[Bug]: de.beyondco.herd.helper (Not Responding) #1784

Activity

1. issuelabeler

   added

   macOS

   on Oct 24, 2026
2. diegoandornoorangepix commented on Oct 28, 2026

   

   diegoandornoorangepix

   on Oct 28, 2026

   More actions

   Same helper, Herd 2.0.3 (72), but in my case it crashes and nginx/dnsmasq never start.

   Environment: macOS 26.2 (25C56), arm64. Herd.app last modified 2026-10-26; the helper binary was
   installed 2026-10-20; helper bundle version 1.0.53.

   Symptoms: after launching Herd (or "Restart all"), php-fpm and the devtools helper start, but
   nginx and dnsmasq do not. Nothing listens on :80, :443 or :53. The .test resolver points to
   127.0.0.1, so DNS times out and every .test site is unreachable.

   launchctl print system/de.beyondco.herd.helper -> state = not running, runs = 2.

   Crash reports in /Library/LaunchDiagnostics... (/Library/Logs/DiagnosticReports):
   de.beyondco.herd.helper-2026-10-28-095647.ips, ...-090531.ips (today, at Herd launch) and
   2026-10-27-091211.ips. Exception in all of them:
   EXC_BAD_ACCESS / SIGSEGV, KERN_INVALID_ADDRESS at 0x8000000000000008 -> 0x0000000000000008
   (possible pointer authentication failure), terminated by SIGNAL 11. The helper was launched at
   09:56:32 and crashed about 6 s later. The crashed thread has no frames in the report.

   Restarting the services from the Herd app does not help. First crash report is the day after the
   app was updated. Attached the .ips files.

   herd-helper-crash-redacted.zip

   👍React with 👍1Reacted by Mike
3. mikescott commented on Nov 12, 2026

   

   mikescott

   on Nov 12, 2026

   Last edited by mikescott

   Author

   More actions

   I've updated to Herd 2.1 and this process still shows as not responding
4. diegoandornoorangepix commented on Nov 13, 2026

   

   diegoandornoorangepix

   on Nov 13, 2026

   More actions

   Same here on 2.1.0. The helper still crashed on the first boot after the update, so nginx and dnsmasq never started and every .test site timed out.

   2.1.0 still ships the old helper. The helper inside Herd.app 2.1.0 is still bundle version 1.0.53, the same binary I had before (image UUID 7bef5732-466b-33c0-a9b7-b12d37b4f2fe). The update never replaced /Library/PrivilegedHelperTools/de.beyondco.herd.helper (dated 2026-10-20).

   The faulting thread now has a stack. It is identical, offset for offset, in today's crash (2.1.0) and in the one from Nov 10 (before the update):

   EXC_BAD_ACCESS (SIGSEGV) KERN_INVALID_ADDRESS at 0x8000000000000008 (possible pointer authentication failure)
   Thread: com.apple.root.default-qos
   0  libswiftCore.dylib       +0x923cc  swift_isUniquelyReferenced_nonNull_native
   1  de.beyondco.herd.helper  +0x1039c
   2  de.beyondco.herd.helper  +0x2fe0
   3  de.beyondco.herd.helper  +0x2ef8
   4  de.beyondco.herd.helper  +0x301c
   5  de.beyondco.herd.helper  +0x2f30
   6  de.beyondco.herd.helper  +0x83940
   7  de.beyondco.herd.helper  +0x65230
   8  de.beyondco.herd.helper  +0x8271c
   9  de.beyondco.herd.helper  +0x80b68
   10 de.beyondco.herd.helper  +0x7ef2c
   11 de.beyondco.herd.helper  +0x11d98
   12 libdispatch.dylib        +0x1b5c   _dispatch_call_block_and_release
   

   Timing from the unified log (crash at boot):

   09:00:12.789 launchd  Successfully spawned de.beyondco.herd.helper[5912] because ipc (mach)
   09:00:24.660 helper[5912:4bbd] Starting DNSMasq with command: ...
   09:00:24.662 helper[5912:4bbb] Starting Nginx with command: ...
   09:00:24.665 kernel   de.beyondco.herd.helper[5912] Corpse allowed 2 of 5
   09:00:24.666 launchd  service inactive: de.beyondco.herd.helper
   

   DNSMasq and Nginx are started from two different dispatch threads 2 ms apart, and the helper dies 3 ms later in swift_isUniquelyReferenced_nonNull_native. That looks like a data race: two concurrent blocks mutating the same Swift collection, for example a dictionary of running processes.

   The successful run fits that reading. Later, herd restart from the CLI recovered everything. In that run the two starts were about 300 ms apart and the helper did not crash:

   12:17:34.612 helper[63490:97170] Starting DNSMasq with command: ...
   12:17:34.905 helper[63490:97172] Starting Nginx with command: ...
   

   My guess is that the "Not Responding" state in the original report is the same race ending in a hang instead of a crash, but I can't confirm that from here.

   Environment: macOS 26.2 (25C56), Mac15,4 (arm64), Herd 2.1.0.

   Attached: both .ips reports (machine identifiers redacted) and the relevant unified log lines.

   herd-helper-crash-2.1.0-redacted.zip