Since installing Catalina yesterday, multiple crashes from userspace watchdog timeout

I have two Mac Minis that I run as headless servers, both started exhibiting this behavior after upgrading to 10.15.6. I never ran into this before, so I thought it was a problem with this latest version of the OS. But now after reading this thread and others, I'm seeing the problem has existed quite a bit longer. I'm not sure why it just started happening for me.

As I said, I was using these as headless servers so there is no monitor or keyboard attached. I purchased a couple HDMI display emulators, to trick the Minis into thinking there is a monitor attached, thinking that might help alleviate the situation, but it hasn't made any difference.

On one computer (Mac Mini Late 2012) the crash and reboot happens multiple times a day, usually once overnight and 1 or 2 more times throughout the day. On the other one (Mac Mini 2018) the crashes happen less frequently, usually once every day or two.

I've read that one possible solution is to display Power Nap. I did that on both, but the 2012 mini still crashed today after making that change. So I don't think that is the solution. I've also read others post about issues with external monitors connected to a MacBook Pro, but that obviously doesn't apply in my case.

Hoping against hope that someone might have a solution, but moreso, I just wanted to add my voice to the chorus. Hopefully someone at Apple will take notice, and we'll get a fix sooner than later.

@BlakeEiseman, try unchecking "Put hard disks to sleep when possible" in the Energy Saver System Preferences. That's what I tried last week, as suggested by an Apple support tech, and I haven't encountered a single kernel panic since.

Thanks for the comment, @gzazu ... unfortunately I already had that option unchecked, and I'm still getting the kernel panics.

The odd thing with this is, it seems like the recipe is different for different people. What works for some, doesn't work for others. Still trying to figure out what works for me on these CPUs.

a huge issue for me is when I have hardware acceleration enabled on any apps, that causes a lot of bugs for me. Apps like Spotify, Chrome, Zoom all have acceleration as options that I've turned off. Things have been better (not perfect) since then.

It appears that macOS 10.15.6 doesn't like virtual imaging software so much.

After installing the latest versions of VirtualBox and VMWare Fusion Pro the system refuses to shut down & reboot: it hangs for a half of an hour and crashes with a "watchdog timeout" kernel panic.

Hardware acceleration has nothing to do with the watchdog timeout kernel panic.

The problem is likely associated with a kext conflict as result of a memory leak. It has been reported by users of both apps to their respective developers who are already aware of. Hopefully this issue will be corrected on an upcoming release followed by a supplemental update from Apple.

dragontorc wrote:

... as result of a memory leak.

Don't think it's memory leak, as I've had the WindowsServer hang and Watchdog reboot as the machine was starting up following a hang/reboot.

On the up side ... 10.15.6 (19G73) Public Release does seem to have fixed the issue for me. It's been a few days now since hangs, where I was seeing almost one an hour.

( iBin Mac Pro 2013 with D300 )

Unfortunately 10.15.6 definitely hasn't fixed the issue, but maybe your user behaviour isn't triggering it as much (yet).

I drove away for the entire weekend with my laptop left at home, physically open with the screen locked. Fully expected to come home on Sunday evening to a rebooted login screen with the usual enter password, TouchID disabled after reboot message. But it wasn't, it had sat there the whole time quite happily. Got through yesterday without a freeze and most of today until about 45 minutes ago.

Frankly I'm astonished this isn't getting more public attention. Unlike a separate issue the 2020 13" model that I have has, this issue seems to be affecting users across vast swathes of different hardware profiles; how this was not detected in testing I don't know.

I agree with your observation about avoiding VMWare Fusion. I had no crashes today after avoiding running a Windows 8 VM in VMWare Fusion (as I normally do). And I have all of the bad-for-this crash problem energy saver items checked (i.e., turned on): Put hard disks to sleep, Enable Power Nap, and "Prevent computer from sleeping" as unchecked. I still have all Sidecar items unchecked as I mentioned in previous posts. I have 3 external monitors connected and have been running MS Teams in the 3rd monitor, which is usually the kiss of death wrt causing an upcoming crash.

I repeatedly got this error recently after updating the last Catalina version 10.15.6 and running VMware Fusion with WIndows 10 for sometime. MacOS then automatically restarted. Previous Catalina version doesn't have this error. Please fix, thanks.

I have had no problems with VMWare Fusion (CentOS7 and VoidLinux) triggering this watchdog panic. Since the 10.15.6 update I have had no crash/reboot instances whether running Fusion or not. Under 10.15.5 I had watchdog crashes related to the system going to sleep while connected to an external monitor in clamshell mode. While I was often running Fusion at those times, the system *never* crashed while I was actively working, but rather only when I tried to wake it up after a sleep. When I successfully implemented the 'caffeinated -d -u' workaround advertised in this thread the system never crashed even under the more problematic 10.15.5 while running Fusion and/or after sleeping.

I heartily agree with danpalmer that it's rather shocking this issue isn't a lot more public at this point. I first tuned in due to my own clamshell/extmonitor crashes which began with the 10.15.5 update, but it seems more and more that there are a LOT of conditions which trigger this other than my own (which I am certainly delighted seems to be fixed for the moment by 10.15.6). This is atrociously unacceptable OS performance for a major vendor. Hope Apple starts selling more watches -- between the POS broke-keyboard 2016 MBP they sold me (which I had to ship off for factory repair THREE TIMES), this fiasco, and the long-time-broken Terminal app that took them about a year to fix, my estimation of Apple's hardware and software capabilities have plummeted over the last 3 years.

For my MBP running 10.15.6, I think it's the combination of running VMWare Fusion-with-Windows AND MS Teams simultaneously. If I run one or the other alone (with the rest of MS Office running, Apple Mail, Safari, etc), I'm not having crashes. When running them together, my system will always crash in the middle of a MS Teams conference call.

I'm not running VMWare Fusion at all. But I am running VitualBoxVM with an Ubuntu VM. And it happens multiple times per day.

As much as I hate to do this, I'm considering trying to do a timed reboot every couple hours, since that is more elegant than having a crash. I've had a few corrupted databases within the Ubuntu VM, which I've had to restore from a backup. At least with the timed reboot, it would be a graceful shutdown rather than a crash and potentially corrupted files.

POTENTIAL SOLUTION

After it getting so bad that last night I had a whole 14 minutes in between kernel panics, I did some more online research this morning and found a bunch of threads of various forums claiming issues with Google Drive File Stream, which I have installed (along with VMware Fusion, Virtualbox etc that have been mentioned in this thread).

The solution that has worked for thus far since implementing it this morning has been to go into System Preferences, Spotlight, Privacy and add any GDFS roots as exclusions from the locations Spotlight should index.

I’m not sure whether this is something to do with GDFS files not actually being present on the system and the copy on write functionality of APFS which Catalina requires failing to index them properly and somehow causing watchdog to crap itself, but regardless, so far it has apparently solved the issue.

If this is indeed the case then it's plausible similar solutions like Dropbox's Smart Sync may also need excluding (I don't have Smart Sync switched on mine so can't confirm).

Thanks for this suggestion, @danpalmer. I don't use GDFS, but I do have an rclone mount. I don't know if it will help, but at this point, I'm willing to try anything. So I've added the rclone mounted drive to the Spotlight exclusion list, and we'll see if it helps. I'll report back my findings (good or bad).