Leopard Install Problems -Essentials Pkg - RAM issues

Okay, let me put it this way.

I have 30+ years of experience with personal computers, including my share of programming bugs. I have seen countless null and bogus pointer dereferences that behave erratically just like this one. I am also aware of large software development projects with an astonishing number of unfixed NULL pointer dereferences. It is a very common bug in C/C++.

So maybe I should have said that while my analysis is not absolutely certain, I think it is very highly likely given the observed facts. Apple simply didn't sufficiently test the Leopard distribution on the G5.

If you have a better explanation, I'd like to hear it. And I'd be very interested in your explanation of how RAM that has worked fine on a machine for years can suddenly fail when a new operating system is installed.

KA9Q wrote:
Okay, let me put it this way.
If you have a better explanation, I'd like to hear it. And I'd be very interested in your explanation of how RAM that has worked fine on a machine for years can suddenly fail when a new operating system is installed.

Your surmise that the RAM failed suddenly is in error. Leopard simply detects the marginal RAM and does a more stingent test on RAM than did Tiger. Leopard runs faster and bigger (full 64-bit) so it is much more demanding of RAM.
Let's say some RAM does not meet Apple's spec, but it's close enough that TIger runs OK.
Now, a few years later, we retest the RAM using Leopard's test which tests right to Apple's spec for that machine. Some marginal RAM will not pass. No, the RAM has not "failed" but is simply not good enough.
If you reinstall Tiger, that RAM will work fine.
All Leopard is doing is verifying if the RAM meets the original spec. Failing Leopard's test does not make the RAM bad, the RAM hasn't changed.
I am sure you are aware that RAM degrades over time, like most everything else. Tiger accepts poorer RAM and does not complain. Apple wants to be sure there are no RAM-related problems wiht Leopard, so the make sure the system RAM meets their spec. Obviously some RAM does not. The original Apple supplied RAM works fine.

I probably have more experience than you in the electronics fields, I have several patents, etc., but who cares? i have programmed IBM 360s, used Mark Williams Coherent, was one of the original alpha/beta testers for the first MSN. I also hold an original 2-letter call. I program in almost every language known to man 🙂 Forth anyone?

I installed Leopard on my dual G5. It installed perfectly the first time and runs well.
All your data are from a few G5s that have had problems. Out of 2million plus copies of Leopard, I am sure many went on G5s, with zero problems, or we would be hearing about them.

You are too quick to knock Apple and to quick to defend what is probably questionable RAM. Again, I fail to see your logic. You seem to be hung up on a C/C++ bug. I think your analysis is highly unlikely because you have no basis whatsoever. If you were right, there would have been an immediate patch for it.

If replacing the RAM solves the problem, how did that fix your sloppy coding bug?
If Leopard runs with Apple-supplied RAM and gives kernel panics with other RAM, you really think that's a code bug? I wouldn't want you to work for me if you really believe that.

No matter, in the end, we will know what the problem is/was. Leopard is humming away on my MBP and G5 without a burp.
73, OT
Enjoy

I don't understand why you're so offended, I simply offered what I think is a far more likely explanation than failed RAM.

With all due respect, your statement that Leopard is more "demanding" of RAM than Tiger because it runs "faster" or in 64-bit mode is quite simply nonsensical. The system hasn't even been installed yet! If a memory tester is properly written it will report bad memory if it is in fact bad. Memory testing is in fact quite hard to do well because it is obviously out of the question to test all possible states. Furthermore, if the memory tester itself is buggy it could report bad memory when it is in fact good.

Digital electronics, properly designed and operated, can operate without "degradation" for a very long time. It certainly doesn't suddenly "degrade" when the software changes. All of the RAM in my system was made to Apple specs and it has functioned perfectly over its entire lifetime. I refuse, as should every Apple user, to fall prey to Apple's FUD (fear, uncertainty and doubt) about third party RAM to scare people into buying their own overpriced RAM.

By the way, I tried removing all my third party RAM and using only the original RAM that came from Apple. The Leopard install fails at the same point. So much for your theory, eh?

I can only think of one possible way that new software change could reveal (note - not cause) a memory problem, and that is if the earlier software never exercised all of the RAM so it never hit the bad region of memory. However, this is extremely unlikely on any UNIX-like operating system, including OSX. For example, Linux users are often recommended, when building a new system, to build a kernel as an first test. This is a sufficiently demanding operation that it is almost certain to find any memory problems. Darwin, the UNIX-based OS under Mac OSX, is very similar to Linux in how it allocates and uses RAM, so the probability that bad RAM would go undetected in a system like mine that has been heavily used daily for four years for every kind of application, program development, etc, yet be immediately detected when I install Leopard is vanishingly small.

Darwin, the UNIX-based OS under Mac OSX, uses RAM for disk buffering as well as program execution, and it will use as much RAM as is on the system for this purpose. If there is faulty RAM, it will find it. Quickly.

If you want to trade credentials, I have two EE degrees from major universities and have worked in the computer and telecommunications industry for my entire career. I've been a ham since 1971, a commercial radio operator since 1973, and I was one of the original contributors to amateur packet radio and the Internet. However, that's all irrelevant, as are your credentials.

You seem to take umbrage at the notion that Apple might make an extremely common programming error in a very large and complex (and brand new) software package. This strongly suggests to me that you have little or no experience with large software development projects. No one is immune to such bugs, not me, not you and not Apple. As for the immediate patch, I point out that this bug is on the distribution DVD, and it prevents the OS from being installed. I am installing on virgin drives as I'm taking the opportunity to put my previous system disks on the shelf for safekeeping and install on larger drives. Please tell me how I am supposed to patch my DVD.

The mere fact that so many people are reporting the very same problem with Leopard -- but not Tiger -- is even more evidence that the problem is a software bug. I have already explained exactly how invalid pointers can cause sporadic, difficult-to-reproduce faults, and even how changing RAM might seem to make it go away. I have seen exactly this kind of bug many times over my career, and I am not ashamed to say that I've made it myself. Rather than repeat my explanation I suggest that you go back and read it.

There is another possible explanation for the failure of the Leopard install on the G5 that is very similar to my first. It is also an invalid pointer software bug. But rather than depending on the contents of uninitialized RAM, it depends on the effects of the memory randomization that has been added as a security feature.

One of the interesting features in Leopard, which is also being deployed in other operating systems such as Linux, is memory (or library) randomization, the random placement of program elements in memory. The idea is to make things more difficult for malware writers who exploit buffer overruns, basically bugs in programs where program inputs are not properly checked. Because the layout of a program can change from one execution to the next, the effects of an invalid pointer could also change from one execution to the next. That Leopard seems to install after RAM is removed or changed might simply be a totally random coincidence.

I note that Apple would not be doing library randomization in the first place if they were confident that they wrote software without bugs. You really ought to get up to speed on large software development projects. Your eyebrows may be raised by the fact that there can be thousands of unfixed bugs in any given piece of software, especially proprietary closed-source software like OSX where the world at large is not given the opportunity to peruse the source code looking for such bugs. Fortunately, most such bugs do not manifest themselves most of the time or the world of computers would come to a screeching halt. At the same time, that makes such bugs rather difficult to find and fix. One interesting side effect of library randomization may be to make invalid pointer bugs more obvious and therefore easier to detect and fix, and that will be a good thing.

Oh, yet ANOTHER proof that the install problem cannot possibly be bad RAM: on a different set of hard drives, on which I've been running Tiger on my G5 for many years, I upgraded to Leopard using a separate Leopard update disk (one that I got for my new Mac Pro) that worked just fine with all 3.5 GB of existing RAM. I did have the problem with my Administrator account being downgraded to Standard, but the Tiger->Leopard upgrade otherwise went fine. And Leopard itself ran fine on my G5 with all its RAM.

I only had the problem when installing from the Install DVD onto a pair of virgin 1TB drives. I was doing that because I wanted to put my 500GB drives on the shelf as backups and use a new set of 1TB drives to get the extra space. This switch to a new major version of Mac OSX seemed like as good a time to do that as any.

So I think it quite clear that all the RAM on my G5 is just fine. The Leopard installation failure is unquestionably a bug in the installation software. Since this is a fatal problem that keeps the OS from being installed at all on the G5, Apple is going to have to fix this with a new DVD if we are to be able to install Leopard on virgin disks with no operating system at all.

Perhaps they can let us download some sort of bootstrap operating system that we can burn onto a CD and boot ahead of the install DVD so they don't have to send us new install DVDs, but that seems to be getting rather complicated.

im having the same problem with a friends macbook pro, he had 2 partitions one with windows xp and one with mac os. I repartitioned the disk and made 1 single partition. And after 20 minutes of installing I get the same error. Shoukd I remove the ram drive?

i have the same problem. i have to RAM sticks of 1GB each that my father brought me from china. should i try removing one stick and then try? i cant use the original apple RAM because i threw them away

I don't think we have much alternative but to wait for Apple to come out with a new DVD to replace this botched abortion of an OSX release. I can't believe how bad this one is; they must not have tested Leopard on G5s at all.

It's not just a G5 issue. I have the same issue here with an Intel (2.16GHz) iMac. 2-3 attempts with no result. Jotted the DVD into an old eMac - runs just fine. No RAM or anything could explain such an odd behaviour. Called Apple Support today, which ***** this side of the Atlantic - 15mins wait time to get through to a guy working in home office being connected to 2nd lvl support by chat!! All I said, he translated to english and then forwarded to some guy in 2nd level. Spent 25mins on the phone with NO RESULT other than that they send me a new DVD. Gosh, what has become of Apple??

cheers

Torsten

The probality of bad RAM tanking an install is in my opinion quite small as all mac OS's perform a RAM check during the bootstrap process which goes like this (borrowed from OS X Daily):
You turn on your Mac, and this is what happens:
* Power is turned on.
* OF or EFI code is executed.
* Hardware information is collected and hardware is initialized.
* Something (usually the OS, but also things like the Apple Hardware Test, etc.) is selected to boot. The user may be prompted to select what to boot.
* Control passes to /System/Library/CoreServices/BootX, the boot loader. BootX loads the kernel and also draws the OS badges, if any.

So... if RAM fails the memory test, your machine should at least tone to tell you... After several instals in our Authorised Apple Service facility, and three on my own Dual G5 towers, I wrote some guidelines which seem to work for me...
PLEASE REMEMBER: BACK UP YOUR DATA FIRST!
If you do not own an external Hard Drive, you may want to consider purchasing one for your backup before installing-Also you’ll need an external Hard Drive of somewhat larger capacity than your internal Hard Drive to run Leopard’s Time Machine” Automatic Backup function. If you’re going to install on mirrored pairs, we’ve found there is one way...it is not particulary quick & easy and you should
allow several hours to do it - but if you don’t want to regret the install, we strongly suggest this method as being the absolute safest way of doing this:

#1- BACK UP YOUR SYSTEM & ALL DATA by Carbon Copy Cloner- preferably to a mirror pair drive box if available-if not use your external FW drive. (CCC version three is on the web.) After a successful Clone, Start the install and do a “NEW INSTALL” (it's a good time to low level format your HD’s if you want a totally sanitary install- then you will need to remake a RAID mirrored pair
using the Leopard Disk Utility) After the install is complete, return to the Install DVD and install the EXTRAS.
Restart your computer, create a ROOT account, log out, log in as ROOT and launch the “data Migration” Assistant. It will ask you to connect the External Mirrored
Drive (or single drive if you’re feeling lucky) as this will be the source for your saved data. Select "External Drive” and begn. This may take several hours depending on the quantity of data. I have used this process on three G5 Dual towers with RAID Mirrors with good success.

It's the RAM...at least, it was for me.

Bought the family pack of Leopard. Installed (upgraded) fine on MacBook Pro Core 2 Duo, MacPro, and 2 iMacs. Upgraded the RAM in my MacBook Pro from 2 sticks of 512 to 1 of 1GB (Apple) and 1 of 2GB (3rd party, OWC).

MacBook Pro started acting wonky, so I decided to do a clean install of Leopard. Hung up at the Essential Package error message multiple times. Apple Care was nice, but useless. This forum gave me the idea to pull the 3rd party RAM, which I did. Did an Erase and Install using just the 1GB stick. Worked fine, first time. I've reinstalled the 2GB stick and the computer seems to be running fine. I'll see how long that lasts, but if it crashes the first thing I'm going to do is pull the 2GB stick.

This doesn't speak to the dual layer drive issue that some people have reported, only the 3rd party RAM issue.

Leopard should add Memtest and while there is a POST, it would be nice if along with AHT there was similar to Vista, an option to schedule a memory test, or boot into a diagnostic, recovery mode.

Same with fsck as well, schedule a full check, with option to repair bad blocks if detected.

Better those than taking 10 minutes to scan the optical media of the installer (which I would rate 3rd if at all).

As for RAID and mirrors, there are threads, SoftRAID's 3.6.4 "broke" and depending on when the RAID was created and with what, I always create and start over and gradually reformat/erase or partition all my drives with new OS (and again a year later).

I don't think it can be said too often or enough, about backups. Thanks.
BootCamp has a PDF and stresses backups - in bold - yet goes unread and unheeded.

Thanks!

(Now if only I could reset-nvram in EFI the way we could in Open Firmware, which was always step #1 on troubleshooting hardware.)

PS: Not sure I agree with the need, necessity, or being desirable even to create a root account. I do just fine without and have very very rarely in 7 yrs now found the need for the account, ever.

nerowolfe wrote:

You are too quick to knock Apple and to quick to defend what is probably questionable RAM. Again, I fail to see your logic. You seem to be hung up on a C/C++ bug. I think your analysis is highly unlikely because you have no basis whatsoever. If you were right, there would have been an immediate patch for it.

Enjoy

Can you chill out with your "pro apple" avengelism man., your starting to behave like a "a r s e" licker.,

We all understand that Apple actions issues posted here as well as listens to users and there new feature demands., but one things that strikes me is when knuckle heads have nothing better to do but hang around here to troll, why don't you take the day off already, all you seem to do is defend Apple from any issues raised when in fact we are on Apple discussion forums.

"OS" system installs should have a integrity check if upgrading a system as well as checking a user migration b testing every plug, 3rd party app etc, Leopard is supposed to do this and it fails.

I am sure we all appreciate your help, but please accept that Apple is a company made of people for people not some extraordinary person doing it all above us.

My installation was fine but operation slow. Checked RAM to find that it saw my original 2x512mb Apple chips but would not see my 2x1GB Crucial chips, which have been working fine under Tiger since February. Turns out the Crucial chips are Ballistic, which they NOW tell me are gaming chips and will not work with any other chips. They will replace under warranty but these chips are no good for me. So NOT guaranteed compatible. Catch 22.

So running on original Apple chips now.

Bill

I tried to install via upgrade a 12" PB G4 with the original 256MB + Viking 1GB in slot, and got a kernel panic near the end of the install. I retried doing a clean install, and it seemed to work okay, but I eventually got another kernel panic. It ran okay for a couple of days, but the kernel panics were occurring more frequently. Yesterday it would not run more than 10 minutes with a kernel panic, so I decided to try tracking down the problem. The panic log files pointed to the metadata server sometimes, but other times a kernel call to the graphics system. I replaced the Viking 1GB stick with the original 256MB from Apple (for a total of 512MB now) and it didn't make any difference. Booted from the DVD and did a disk check, all okay. So, I tried another clean install and this time it got stuck (twice) at the Essentials Pkg problem. I assumed I had some nasty hardware problem, but just for grins I decided to try a clean install of Tiger. That worked without a hiccup, and it has been running flawlessly ever since. It might be a RAM issue, but both the third party RAM and the original Apple RAM failed. I'm not going to worry about it and stick with Tiger, the GF is happy with Tiger and it's her machine. I really would like to know what's going on, though.