This discussion is locked
Fleur DeTran
Level 1 (45 points)

Q: Security Update 2007-5 damaged Powerbook G3-400 Lombard hard drive

Reporting unfortunate, possibly expensive, very time-consuming crash of my Powerbook G3-400 (Lombard) when I installed Apple's Security Update 2007-5 a couple of days ago:

• Logged on as admin user, no other app running (that I remember), but still had USB cables in (they were for iPod - not connected, and for external USB jump drive, connected). OS was 10.3.9, with all updates up to that date. Connected to broadband by Ethernet to switch to cable modem, NOT wireless.
Downloaded, installed, but then during optimization phase got spinning beachball with no progress for over 2 hrs.
• Side comment: HD has 3 partitions: OS X, OS9, and Storage for Users folders. Symlink from OS X's Users folder to actual user folders on the 3rd partition.
• Forced warm boot (Ctrl-Cmd-Pwr), then got nowhere - no OS load.
• Restarted with OS9 CD (no longer have OS X CD, only DVD that does Lombard won't read under OS X): OS X system partition flaky, unable to select it as Startup partition.
• Opened computer, removed hard drive, put it into external case with FW & USB, inspected HD with another FW computer: while OS X partition could be read (at first), running Disk Util showed it was a bad partition, and had to be re-initialized. Lucky I had backed up all non-OS X applications first!
• Re-installed OS X from DVD to the external HD case, applied updates.
• Moved HD back from external case to the PowerBook: now booted OK.
• Disk Util on HD now shows 3rd partition ("Storage") bad.
• However - when went to tighten screws of HD carrier and close Powerbook, lost the HD all over again. Best I can figure out now, somewhere in that process, the orange power/data cable connecting the HD to the IDE/Motherboard quit working (broke?), so I now have a dead Powerbook, until the hardware issue is fixed.

Bottom line: Applying this Security Update led to massive frustrations, frittering of humongous gobs of time (waiting for OS9 to load from CD whenever I tried to test and see if it recognized the internal HDD), and now a dead Powerbook G3. Hopefully a replacement ribbon cable will do the job; although if it's downstream from it (i.e., motherboard connector pins or IDE controller), that's going to be VERY expensive.
Reading this forum and others, perhaps I should have removed the attached USB devices (cable for iPod but without iPod attached, and USB thumbdrive). Also read somewhere that others have had problems with moving Users folders to separate partitions and symlinking to those cause problems with Security Update.
However: wish Apple's Security Updates were better behaved, this experience is making me much more scared of applying any updates in future…

12" iBookG4, iMacG3-400, 15" PowerBook G3-400   Mac OS X (10.3.9)   Toying at OSX Server networking

Posted on Jun 6, 2007 11:24 AM

Close
Fleur DeTran

Q: Security Update 2007-5 damaged Powerbook G3-400 Lombard hard drive

  • All replies
  • Helpful answers

  • by Grant Bennet-Alder,

    Grant Bennet-Alder Grant Bennet-Alder Jun 6, 2007 12:54 PM in response to Fleur DeTran
    Level 9 (61,292 points)
    Desktops
    Jun 6, 2007 12:54 PM in response to Fleur DeTran
    I hear what you are saying.

    Let's talk about failure probabilities over large populations. These trends may not apply to your specific case, since you do not have a large sample, you have one.

    Of the components you mention, the ones least likely to fail are those that consist only of wires, such as the Hard drive cable. As I recall, it is a flex cable, which is even less likely to die in normal use unless you manage to stretch or pinch it -- it is made to be flexed.

    Next least likely to fail is the motherboard electronics. Those drivers and connectors are meant to be plugged and unplugged a few times, but admittedly are not intended for harsh service, and are sometimes subject to zapping by static.

    Most likely of all these components to fail is the spinning metal disk whose moveable heads fly much closer to the rapidly spinning surface than a smoke particle, a fingerprint, or a human hair. One good shock, or a few dozen smaller ones, and the head could crash onto the surface of the disk, scrubbing off some of the oxide and spraying it over the inside of the drive to cause more problems later.

    If this is the original drive, it was manufactured in about 1999. The manufacturer probably sold the same drive at retail with a one-year warranty. Eight years later it has failed. Even if it is a replacement, it is the most likely the Drives that will fail. Thats why you were wise enough to make a backup.

    I expect that the trouble you are having with multiple partitions is an indication that the disk is dying. The way to determine this is to re-write it with known patterns, and read back all those patterns and note which sectors require hardware error correction (to check for errors). A working drive has a number of spare sectors that can be low-level substituted for ones that are found to be bad during this test.

    The way to accomplish this feat under Mac OS X is to initialize the drive with the "Zero All Data" option. It will take several hours to complete. If it passes, your disk has all good sectors. If it fails, running the test again sometimes does a second level of substitution and allows the drive to initialize OK.

    I do not know your exact situation, but over a large sample, it is more likely that your drive died, and more likely that it is mostly a coincidence that it happened during this update, except that the optimize is known to read and write a lot of data in different parts of the disk.

  • by Fleur DeTran,

    Fleur DeTran Fleur DeTran Jun 7, 2007 6:10 AM in response to Grant Bennet-Alder
    Level 1 (45 points)
    Jun 7, 2007 6:10 AM in response to Grant Bennet-Alder
    Thanks for taking the time. This is a replacement, 40 GB drive. Even though it failed the Disk Util check on 2 of the partitions when connected through external Firewire case (not both at once), I had erased the bad 8GB startup partition 1, and re-loaded OS X on it through the external FW case; which is why I (rightly or wrongly) assumed the erased partition is still good (it still passes Disk Util now when I returned it to the external FW drive).
    Great point about the Security Update optimization writing over a lot more sectors than in usual operations, I did not know about that.

    Operationally, I still have a PB that does not start up from HD with not only this HD but also with another (older) HD; after it did so without problem and the only intervening change was my removing the HD carrier completely (ribbon cable and all) to apply the side screws that hold the HD in the carrier. Did not feel/notice any static discharge at time I re-installed that drive. Moreover, when I power on, it goes immediately to CD -do not hear any HD spinup sound at all. So my next cheapest fix might be to replace the ribbon cable, if it's not the HD itself.

    Meanwhile, I'll try to do the low-level reformat as you kindly suggested. Will let you know how that works out.

    Again, Thanks for the rational, reasoned tips!

    12" iBookG4, PowerMac G4, 14" PowerBook G3-400   Mac OS X (10.3.9)   Toying at OSX Server networking

  • by Grant Bennet-Alder,

    Grant Bennet-Alder Grant Bennet-Alder Jun 7, 2007 10:03 AM in response to Fleur DeTran
    Level 9 (61,292 points)
    Desktops
    Jun 7, 2007 10:03 AM in response to Fleur DeTran
    One should not be too satisfied when a disk passes Disk Utility's "Repair" function. Disk Utility's "Repair" function does not do much checking of the actual data blocks that hold your files.

    What the Repair function does is verify the integrity of the directory and the free-list. The directory is the table of disk addresses and pointers indicating where fragments of files are stored, and how many blocks each fragment occupies. It also verifies that sectors listed as used, and sectors listed as available are in correspondence, to avoid mistakenly overwriting blocks of good data that appear to be free. While it is doing these checks it also checks to be sure files do not overlap (are not "cross-linked").

    What the repair function does NOT do is check the integrity of the data blocks that hold your files. These could be in quite bad shape, but as long as the directory is not messed up, the Repair function would pass.

    In Mac OS 9 and earlier Drive Setup, there is a "Test Disk" function, which does read every block and checks the drive's status for read errors, which are usually corrected by the drive itself. It takes several hours to do this on an average sized disk. Mac OS X has not implemented this function, relying instead on the "S.M.A.R.T." Disk status that is pervasive in today's IDE/ATA drives. It is important to look at the "S.M.A.R.T." Disk status that appears when selecting a Hard Drive in Disk Utility. It is expected that a bad "S.M.A.R.T." Disk status would be reported as a System Error, but I have never seen such an error, or heard from users who have seen such an error.

    Whether in the long run the "S.M.A.R.T." Disk status is sufficient to determine whether your drive is okay remains to be seen. There are several anecdotes posted here that indicate drives may become completely unusable before the "S.M.A.R.T." Disk status indicates they have a problem.

    Executive summary: Passing Disk Utility's Repair function is good, but not sufficient to tell whether your Disk Drive has some serious problems.

  • by Fleur DeTran,

    Fleur DeTran Fleur DeTran Jun 21, 2007 11:50 AM in response to Grant Bennet-Alder
    Level 1 (45 points)
    Jun 21, 2007 11:50 AM in response to Grant Bennet-Alder
    In the end, turned out that the motherboard had a short-circuit right in the area of the IDE controller chip, so it was neither the software update nor the hard drive's fault. The repair tech noticed the short overlay an area of a crack in the bottom of the case, which had over the last couple of years widened to a hole about 1" (2.5 cm) or so wide.
    I do not know if the final burnout of the IDE disk controller chip occurred due to the extra stress related to writing and implementing the security update, or if it was sheer coincidence.
    Thank you, Grant, again, for your judicious comments.