new user; 4) installed the latest version available of Retrospect (5.0.238 with the 3.3.104 driver).

 

 

 

Guess what? The problem still occurs precisely as it occurred before: two crashes of the fix_prebinding process at a twenty second interval.

 

 

 

Instead of asking me to do your job, you should read carefully every one of my posts and do your testing more thoroughly.

 

 

 

First, this is not an issue people will spontaneously tell you about, because it does not affect Retrospect directly. It affects the OS. More precisely, Retrospect makes a system process crash because the paths used for prebinding are erroneous.

 

 

 

Second, I reinstalled all of my system ten days ago and I have deleted and reinstalled Retrospect four times since then. Don't you think it's about time you start looking for answers on your side instead of in my computer?

 

 

 

Third, I suspect Retrospect to have caused the problems I had when I updated my system to 10.2.3 (not when I reinstalled everything from scratch, but the first time I tried updating). Retrospect was running at the time of the first update attempt. If Retrospect makes a system process crash, and no other application does, it is reasonable to think that Retrospect could have caused the problem.

 

 

 

Here is what you have to do in order to reproduce the problem:

 

 

 

1) Find a computer with Mac OS 1.2.3 on which Retrospect was never installed, preferably an iMac DV 450 MHz or similar machine; if you do not have one, then format the hard drive on a computer (after all, I did it and I am not paid for it!).

 

 

 

2) Reinstall 9.2.2, then Jaguar as well as the latest system update (10.2.3).

 

 

 

3) Install the French version of Retrospect.

 

 

 

The process is important, because if you start from a computer on which Retrospect was previously installed and update to 10.2.3 without reformatting the drive, the prebinding information is still cached in system files and does not have to be updated, therefore the fix_prebinding process is not involved.

 

 

 

I think that I have given Dantz plenty of thoroughly verified information in my posts. It's your turn to act now.

Link to comment
Share on other sites

What a fascinating, rude post.

 

 

 

>Here is what you have to do in order to reproduce the problem:

 

 

 

Have _you_ reproduced the problem this way?

 

 

 

I know that these are the steps you remember, and that information is _very_ important to know (and is the sort of detail that is sorely lacking on most people's posts).

 

 

 

But have you done this again, and found the same result?

 

 

 

Not your job, I know.

 

 

 

1) Find a computer with Mac OS 1.2.3 on which Retrospect was never installed, preferably an iMac DV 450 MHz or similar machine; if you do not have one, then format the hard drive on a computer (after all, I did it and I am not paid for it!).

 

 

 

Done.

 

 

 

2) Reinstall 9.2.2, then Jaguar as well as the latest system update (10.2.3).

 

 

 

9.2.2? What does this have to do with this, do you suppose?

 

I have an iMac 450 with 5 partitions. No problem to wipe one with Disk Utility.

 

I Installed 10.2 and then Apple's Combo 10.2.3 updater. I don't know where my OS 9 CD is.

 

 

 

3) Install the French version of Retrospect.

 

 

 

Done.

 

 

 

Launching Retrospect French did not crash any processes on this machine.

 

 

 

I suspect Retrospect to have caused the problems I had when I updated my system to 10.2.3 (not when I reinstalled everything from scratch, but the first time I tried updating). Retrospect was running at the time of the first update attempt.

 

 

 

You left this step out in your original list. So perhaps the test steps should be:

 

 

 

1- Install 10.2.x on fresh, formatted volume (what OS version were you running before you updated?)

 

2- Install Retrospect 5.0.238 French

 

3- Launch Retrospect and enter your serial number

 

4- With Retrospect still running, run the appropriate 10.2.3 updater

 

5- Restart Mac OS from the installer and launch Retrospect.

 

 

 

If you were a regular reader of this Forum you would know that there are many very careful, bright contributors. If an OS X process is crashing everytime the program launches you can be sure it would have been reported here more often then this one time.

 

 

 

In none of your messages have you been clear about what you have done. You talk about having installed everything from scratch, then you bring up suspicions about something having happened before reformatting your drive. After 8 posts, I remain confused as to your exact steps.

 

 

 

Using your steps 1-3 causes no problem for me (although I'm open to doing it again with the OS 9 component if I were even the least bit suspicious that it might make a difference).

 

 

 

My steps 1-5 sound like fun; perhaps I'll give them a shot next year...

 

 

 

Dave

Link to comment
Share on other sites

In reply to:

My steps 1-5 sound like fun; perhaps I'll give them a shot next year...


 

 

 

 

 

Here's what I did to try and reproduce this error:

 

 

 

1- Install 10.2.2 on fresh, formatted volume

 

2- Install Retrospect 5.0.238 French

 

3- Launch Retrospect and enter my serial number

 

4- With Retrospect still running, run the appropriate 10.2.3 updater

 

5- Restart Mac OS from the installer and launch Retrospect.

 

 

 

After this I was able to launch Retrospect with no problems, and run the prebinding routines with no reports of errors from Retrospect files.

 

 

 

Since Dantz says that they cannot reproduce this, and no one else here (or on the RetroTALK email list) has ever reported a similar experience, your specific OS X install becomes suspect.

 

 

 

No one else is going to be able to learn what went wrong for you without your continued input. What _exactly_ did you do when you rebuilt your system?

 

 

 

Dave

 

 

 

Dave

 

 

Link to comment
Share on other sites

My rudeness is a consequence of Dantz's frustrating and unfair support policy. Unless you work for Dantz, you should not feel targeted in any way. My last message was a reply to a message from Dantz's forum moderator.

 

 

 

For the record, here are the steps I followed, which I believe to be in accordance with Apple's and Dantz's guidelines.

 

 

 

1. Booted from the OS 9 CD.

 

2. Formatted the hard drive with only one partition.

 

3. Installed 9.2.2 (because I have only one partition and this step needs to be done first in this case, unless you don't need the Classic environment).

 

4. Installed Jaguar from the CD.

 

5. Installed the 10.2.3 update from the Combo update disk image downloaded from Apple's Web site.

 

6. Installed Retrospect 5.0.238 and the latest driver.

 

7. Launched Retrospect and entered my serial number.

 

 

 

Since then, no matter what I do, the very same fix_prebinding process crashes occur consistently every time I launch Retrospect. No other application on my computer causes the fix_prebinding process to crash. And, to top it all, Pacifist confirms that the prebinding of the Retrospect application is impossible, due to erroneous directory paths.

 

 

 

As far as your testing is concerned, you seem to be using a very different configuration from mine, namely a partitioned hard drive. In my view, this is not a reliable way to reproduce a bug.

 

 

 

One other thing, concerning your last test: At the time of my 10.2.3 update which ended up in a major system failure, I did not have the latest version of Retrospect. But this is irrelevant, since the bug we are talking about now is the one which definitely causes a fix_prebinding crash upon launching Retrospect, in my computer. My anterior problem with the 10.2.3 update is what made me look deeper into Retrospect's behavior, but is not the problem I am trying to solve now.

 

 

 

I will be testing Retrospect on an iMac G4 with Mac OS 10.2.3 soon. If I can reproduce the problem on a different machine, I will most certainly let every one know.

Link to comment
Share on other sites

My rudeness is a consequence of Dantz's frustrating and unfair support policy. Unless you work for Dantz, you should not feel targeted in any way. My last message was a reply to a message from Dantz's forum moderator.

 

 

 

One need not be the target of rudeness to feel its sting. AmyC wrote to try and help. She was cordial and direct (as she is in all her posts here). She should be treated with the same respect that any other person is due.

 

 

 

For the record, here are the steps I followed, which I believe to be in accordance with Apple's and Dantz's guidelines.

 

 

 

{steps snipped}

 

 

 

As far as your testing is concerned, you seem to be using a very different configuration from mine, namely a partitioned hard drive. In my view, this is not a reliable way to reproduce a bug.

 

 

 

I don't believe that simply having other volumes in /Volumes/ would effect such mangled behavior, but if it _does_ then the solution to your problem would be to partition your drive!

 

 

 

I did all my testing with the other partitions unmounted. The're gone from the Desktop, gone from the /Volumes/ directory. Installers and unix processes to not know that they exist.

 

 

 

I will be testing Retrospect on an iMac G4 with Mac OS 10.2.3 soon. If I can reproduce the problem on a different machine, I will most certainly let every one know.

 

 

 

I hope you'll report your experiences no matter what the results.

 

 

 

Dave

Link to comment
Share on other sites

For sure I will post in this forum the results of my tests on the iMac G4 (which is not the same machine), no matter what the results. If I reproduce the problem, it will be one more proof that people in Dantz labs are not even trying. If I don't reproduce it, it will just mean I must find another machine similar to mine for testing.

 

 

 

But in the mean time, let me give you a demonstration of the stability and reliability of my 10.2.3 system as it is now. After booting from a Jaguar CD and doing a weekly routine check of the hard drive which showed only one minor repair had to be done (and no error on the second check), I booted again from the HD and launched a total of 70 processes, all running at the same time, including Mail, IE, iCal, Address Book, Backup, iTunes, iMovie, AppleWorks, Word, Excel, PowerPoint, Entourage, Virtual PC and others. These processes used up all 768 MB of RAM plus a large chunk of virtual memory. And to top it all, I played a couple of Quicktime movies at the same time. The Console reported no error message. No crash. No application quit unexpectedly. No complaint. NOTHING!

 

 

 

Then I launched Retrospect. Guess what? The same thing as always occurred: two fix_prebinding crashes.

 

 

 

Retrospect is the only application, after having run 70 processes all at the same time, which caused anything abnormal.

 

 

 

I am dead sure there is a bug in Retrospect, and I will continue writing in this forum until somebody at Dantz wakes up. They are the ones who are rude, taking money from people and not even providing the needed support.

 

 

 

 

Link to comment
Share on other sites

In reply to:

I am dead sure there is a bug in Retrospect, and I will continue writing in this forum until somebody at Dantz wakes up.


 

 

 

You should absolutely keep writing. As I mentioned, we've been unable, still, to reproduce the issue you are experiencing with your exact same setup. Any information you can provide is useful. At this time, we can't reproduce it with the steps you provided in the same configuration, we aren't receiving reports of other users with other configurations seeing this issue, and we haven't yet seen it in the lab.

 

 

 

Without being able to reproduce the issue, the ability to address the problem becomes that much more difficult. The problem may be in the software or the problem may be in your environment. Once we can reproduce the issue in a controlled lab enviroment, we can make that determination.

 

 

 

In reply to:

They are the ones who are rude, taking money from people and not even providing the needed support.


Should you have opened a support incident that ultimately proved to be a bug in the software, then your incident charge would be promptly refunded, as is policy. With no calls to technical support directly, and one valid request on this forum regarding a step to try, I fail to see how anyone was rude to you. I'm sorry you feel that way. This forum is not an official means of contacting Dantz Technical Support. The quickest coarse of action would be to open an incident with Technical Support so we can work with you directly. As mentioned, if the problem ends up being a bug in Retrospect, your charges will be refunded. You can also continue using the forum resources if you choose.

Link to comment
Share on other sites

Open an incident for 69,95 USD? No thanks. I paid 293,80 EUR in October for Retrospect Workgroup. Over the past two years, I have paid more than 500 USD for Dantz software. That is enough in my view to get a proper response when I need help. So, when we are talking about rudeness, this is what I mean. And the person at Dantz's support I talked to on the phone told me that I was far from being the only one complaining about this policy. Eventually, Dantz might say the bug is in Apple's system software or keep pretending that they can't reproduce it. And I will have to pay the bill.

 

 

 

If the fee was something like 9,95 USD per incident, for example, it would be enough to deter people from phoning all the time without even looking for the solution in other resources, and such a fee would be respectful of loyal users. The registered user could be allowed three incidents per year at this rate, then have to pay a higher fee for following incidents. I don't think this would cause Dantz to go bankrupt.

 

 

 

By the way, the problem could be the files I downloaded from Dantz's site, which is a real mess when it comes to versions: broken links, wrong version numbers, etc. For example, the French updates page lists the latest Client update as "Retrospect Client 5.0.540". But when I download the file, I get an archive called "retrospect_50238_clients_fr.sit". Then, when I decompress it, the folder is called "Retrospect Client 5.0". And when I dig into the folder and subfolders, I find an updater called "Mac_Client_Update_FR_5_0.rcu". So, where is "5.0.540" now? Another example? When I download the French 5.0.238 updater, I get a file called "retrospect_50238_workgroup_fr.s". The system tries to open it with Word! You must manually change the extension to "sit" in order to decompress the files. Any experienced Mac user knows this, but it certainly reflects a certain carelessness on the part of Dantz. An FTP site including all US and international versions, in properly named files and folders, would be an efficient way to distribute updates.

 

 

 

Step one of proper technical support would have been to make sure that I have an uncorrupted update file for the 5.0.238 French version and the latest version of the driver. I don't even know if, for 69,95 USD, I would get the proper files. And if I did, and it solved the problem, this will most certainly not be considered a bug. I would have lost 69,95 USD due to Dantz's poor management of its Web site.

 

 

 

Furthermore, I am skeptical about the seriousness and completeness of your so-called tests. And I have good reasons to be, given your erratic answer recently to a user inquiring about Retrospect's compatibility with Mac OS 10.2.3. You were very quick to answer it was compatible, until the user said he had got conflicting information on the phone. There seems to be a communication gap between you and Dantz's lab.

 

 

 

Considering that the fix_prebinding crash problem is very precise and consistently occurs the same way, it should be quite simple for a programmer to check the code and determine if there is a possibility that the fix_prebinding process gets the wrong directory path from Retrospect, in Mac OS 10.2.3. Judging by the posts in various forums, there are quite a few file system problems related to various applications, which were not ready for 10.2.3 (for example, Virtual PC 6). Retrospect could be another one of them, although probably to a lesser degree.

 

 

 

We all agree that Retrospect is a fine product, in general, but I know I am not the only user who is sadly disappointed with Dantz's technical support policy.

Link to comment
Share on other sites

I am just back from testing Retrospect on an iMac G4. I was able to reproduce the fix_prebinding crash exactly as it occurs on my iMac DV. Here is the procedure I followed.

 

 

 

The iMac G4 was bought new about a month ago. The Mac OS 10.2.3 system update was done about ten days ago through Software Update. The update went smoothly, according to the owner of the computer, which has Classic (9.2.2) installed. Retrospect had never been installed before on this computer.

 

 

 

1) I checked the crash logs for the computer. Only two crashes had been reported: one IE crash and one MS-Word crash. No system crash.

 

 

 

2) I installed Retrospect Workgroup 5.0.238 French with driver update 3.3.104.

 

 

 

3) I launched the Console and set it to report crashes automatically.

 

 

 

4) I launched Retrospect and entered my licence number.

 

 

 

5) I waited for a minute without doing any change in Retrospect's default settings.

 

 

 

---- No fix_prebinding crash occurred at this point, which was expected since another process than fix_prebinding does the prebinding when an application is launched for the first time. This process is called update_prebinding. Every time I reinstalled Retrospect from scratch on my own computer, I never observed the fix_prebinding crash on the first launch. [And anybody who is serious about trying to reproduce a bug would certainly do more than one attempt anyway.] ----

 

 

 

6) I quit Retrospect and launched it again.

 

 

 

7) After a few seconds, the Console reported a fix_prebinding crash.

 

 

 

8) I relaunched Retrospect twice and observed the fix_prebinding crash each time.

 

 

 

---- Only one fix_prebinding crash occurred on each launch, instead of two, which could be explained by the absence of programmed backups. But the fact is that at least one fix_prebinding crash occurs each time Retrospect is launched. ----

 

 

 

9) I quit Retrospect.

 

 

 

10) I quit the Console.

 

 

 

11) I deleted all Retrospect files from the iMac G4.

 

 

 

Bug confirmed. Do your homework, Dantz. Am I clear enough?

Link to comment
Share on other sites

In reply to:

And I have good reasons to be, given your erratic answer recently to a user inquiring about Retrospect's compatibility with Mac OS 10.2.3. You were very quick to answer it was compatible, until the user said he had got conflicting information on the phone. There seems to be a communication gap between you and Dantz's lab.


 

 

 

Please note that you are referring to a response by the Customer Service team - a non-technical department - who gave out misinformation that was immediately corrected.

 

 

 

I understand that you are frustrated and that you are experiencing the issue that you are outlining in your posts. We will continue testing. Information will be posted to this forum when it becomes available.

 

 

 

Amy

 

Dantz Technical Support

Link to comment
Share on other sites

An FTP site including all US and international versions, in properly named files and folders, would be an efficient way to distribute updates.

 

 

 

ftp://ftp.dantz.com/pub/fr/workgroup_mac_50_fr.sit

 

 

 

Pretty efficient ftp path, IMHO, that downloads a StuffIt archive named "Retrospect 5.0.238" and installs the correct French version.

 

 

 

The download link referenced on the French web site:

 

http://www.dantz.com/fr/support/updates.html

 

 

 

is:

 

 

 

ftp://ftp.dantz.com/pub/updates/fr/retrospect_50238_workgroup_fr.sit

 

 

 

Which is a perfectly valid URL. Unfortunately, Microsoft Internet Explorer is unable to name the file properly (all of the Microsoft programs for OS X are limited to 31 character file names). This is not a failing of Dantz (since they appear to be following their existing naming convention), it is a failing of the web browser you are using. Any experienced Mac user would know this.

 

 

 

I am skeptical about the seriousness and completeness of your so-called tests. And I have good reasons to be, given your erratic answer recently to a user inquiring about Retrospect's compatibility with Mac OS 10.2.3. You were very quick to answer it was compatible, until the user said he had got conflicting information on the phone.

 

 

 

Erratic? What posts are you reading?

 

 

 

In response to a query regarding the newly released OS update, AmyC wrote:

 

"Our preliminary testing has not exposed any incompatibilities between 10.2.3 and Retrospect. "

 

 

 

Later, when a poster informed her that her Customer Service department was giving callers different information, AmyC apologized for the confusion and reported that Customer Service had been given more up-to-date information. She went on to reiterate that her testing had shown no problems (she never, ever stated that there were no problems to be found, only that none had been found in the few days that the final version of the OS update had been available).

 

 

 

The next post was from you, where you assured readers that there was a problem with 10.2.3, giving as evidence the fact that you were having a problem with it.

 

 

 

Yet so far no one else has reported the problem here, and according to AmyC no one has called tech support about it, and that her testing has not been able to reproduce it. What is erratic about her saying that?

 

 

 

You're skeptical about her tests because they do not produce the results that you're getting. Yet each of your posts has included different information about what you did, making it difficult for anyone to attempt to reproduce your environment.

 

 

 

It's obvious that _you_ are having an issue. It's obvious that it happens each time you launch Retrospect (no matter how many other processes you launch first). No one ever doubted it. Now you suggest that you might have a corrupt download (yet another fact you left out of earlier posts) which might be the cause.

 

 

 

Today you report seeing the same behavior on another computer. Certainly interesting information, and probably helpful to Dantz as they continue whatever testing they're doing. But what about the source software you're using? Did you use your same suspicious download when you tested on the G4, too? If not, what was the source of your test software? And what about the driver update? Where did you get that? Did you try without that RDU?

 

 

 

I believe that the really important question about all this is, why all the passion? What do you expect from Dantz? If AmyC is eventually able to reproduce this by finding whatever specialized condition is necessary to bring it forth, will you expect an immediate fix from them? Considering that it's obviously not a wide-spread issue, and although it's annoying, it does not effect the behavior of backups or restores, what will satisfy you?

 

 

 

Are you going on Microsoft's support sites and complaining as loudly that the web browser they wrote is unable to properly download a valid file?

 

 

 

Dave

 

 

 

 

Link to comment
Share on other sites

Here are the steps that I did this morning to try and see this:

 

 

 

iMac Slot Loading 600 Mhz (Graphite) 256Mb CD-RW

 

 

 

1. Boot from OS 9.1 CD

 

- Initialize internal HD using Disk Setup (single volume)

 

- Install OS 9.1 using default settings

 

- Run 9.2.1 update (from server)

 

- Restart into 9.2.1; complete Setup Assistant

 

- Update to 9.2.2 from Software Update control panel

 

- Restart and run Software Update again

 

- Update AppleScript, Authoring Tools and CarbonLib

 

- Insert OS X 10.2 CD and launch installer; restart at the prompt

 

2. Install OS X 10.2

 

- Run installer using default settings

 

- Have breakfast

 

- Step through Setup Assistant

 

- Configure System Preferences:

 

-- Date Time; configure for Network Time

 

-- Sharing; enable Remote Login

 

- Open Console.app and set preferences to display crash logs

 

- Update to 10.2.3 6G30 using Software Update

 

- Restart

 

- Configure Classic preference pane; confirm System Folder and start/stop Classic (install system components when prompted)(intended to do this before the 10.2.3 update, but oh well)

 

3. Download software

 

- Use Internet Explorer to download Retrospect 5.0.238 FR from:

 

ftp://ftp.dantz.com/pub/fr/workgroup_mac_50_fr.sit

 

- Use Internet Explorer to download Retrospect Device Update 3.3.104 from

 

ftp://ftp.dantz.com/pub/updates/rdu33.sit

 

4. Run the French Retrospect Installer

 

- Place French RDU file in Retrospect 5.0 folder (after trashing existing file)

 

5. Launch Terminal and type "top -u" (to watch processes as they start)

 

6. Launch Retrospect and authenticate. Leave default checkbox to always authenticate.

 

7. Enter valid Workgroup code

 

8. Quit Retrospect

 

9. Launch Retrospect

 

Repeat steps 8 & 9, waiting at least 5 minutes before quiting.

 

10. Restart. Repeat steps 8 & 9

 

11. Start Classic. Repeat steps 8 & 9

 

12. Disable preference to "always authenticate" Retrospect. Repeat steps 8 & 9

 

 

 

I saw no crashes in the console, no files were written to /Library/Logs/CrashReporter/

 

I never saw either update_prebinding or fix_prebinding try and run.

 

 

 

I used the Terminal to run the 'update_prebinding' process (which is what Pacifist does; type: sudo update_prebinding -root /) and had no errors on any Retrospect files.

 

 

 

I also tried the -verbose, -debug and -force flags on the update_prebinding command with no errors (for Retrospect) reported.

 

 

 

I tried running some Apple Installer packages (StuffIt 7 and QuickTime 6) and restarted.

 

 

 

I have no idea what might be different between my test machines (4 so far) and the two machines where you have seen a problem

 

 

 

Dave

Link to comment
Share on other sites

In reply to: "I saw no crashes in the console, no files were written to /Library/Logs/CrashReporter/

 

I never saw either update_prebinding or fix_prebinding try and run."

 

 

 

Even if you probably know it, I thought I would mention that the fix_prebinding crash reports should be in the root Library, not the user Library. Also, to be really sure, you could do a file search on the whole hard drive for "fix_prebinding".

 

 

 

The one thing my iMac DV and the iMac G4 have in common is that they both have French as the Mac OS interface language... now, don't start reinstalling Mac OS in French. But, I am starting to think that it could be a language specific Mac OS bug. At least one other bug related to international versions of Mac OS X was uncovered in 10.2.3. It had to do with iTunes' version number, which had not been updated in languages other than English and caused Software Update to constantly want to update iTunes, even just after updating it. The bug was reported on MacFixit with a workaround.

 

 

 

But why would Retrospect be the only application to cause the crash?

 

 

 

I will post a message in Apple's Mac OS X forum to see if people using other interface languages than English observe fix_prebinding crashes using Retrospect or other applications.

 

 

 

In the mean time, someone at Dantz [not you, Dave] could be helpful for a change and try Retrospect Workgroup French on Mac OS 10.2.3 with the French interface.

 

 

 

And, by the way, even if Microsoft makes poor software, it happens that Apple has chosen IE as the default browser on all of its computers sold for the past 4 or 5 years. Therefore, if I was Dantz, I would stick with 31-character file names (which is not that hard to do) in FTP and other download sites.

Link to comment
Share on other sites

Archived

This topic is now archived and is closed to further replies.


×
×
  • Create New...