MacOS-X File System Layer is faulty.
Here is the case:
after I double-tested that the file system of my Ti 400 was clean, I enabled journaling and ran happy ever after - until...
... yesterday, after a record 45 days of uptime, drag-and-drop stopped working systemwide. So I decided to reboot the machine (which went fine), install the 8 updates that had accumulated in between boots and restarted.
Just a spinning rainbow cursor. No disk util is able to repair because of overlapping catalog extends.
OK, I have the machine running on Panther right now (my Jaguar install disks are 650Km away) and I am wondering. How could that happen?
There seems to be only one logical conclusion: Apples implementation of the HFS+ layer in MacOS-X is faulty, corrupting the file system while running.
Apple, this is really, really poor :-(
after I double-tested that the file system of my Ti 400 was clean, I enabled journaling and ran happy ever after - until...
... yesterday, after a record 45 days of uptime, drag-and-drop stopped working systemwide. So I decided to reboot the machine (which went fine), install the 8 updates that had accumulated in between boots and restarted.
Just a spinning rainbow cursor. No disk util is able to repair because of overlapping catalog extends.
OK, I have the machine running on Panther right now (my Jaguar install disks are 650Km away) and I am wondering. How could that happen?
There seems to be only one logical conclusion: Apples implementation of the HFS+ layer in MacOS-X is faulty, corrupting the file system while running.
Apple, this is really, really poor :-(
Comments
Originally posted by Smircle
...OK, I have the machine running on Panther right now (my Jaguar install disks are 650Km away) and I am wondering. How could that happen?
...Apple, this is really, really poor :-(
did you say ur running panther?
unless i read wrong you have no right to copmlain to apple, i am going to assume your not a developer and if you are you should have known better
(sorry, perhaps i'm being a little mean but its the truth)
Originally posted by Smircle
Here is the case:
after I double-tested that the file system of my Ti 400 was clean, I enabled journaling and ran happy ever after - until...
... yesterday, after a record 45 days of uptime, drag-and-drop stopped working systemwide. So I decided to reboot the machine (which went fine), install the 8 updates that had accumulated in between boots and restarted.
Just a spinning rainbow cursor. No disk util is able to repair because of overlapping catalog extends.
OK, I have the machine running on Panther right now (my Jaguar install disks are 650Km away) and I am wondering. How could that happen?
There seems to be only one logical conclusion: Apples implementation of the HFS+ layer in MacOS-X is faulty, corrupting the file system while running.
Apple, this is really, really poor :-(
There are over 7,000,000 users of OS X...and here is the first complaint about HFS+ being unreliable... and you are using a beta of Panther. Strikes me as being a gargantuan leap in logic.
He also said that he had an uptime of 45 days... Panther hasn't been out for 45 days, so there's no way he could have been running it when he crashed. He just doesn't have anything better at the moment.
However, even if you were using Jag, I think it's a little hasty to jump from "My Hard Drive Crashed" to "OS X has a faulty implementation of HFS." Something got corrupted. That's all.
Originally posted by McCrab
There are over 7,000,000 users of OS X...and here is the first complaint about HFS+ being unreliable... and you are using a beta of Panther. Strikes me as being a gargantuan leap in logic.
Now then here is the second complaint: I had several disk related problems with OS X which I haven't had with OS 9. It was not hardware related (occurred on more than 1 mac and with different harddisks). But I have to append that the problems only occurred when I "cross-used" OS 9(/classic) and OS X. With OS X only the problems have not occurred again.
It also sounds like your disk needs repairing and permissions fixing.
Guys, give him a little slack. He's *NOT* running Panther, but Jaguar. Also, his situation sucks, I've been there.
Smircle, it is much more likely that you have a bad disk, or bad RAM (that was my problem - the RAM was being corrupted, corrupting the catalog info, which was then being written out faithfully to disk), than you're the only person who has been bitten by some obscure but catastrophic HFS+ bug, no?
If you can, boot into 9 somehow and run Gauge Pro, a RAM checker available on versiontracker.com. It'll eliminate one variable that most people never think of. Not only that, but RAM is cheap.
If the RAM checks out OK (and no, you probably can't just run Apple's RAM checker under X, it fails to perform a specific test that mine was failed on), then have the drive checked out by DriveX, Tech Tool Pro, etc.
I'll bet it's one of these two items, almost certainly, and this will save you the four months it took for me to diagnose it.
I didn't enable journaling, as I have no idea how to do that (nor do I believe that I require that feature). If there is some kind of fault with the file system then its something that can't remain undocumented for very long.
The solution in my case was to have the techs replace the 20Gb drive with a 40Gb drive and reinstall Jag. I just wish that I knew what went wrong....
- No, I was running 10.2.5, *not* Panther. I am no ass to put a pre-release OS on a production machine.
- The HD tested physically OK (of course thats no 100% proof).
- After spending most of the weekend trying to repair my file-system from every OS left to me (thats 9 and 10.3), I eventually pulled my home folder off the partition and reformatted. Lucky thing I never trusted Apple enough to put valuable data on the boot partition but still keep partitioning my drives like in the old days.
- Toast would not run under Panther and the 10.3 Finder would not burn toast images of 10.2 I obtained from a friend. 9 to the rescue (what the heck I am going to do with my next Mac which will not support 9 anymore I wouldn't know).
BUT the interesting thing is: I am not the only one. I observe a lot of Mac-Users having the same problems eventually and I always put the blame on crashes and not enabling journaling. Now it bit me with journaling enabled. I do not believe that any application short of low-level disk tools should be able to damage a file system in a Unix-kind-of-OS, so I still believe Apple botched the HFS+ layer in OS-X.
Originally posted by T'hain Esh Kelch
I'll try again, Bit torrent????
Nope
And of course, an high-level application like bittorrent should not be able to corrupt a file system.
Originally posted by Smircle
OBUT the interesting thing is: I am not the only one. I observe a lot of Mac-Users having the same problems eventually and I always put the blame on crashes and not enabling journaling. Now it bit me with journaling enabled. I do not believe that any application short of low-level disk tools should be able to damage a file system in a Unix-kind-of-OS, so I still believe Apple botched the HFS+ layer in OS-X.
Why are you so sure that there isn't something wrong the the hard drive itself? Journaling won't save you from hardware errors.
Originally posted by JLL
Why are you so sure that there isn't something wrong the the hard drive itself? Journaling won't save you from hardware errors.
Of course, I cannot be 100% sure. Thats a given.
But:
- I believe, about halve of the MacOS-X users I know had a damaged file system one time or other. For some time, I did attribute it to installing KEXT's and stuff, then to crashes and not performing fscks afterwards.
- Formatting and reinstalling did not produce any drive error.
- During the whole 2 years this very HD is in operation, it never lost data, crashed the system or something.
- In some 10 years of using various Macs and PCs, I have only once witnessed a hardware error destroying a FS (an Athlon board). However, I have had my share of faulty FS implentations in 7.5.x and troubles when using RAMDoubler on 7.6.
It's Occams razor: software bugs are more likely than hardware faults, so it's a software bug unless proven otherwise.
I started this thread mainly so others have a chance to be alert when their Macs just out of a sudden refuse to work. This may not be the fault of some crash, some application going wrong, but Apples HFS+ implementation.
Originally posted by Smircle
It's Occams razor: software bugs are more likely than hardware faults, so it's a software bug unless proven otherwise.
It doesn't have to be a software bug in HFS+. Remember that HFS+ is years old by now and a corrupted file doesn't have to be corrupted by the file system - the application writing the file can corrupt the file.
Btw. the missing drag'n'drop you mention is usually fixed by deleting you cache files.
Originally posted by Smircle
I believe, about halve of the MacOS-X users I know had a damaged file system one time or other
What on earth are you doing?
Originally posted by JLL
It doesn't have to be a software bug in HFS+. Remember that HFS+ is years old by now and a corrupted file doesn't have to be corrupted by the file system - the application writing the file can corrupt the file.
Only that no file was corrupted, but the catalog extends. I don't exactly know what it is, but it sounds like part of the disk catalog.
In a Unix-like environment, sure as hell no higher-level application should be allowed to write directly to the disk catalog (that was not even the case with the Apple ][ 20 years ago).
just a thought.
Check. Your. RAM. Seriously.
My drive was 100% fine.
I had catalog extent problems.
They came out of nowhere.
It sounds *suspiciously* familiar.
The catalog extent is a file, just like any other, it just happens to be controlled by the OS filesystem layer, that's all. If it gets corrupted at any point, (including in RAM), then you're hosed.
Considering that you've probably eliminated the drive as the problem, (I assume you also checked the integrity of the cabling?) and that, as many people have pointed out, the likelihood of you getting bitten by a software bug in a battle-tested layer is pretty slim...
Check. Your. RAM.
I can't say this enough.