OWC Thunderbay 2: Kernel Panic on power off?

[Makosuke]

I’ve got an OWC Thunderbay 2 and an OWC Thunderbay Mini hooked to my 5K iMac, both with SoftRAID arrays running in them. I keep the Mini on all the time, but the big one is for bulk storage that I don’t use often, so I usually keep it powered off to reduce noise, power use, and maybe wear on the drives. That’s fine--SoftRAID volumes eject just like any others.

Problem: Every time I power off or disconnect the Thunderbay 2, my iMac immediately kernel panics. The drives have been properly ejected, but the OS goes down hard anyway.

If I’m reading the panic logs correctly, it’s not SoftRAID that’s causing the problem; it appears to be the Thunderbolt to SATA controller in the Thunderbay:

Kernel Extensions in backtrace:
         com.apple.driver.AppleAHCIPort(3.1)[93E1D2EC-50EE-3340-9676-0EA1E0B1AD45]@0xffffff7f929ed000->0xffffff7f92a07fff
            dependency: com.apple.iokit.IOAHCIFamily(2.7.5)[DAF0353B-16D5-385A-A0F3-FD5CF3BA07F2]@0xffffff7f9185f000
            dependency: com.apple.iokit.IOPCIFamily(2.9)[56AD16B5-4F29-3F74-93E7-D492B3966DE2]@0xffffff7f90d26000

Can anybody else with a Thunderbay help out with whether this is happening to you as well, or if powering off the connected device works as expected? I’m going to get ahold of OWC support next week, but I’m curious if this is a widespread issue or specific to my system.

 

[matreya]

Can anybody else with a Thunderbay help out with whether this is happening to you as well, or if powering off the connected device works as expected? I’m going to get ahold of OWC support next week, but I’m curious if this is a widespread issue or specific to my system.

I have an iMac 5k and 2 Thunderbay IVs (the Thunderbolt1 original version) and if I turn off or disconnect them while the iMac is still running, I too get a Kernel Panic. This doesn't bother me so much as I never turn them off, but my procedure for whenever I want to change drives in one of them is to shutdown my iMac :/

Edit: I don't have SoftRAID, so it's definitely either a bug in the enclosures or the OS.

 

[Makosuke]

Thanks--useful data point. So definitely not just me.

I do wonder if it is just the 5K iMacs and at least some other machines are okay; it seems like if it was causing panics with everything they’d have noticed and at least put a warning in the manual or something...

 

[matreya]

Thanks--useful data point. So definitely not just me.

I do wonder if it is just the 5K iMacs and at least some other machines are okay; it seems like if it was causing panics with everything they’d have noticed and at least put a warning in the manual or something...

Actually, now that I think about it, I used these same enclosures on a late 2012 27" iMac and never got a KP when unplugging them...

 

[ColdCase]

Thanks--useful data point. So definitely not just me.

I do wonder if it is just the 5K iMacs and at least some other machines are okay; it seems like if it was causing panics with everything they’d have noticed and at least put a warning in the manual or something...

What OS are you running? If either my LaCie big 5 or OWC TB1 loses power or the cable has a disconnect, I typically have to restart my rMBP15 with 10.9.5 to recover.

I didn't check the log, but the strange thing is the screen is alive and the mouse moved about... but nothing else is happening. I assumed it was a TB dock thing, but perhaps not.. Since that happened maybe twice over the previous 6 months, not a big issue for me

I have my drives set to spin down so they save perhaps a bit of power/wear and tear.

 

[Makosuke]

10.10.2 in my case. And it’s not a hang for me, it’s a flat-out kernel panic with immediate reboot.

Your description makes it sound more like the system is hung waiting on some sort of I/O. Do you get a progressive series of beachballs, like when a drive stops responding but doesn’t go completely dead, or is it really just a “mouse moves, but nothing at all responds to clicks” situation?

 

[ColdCase]

No beach balls. For grins I let it set for an hour the last time, but no joy. When it happens again I'll take a closer look.

 

[ColdCase]

In the end I don't think I'm much help to you.

The LaCie is connected to Port one, via a TB dock. A DP monitor daisy chained after the LaCie. So when I turn the LaCie off and then back on I get the active screen, active mouse but nothing else (no secondary screen) symptom. Log does not seem to have a kernel panic but seems to indicate a problem with a log in window

3/1/15 11:44:46.106 AM launchservicesd[191]: Application App:"loginwindow" asn:0x0-1001 pid:180 refs=7 @ 0x7ff26b42a970 tried to be brought forward, but isn't in fPermittedFrontApps ( ( "LSApplication:0x0-0x269269 pid=6558 "ScreenSaverEngine"")), so denying. : LASSession.cp #1481 SetFrontApplication() q=LSSession 100004/0x186a4 queue
3/1/15 11:44:46.106 AM WindowServer[211]: [cps/setfront] Failed setting the front application to loginwindow, psn 0x0-0x1001, securitySessionID=0x186a4, err=-13066
3/1/15 11:44:49.784 AM loginwindow[180]: ERROR | -[LWBuiltInScreenLockAuthLion closeAuthAndReset:] | Attempted to remove an observer when not observing


There is also a number of disk not present entires.

The OWC Thunderbay 2 is directly connected to the second TB port. When I turned the OWC Thunderbay off and on today, I did not see a problem recovering. System just kept chugging along, the disk not present entries seemed to clear themselves.

Are you daisy chaining devices?

 

[theSeb]

Here is my thread about the same issue. I have tested different combinations and my findings are in there. I believe that this was introduced by 10.10.2

https://forums.macrumors.com/threads/1850692/

 

[Makosuke]

The OWC Thunderbay 2 is directly connected to the second TB port. When I turned the OWC Thunderbay off and on today, I did not see a problem recovering. System just kept chugging along, the disk not present entries seemed to clear themselves.

Are you daisy chaining devices?

My devices are not daisy-chained--one per port. Based on theSeb's testing in the linked thread (also coincidentally posted yesterday), it looks like it's intermittent, which might explain why you sometimes have problems and sometimes don't.

I haven't tested thoroughly enough to tell if my crash is 100% of the time (I try to avoid panic-ing my Mac when possible...), just that it's frequent.

For reference, "loginwindow" is the process that handles actually logging in your account; if it crashes, you will usually be unceremoniously dumped back to the login screen with no warning at all--an instant, forced logout. (Interestingly, in 10.10 at least, Activity Monitor will actually warn you of this if you try to use the GUI to kill it--I'd never seen that before, although I also don't think I'd ever tried with Activity Monitor.)

From those couple of log entries it looks like it's maybe getting confused and trying to throw up an unlock-on-wake window, but for some reason isn't getting pushed to the front so you can't see it, resulting in a stuck screen. I kind of wonder if it's because of something unusual the drivers for the TB dock you have are trying to do. Could also just be a bug, of course.

Here is my thread about the same issue. I have tested different combinations and my findings are in there. I believe that this was introduced by 10.10.2

Very useful testing. In my case, just for reference, I'm flipping the Thunderbay 2 off with its rear power switch, which is mechanical and therefore should be equivalent to pulling the plug from the wall. I haven't tried yanking the TB cord instead, but in any case it must be a low-level driver problem, and somewhat intermittent. I probably just haven't tried it enough times to realize that it doesn't happen 100% of the time.

 

[ColdCase]

Thanks, drivers do not get installed for the dock, uses whatever is built into the OS. The external monitor is the secondary screen and is not getting a signal (goes black then powers off) under this condition. My primary monitor is the rMBP screen. I have my tray set to pop up, so in this condition the tray doesn't pop up and clicking on any of the top items does nothing. Doesn't seem to be any response to the keyboard short of the power/shutdown button. The screen sits there with the desktop image and top menu displayed but otherwise empty and the mouse moves around as normal.

Its one of those OS 10.9.5 bugs... errr... quirks... errr character things. Perhaps has more to do with the display disappearing than the drives. One of several annoying character flaws, I'm not going to 10.10 any time soon.

 

[Makosuke]

Thanks, drivers do not get installed for the dock, uses whatever is built into the OS.

Right, but even if they’re built into the OS, it’s still a driver. I assume the dock uses controllers that the OS has support for built-in, but that support could be buggy, flakey, or be getting somehow confused by the “disappearing" hardware. None of which are what’s supposed to happen, of course.

In any case, it’s possible that the issues are fixed in 10.10, although given the problems I and others are having with 10.10.2, maybe not (though the two do seem to be largely unrelated, even if what causes them is similar).

 

[theSeb]

When I use the big blue button to power off on the LaCie 2big I never get a kernel panic, but it works like a soft power off, based on what I can see, because it will switch on again if you unplug the tb cord and plug it back in.

I haven't had the time to try and repeat this exercise thoroughly, but since I did the testing in the linked thread the issue seems to have gone away. I have to add though that I have only powered off 3 times since then. Now I am more confused

 

[ColdCase]

Not that it matters much, but I found simply disconnecting or turning the external monitor off does the same thing. Because it was daisy chained through the LaCie, the display must get disconnected when LaCie power is off (makes sense). External display, sleep, and recent Mac OSs can provided loads of entertainment, although this seems to be the lone remaining issue for me.

 

[Makosuke]

I’m wondering if the new 10.10.3 beta fixes this issue, but I’m a little wary of installing beta OS updates on my primary work machine. Probably won’t be too long until it’s released anyway.

 

[kakman]

I’m wondering if the new 10.10.3 beta fixes this issue, but I’m a little wary of installing beta OS updates on my primary work machine. Probably won’t be too long until it’s released anyway.

I'll just confirm I get the same behaviour (kernel panic/hard restart) on disconnection of a Thunderbay 4. I've seen it happened when connected to both a Retina MBP and a current model Mac Mini - both running 10.3.3.

Edit to say it happens whether it's the only device connected or whether it's daisy chained with another drive.

 

[Makosuke]

...connected to both a Retina MBP and a current model Mac Mini - both running 10.3.3.

I assume that was a typo and you actually meant the beta of 10.10.3, right?

 

[Eye In The Sky]

I am experiencing the same problem Makosuke. I also have iMac 5k with 10.10.2. Every time I unmount my ThunderBay 4 and turn power the unit off it immediately causes a kernel panic shutdown. I just installed SoftRAID 5.0.7 and the issue still exists. I hope this issue is resolved sooner than later!

 

[Darby67]

Running latest beta of 10.10.3, issue has gone away for me. Actually since installing the earlier beta about a month or so ago, it has been solid.

 

[MCAsan]

I’ve got an OWC Thunderbay 2 and an OWC Thunderbay Mini hooked to my 5K iMac, both with SoftRAID arrays running in them. I keep the Mini on all the time, but the big one is for bulk storage that I don’t use often, so I usually keep it powered off to reduce noise, power use, and maybe wear on the drives. That’s fine--SoftRAID volumes eject just like any others.

Problem: Every time I power off or disconnect the Thunderbay 2, my iMac immediately kernel panics. The drives have been properly ejected, but the OS goes down hard anyway.

If I’m reading the panic logs correctly, it’s not SoftRAID that’s causing the problem; it appears to be the Thunderbolt to SATA controller in the Thunderbay:

Kernel Extensions in backtrace:
         com.apple.driver.AppleAHCIPort(3.1)[93E1D2EC-50EE-3340-9676-0EA1E0B1AD45]@0xffffff7f929ed000->0xffffff7f92a07fff
            dependency: com.apple.iokit.IOAHCIFamily(2.7.5)[DAF0353B-16D5-385A-A0F3-FD5CF3BA07F2]@0xffffff7f9185f000
            dependency: com.apple.iokit.IOPCIFamily(2.9)[56AD16B5-4F29-3F74-93E7-D492B3966DE2]@0xffffff7f90d26000

Can anybody else with a Thunderbay help out with whether this is happening to you as well, or if powering off the connected device works as expected? I’m going to get ahold of OWC support next week, but I’m curious if this is a widespread issue or specific to my system.

What does the OWC support staff say? Sound like the ball is in their court.

 

[Makosuke]

Adding a potentially useful data point, with the release version of 10.10.3 installed I haven't yet had a panic.

I admittedly haven't powered one of the OWC units on and off all that many times yet, so I'm not willing to declare the problem unequivocally fixed quite yet either, but it's definitely improved at minimum, and may be entirely resolved.