You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
We are seeing a bug on macOS Sequoia related to the test running.
When attempting to run tests, the iOS simulator becomes stuck indefinitely. This can occur whether we run tests for a specific module, all unit tests, or even a single test.
We narrowed down the issue and is due to the OS failing to copy the realm DB named pipes (FIFO):
db.realm.note
db.realm.management/access_control.new_commit.cv
db.realm.management/access_control.pick_writer.cv
We saw in the CoreSimulator log file the following error:
NSCocoaErrorDomain Code=512 ""access_control.new_commit.cv" couldn't be copied
The copy fails and then a new clone of the simulator is created retrying the process. And this goes on and on.
What is happening?
When running the tests for the iOS app on the simulator, under the hood, the OS tries to clone the simulator device.
A list of folders is created, including one Shared/AppGroup. Under the AppGroup folder, the list of multiple UDIDs corresponds to the specific App Group containers created for individual app targets or extensions that are sharing data within that App Group.
One of these folders contains the Realm DB files. There are a few files called named pipes which are invisible in Finder (even if you have hidden files to true). You need to list them with ls -l.
So, when we select the test target and run the tests, the OS clones the permanent simulator and copies all files from its folders. All files are being copied except those pipes.
When the OS attempts to copy the pipes, the operation fails with the error Code=512.
Are there any workarounds available?
Yes. We found that running the tests works if we first “Erase All Content and Settings” from the simulator.
Another workaround is to remove all simulators and reinstall the iOS runtimes. This prevents the issue for a longer period (almost a full day), but eventually, the problem reoccurs.
Alternatively, we can delete the named pipes from the App Groups directory before running the tests.
Have we tried to give full disk permissions
Yes, we tried to give full disk permissions to a lot of things (Xcode, simulator, file paths, directories, etc.) but with no luck. We still get the same error.
Is the issue happening on a specific version of Xcode?
No, it happens for multiple Xcode versions:
Xcode 15.4
Xcode 16.0
Xcode 16.1
Is the issue happening on a specific macOS Sequoia version?
No, it happens on multiple macOS Sequoia versions:
15.0 Beta 8
15.0 RC
15.0
15.1
Stacktrace & log output
Oct 29 17:41:18 CoreSimulatorService[14079] <Error>: New device is stuck in creation state, deleting: Clone 712 of iPhone SE (3rd generation) (58D6DED6-2C55-4E7C-BBB4-E0D661DC41A1, iOS 18.1, Creating)
Oct 29 17:41:20 CoreSimulatorService[14079] <Error>: Error Domain=NSPOSIXErrorDomain Code=22 "Invalid argument" UserInfo={NSLocalizedFailureReason=Device was allocated but was stuck in creation state. Check CoreSimulator.log for more information.}
Oct 29 17:41:20 com.apple.dt.Xcode[90147] <Error>: Error Domain=NSPOSIXErrorDomain Code=22 "Invalid argument" UserInfo={NSLocalizedFailureReason=Device was allocated but was stuck in creation state. Check CoreSimulator.log for more information.}
Oct 29 17:41:30 CoreSimulatorService[14079] <Warning>: Device 6C270BDB-2945-42B5-A985-884F93BFD3E1 encountered in creation state at launch. The device will be re-created.
Oct 29 17:41:38 CoreSimulatorService[14079] <Error>: Failed to clone the device data path, error = Error Domain=NSCocoaErrorDomain Code=512 "“access_control.new_commit.cv” couldn’t be copied to “db.realm.management”." UserInfo={NSSourceFilePathErrorKey=[...]/db.realm.management/access_control.new_commit.cv, NSUserStringVariant=(
Copy
), NSDestinationFilePath=[...]/db.realm.management/access_control.new_commit.cv, NSUnderlyingError=0x600000ffa490 {Error Domain=NSPOSIXErrorDomain Code=45 "Operation not supported"}}
Oct 29 17:41:38 CoreSimulatorService[14079] <Error>: Failed to re-create device that was encountered in the creation state (Clone 713 of iPhone SE (3rd generation) (6C270BDB-2945-42B5-A985-884F93BFD3E1, iOS 18.1, Creating)): Error Domain=NSCocoaErrorDomain Code=512 "“access_control.new_commit.cv” couldn’t be copied to “db.realm.management”." UserInfo={NSSourceFilePathErrorKey=[...]/db.realm.management/access_control.new_commit.cv, NSUserStringVariant=(
Copy
), NSDestinationFilePath=[...]/db.realm.management/access_control.new_commit.cv, NSUnderlyingError=0x600000ffa490 {Error Domain=NSPOSIXErrorDomain Code=45 "Operation not supported"}}
Oct 29 17:41:40 CoreSimulatorService[14079] <Error>: Error Domain=NSPOSIXErrorDomain Code=22 "Invalid argument" UserInfo={NSLocalizedFailureReason=Device was allocated but was stuck in creation state. Check CoreSimulator.log for more information.}
Oct 29 17:41:50 CoreSimulatorService[14079] <Warning>: Device C6DEBFBB-6EFA-4E4C-B51B-9DDA08AF9BDB encountered in creation state at launch. The device will be re-created.
Oct 29 17:41:55 CoreSimulatorService[14079] <Error>: Failed to clone the device data path, error = Error Domain=NSCocoaErrorDomain Code=512 "“access_control.new_commit.cv” couldn’t be copied to “db.realm.management”." UserInfo={NSSourceFilePathErrorKey=[...]/db.realm.management/access_control.new_commit.cv, NSUserStringVariant=(
Copy
), NSDestinationFilePath=[...]/db.realm.management/access_control.new_commit.cv, NSUnderlyingError=0x600000ffb000 {Error Domain=NSPOSIXErrorDomain Code=45 "Operation not supported"}}
Oct 29 17:42:06 CoreSimulatorService[14079] <Warning>: Device 032BAE7E-E345-48F2-86EB-4DF1AD4D5291 encountered in creation state at launch. The device will be re-created.
Can you reproduce the bug?
Always
Reproduction Steps
Run the iOS app on a simulator (e.g. iPhone SE 3rd gen. with iOS 18.1).
Quit Xcode and the iOS simulator.
Select a test target to run.
Select the same simulator you previously ran the iOS app.
Run the tests (CMD + U).
The simulator is now stuck.
Version
10.54.1
What Atlas Services are you using?
Local Database only
Are you using encryption?
Yes
Platform OS and version(s)
macOS Sequoia 15.0 or 15.1. iOS 18.0, iOS 18.1
Build environment
Multiple Xcode versions: 15.4, 16.0, 16.1
macOS Sequoia: 15.0 or 15.1
The text was updated successfully, but these errors were encountered:
How frequently does the bug occur?
Sometimes
Description
Hi Team! 👋
We are seeing a bug on macOS Sequoia related to the test running.
When attempting to run tests, the iOS simulator becomes stuck indefinitely. This can occur whether we run tests for a specific module, all unit tests, or even a single test.
We narrowed down the issue and is due to the OS failing to copy the realm DB named pipes (FIFO):
db.realm.note
db.realm.management/access_control.new_commit.cv
db.realm.management/access_control.pick_writer.cv
We saw in the CoreSimulator log file the following error:
The copy fails and then a new clone of the simulator is created retrying the process. And this goes on and on.
What is happening?
When running the tests for the iOS app on the simulator, under the hood, the OS tries to clone the simulator device.
A list of folders is created, including one
Shared/AppGroup
. Under theAppGroup
folder, the list of multiple UDIDs corresponds to the specific App Group containers created for individual app targets or extensions that are sharing data within that App Group.One of these folders contains the Realm DB files. There are a few files called named pipes which are invisible in Finder (even if you have hidden files to true). You need to list them with
ls -l
.So, when we select the test target and run the tests, the OS clones the permanent simulator and copies all files from its folders. All files are being copied except those pipes.
When the OS attempts to copy the pipes, the operation fails with the error
Code=512
.Are there any workarounds available?
Yes. We found that running the tests works if we first “Erase All Content and Settings” from the simulator.
Another workaround is to remove all simulators and reinstall the iOS runtimes. This prevents the issue for a longer period (almost a full day), but eventually, the problem reoccurs.
Alternatively, we can delete the named pipes from the App Groups directory before running the tests.
Have we tried to give full disk permissions
Yes, we tried to give full disk permissions to a lot of things (Xcode, simulator, file paths, directories, etc.) but with no luck. We still get the same error.
Is the issue happening on a specific version of Xcode?
No, it happens for multiple Xcode versions:
Is the issue happening on a specific macOS Sequoia version?
No, it happens on multiple macOS Sequoia versions:
Stacktrace & log output
Can you reproduce the bug?
Always
Reproduction Steps
CMD
+U
).Version
10.54.1
What Atlas Services are you using?
Local Database only
Are you using encryption?
Yes
Platform OS and version(s)
macOS Sequoia 15.0 or 15.1. iOS 18.0, iOS 18.1
Build environment
The text was updated successfully, but these errors were encountered: