199982 Views Previous 1 … 28 29 30 31 32 … Next 525 Replies Latest reply: Nov 3, 2010 9:22 AM by @martinlabelle Go to original post
Well I couldn't hold out any longer waiting for 10.6.3. I've moved back to Tiger. Yes, the whole machine feels slower, and I miss quick-look, the new XCode, etc., but -- NO BEACHBALLS! Same hardware, same disk, same partitions, same third-party software.
I guess I'll periodically check when Apple makes point releases on Snow Leopard to see if the problem is solved, but the whole thing might have to wait for my next hardware upgrade cycle -- maybe when Nehalem-based notebooks finally come out.
Update on my saga, hopefully the last. I eventually started getting a few beachballs even under Tiger (system booted off external drive, but using most work files on the internal). Neither Disk Utility nor Smart Utility reported any issues with the drive, but I threw up my hands and replaced it anyway, with a different model. Did a block-level clone onto the new drive and guess what? No beachballs even under SL. I've never seen a drive be "bad" by just freezing periodically, while still retaining all of its data. But who knows? I'm sending the drive back, might test in a non-Mac before doing so. I'm still wondering why the whole system locks just because one app waits to access (and why nothing gets logged during these incidents).
Hi Adrian Robert;
It turns out that I had a similar experience many years ago, where the drive tested good by every test I could find but periodically everything slowed to a crawl. I finally trace it to when I was using a particular drive. The drive was still in warranty so I called the vendor. They told me to send it back.
Shortly I received back my disk with all of the data still intact. The explanation they gave was that the on the drive was defective. So whenever I did any that used cache, which is just about everything, the drive would suddenly start operating without any cache.
Sorry I wasn't able to see my old problem in yours earlier. I thought you might find their explanation interesting. After they replaced the cache, the disk worked perfectly fine for many years.
I suspect that Symantec Utilities may be the culprit. I know that Apple stated clearly in the installation that Snow Leopard is incompatible with Symantec Utilities, but the problem is when I uninstalled Symantec using the Uninstall package supplied by Symantec, it said it didn't find anything to uninstall. But in my system log, it clearly showed that the Symantec's daemon is crashing left and right every 10 sec from system log:
com.apple.launchd (com.symantec.symdaemon): Job appears to have crashed: Segmentation fault
com.apple.launchd (com.symantec.symdaemon): Throttling respawn: Will start in 10 seconds
So I finally went to Symantec's web site and finally find this link to download the shell script to completely remove everything Symantec has in sight:
So, I thought I irradicated the culprit, but I spoke too soon.
After the reboot, I get the hang again. Then I looked at the system log and found this (as I always suspect that Microsoft Office Reminder notifier is always causing the hang, which sometimes, when I kill MS Office Reminder during the hang, it wakes up immediately):
Mar 26 14:20:43 Microsoft Office Reminders: start timer
Mar 26 14:20:45 Microsoft Office Reminders: ECF launch successful. Notifying
Mar 26 14:20:47 Microsoft Office Reminders: ECRMenuManager -- ECFWasInitialized
Mar 26 14:20:47 Microsoft Office Reminders: ECRMenu - loading project list
Mar 26 14:20:47 Microsoft Office Reminders: ECSelectionCache projects: invalid filter template assigned. Using blank
Mar 26 14:20:47 Microsoft Office Reminders: ECRMenu - loading category list
Mar 26 14:20:47 Microsoft Office Reminders: ECSelectionCache categories: invalid filter template assigned. Using blank
Mar 26 14:20:50 osascript: Error loading /Library/ScriptingAdditions/Adobe Unit Types.osax/Contents/MacOS/Adobe Unit Types: dlopen(/Library/ScriptingAdditions/Adobe Unit Types.osax/Contents/MacOS/Adobe Unit Types, 262): no suitable image found. Did find:\n /Library/ScriptingAdditions/Adobe Unit Types.osax/Contents/MacOS/Adobe Unit Types: no matching architecture in universal wrapper
Mar 26 14:21:18 Microsoft Office Reminders: ECF shutting down. Notifying
Mar 26 14:21:18 Microsoft Office Reminders: Clearing selection
Mar 26 14:21:18 Microsoft Office Reminders: freeing menu objects
Mar 26 14:29:37 com.apple.kextcache: InternalModemSupport.kext does not declare a kernel dependency; using com.apple.kernel.6.0.
Mar 26 14:29:51: --- last message repeated 1 time ---
Mar 26 14:29:51 com.apple.kextcache: Created prelinked kernel /System/Library/Caches/com.apple.kext.caches/Startup/kernelcache_i386.3F15614A.
Mar 26 14:30:10 /System/Library/CoreServices/CCacheServer.app/Contents/MacOS/CCacheServer: No valid tickets, timing out
Mar 26 14:32:46 Safari: INSERT-HANG-DETECTED: Tx time:114.924372, # of Inserts: 0, # of bytes written: 0, Did shrink: NO
The log shows the missing 30 sec during the system stall was caused by Adobe Unit Types.osax
"osascript: Error loading /Library/ScriptingAdditions/Adobe Unit Types.osax" between Mar 26 14:20:50 and 14:21:18 until Microsoft Office Reminder is aborted.
(So I don't think Mail is causing the hang because I never use Mail, only MS Entourage.)
Then it hangs again by Safari, which is caused by the Kernel cache problem: com.apple.kextcache
So it looks like it is a cascading problem, one thing leads to another that causes the hang.
So I thought, as someone here thinks, that WebKitPluginHost is somehow causing the hang too, because I noticed NetBarrier is detecting WebKitPluginHost is trying to transmit to other host through port 80 (HTTP), 443 (HTTPS), 1935 (Macromedia Flash Communications Server), so I blocked the outgoing traffic to see if it makes any difference. But it doesn't stop the hang.
I really think this system stall is caused by the classic "Deadlock Contention" problem in parallel-processing, when 2 processes are trying to request access to a resource at the same time (in this case, I suspect it is the request for inter-process communication), and neither one is releasing the resource, so both are waiting for the other to release it, but none is willing to release until the other releases it first. So the system hangs, because both are hogging up the resource, until the system times out in 30 seconds.
I think that is why the stall is caused by so many different applications that don't seem to be related in anyway, but they are related in the sense that these applications are requesting for resource access, and none of them are releasing it to the other for access, until the system finally arbitrates it by a timeout after 30 seconds of deadlock.
I also thinks that my Mac that uses FileVault encryption hangs much often than my other Mac that does not use FileVault. That is probably because FileVault is requesting much more system inter-process communication that requires arbitration/contention in the background than without FileVault.
I really thinks sas71 is right about this: http://discussions.apple.com/message.jspa?messageID=10403027#10403027
Apple really needs to fix this bug in resolving the 64-bit Apple Event implementation that did not create problem when it was used in 32-bit mode, as revealed by http://db.tidbits.com/article/10643
BTW: I ran the performance update http://support.apple.com/kb/DL947 and none of my Mac had that hard drive to be fixed, so I don't think the system freeze has anything to do with the disk drive.
Message was edited by: nicoladie
I was so happy to see there is an update to 10.6.3, so I thought this problem was fixed, since it seemed the system responded a bit faster than before. But then, it was short lived. It hangs again.
Here's the system log during the freeze, Safari started to hang as I noticed Entourage is trying to access the internet in the background. Safari's freeze was preceded by Dreamweaver's (CS4 version 10.0.0.4117) bug when it tried to free the memory that was not allocated. Could Dreamweaver be the culprit who brings down the system by corrupting the system memory?
Mar 30 14:08:32 Dreamweaver: Dreamweaver(505,0xa0aae4e0) malloc: * error for object 0x2fac3a90: pointer being freed was not allocated\n* set a breakpoint in mallocerrorbreak to debug
Mar 30 14:08:32 [0x0-0x32032].com.adobe.dreamweaver-10.0: Dreamweaver(505,0xa0aae4e0) malloc: * error for object 0x2fac3a90: pointer being freed was not allocated
Mar 30 14:08:32 [0x0-0x32032].com.adobe.dreamweaver-10.0: * set a breakpoint in mallocerrorbreak to debug
Mar 30 14:10:23 Safari: Periodic CFURLCache Insert stats (iters: 1107) - Tx time:0.285461, # of Inserts: 16, # of bytes written: 276255, Did shrink: NO, Size of cache-file: 181460992, Num of Failures: 0
Mar 30 14:14:26 Safari: INSERT-HANG-DETECTED: Tx time:3.940148, # of Inserts: 2, # of bytes written: 138477, Did shrink: YES
Mar 30 14:17:54 /Applications/Microsoft Office 2008/Microsoft Entourage.app/Contents/MacOS/Microsoft Entourage: Warning: accessing obsolete X509Anchors.
Mar 30 14:20:31 Safari: INSERT-HANG-DETECTED: Tx time:3.859549, # of Inserts: 0, # of bytes written: 0, Did shrink: YES
Mar 30 14:27:54 /Applications/Microsoft Office 2008/Microsoft Entourage.app/Contents/MacOS/Microsoft Entourage: Warning: accessing obsolete X509Anchors.
But, at least, now the system did not completely freeze everything; only Safari was frozen, followed by frozen Entourage; while leaving other applications running and accessible. So it is an improvement, yet it is not a fix of the bug yet.
Note that it hung 2 times consecutively at 14:14:26 and 14:20:31, only 6 minutes apart!
Message was edited by: nicoladie
Ok, I think I traced this annoying bug finally.
Maybe it is caused by the system memory pointer reference counter during either garbage collection when it cleans up memory or allocating the wrong memory when the free memory pointer is hosed by a wrong memory reference counter!
Here is what I did. I suspect the freeze may be caused by the obsolete copy of kernel cache, as I saw a consistent pattern in the system log that indicated kernel cache problem.
So I used Onyx to empty all caches, the kernel cache including, http://www.versiontracker.com/dyn/moreinfo/macosx/20070 and reboot Mac OSX 10.6.3.
Right after the reboot, I immediately see the freeze again, this time it is caused by a totally different application, System Preferences, which is different from any of the usually applications that caused the reported hang before.
Immediately after the malloc reference count underflow by System Preferences (which indicates a memory leak due to the system dereferencing a memory pointer), then it hung Safari again.
Mar 31 17:24:19 [0x0-0x1c01c].com.apple.systempreferences: System Preferences(209,0x7fff709bdbe0) malloc: reference count underflow for 0x20029d040, break on autorefcount_underflowerror to debug.
Mar 31 17:24:19 System Preferences: System Preferences(209,0x7fff709bdbe0) malloc: reference count underflow for 0x20029d040, break on autorefcount_underflowerror to debug.
Mar 31 17:24:59 kextd: updated kernel boot caches
Mar 31 17:27:51 Safari: INSERT-HANG-DETECTED: Tx time:182.356700, # of Inserts: 1, # of bytes written: 21566, Did shrink: NO
Mar 31 17:27:59 Safari: INSERT-HANG-DETECTED: Tx time:5.064829, # of Inserts: 28, # of bytes written: 217560, Did shrink: NO
Hope Apple will look into this memory leak caused by malloc: reference count underflow
I agree that I may have to reinstall SL from scratch, because no matter what I do, such as reinstall 10.6.3 Update Combo, zap the PRAM, clear the kernel cache, startup in 64-bit mode, turn off FileVault, it still cannot get rid of this error at startup:
Apr 2 13:36:30 WindowServer: kCGErrorFailure: Set a breakpoint @ CGErrorBreakpoint() to catch errors as they are logged.
Apr 2 13:36:30 com.apple.WindowServer: Fri Apr 2 13:36:30 WindowServer <Error>: kCGErrorFailure: Set a breakpoint @ CGErrorBreakpoint() to catch errors as they are logged.
I only find this error everytime during startup in the machine that hangs, but not in my other Mac that does not hang.