Loop process ends when app goes to background and is not relaunched until opened manually (4 h overnight outage) – iPhone 18 Pro / iOS 27.0 / Loop 3.14.8 / G7 + Omnipod 5

[GurmeetGhataurey]

Describe the bug
Loop stops running shortly after it goes to the background and is not relaunched by iOS until I open it manually. Overnight this caused a 4-hour period with no automation.

Overnight 29–30 Sep 2026 (times UTC; local = BST, UTC+1):

• 22:01:50Z – Loop (pid 386) logs [lifecycle] APP BACKGROUND. Last successful loop 22:01:51Z.
• 22:01:50Z → 02:04:57Z – no DeviceLog entries at all: no G7 reads, no pod communication, no dosing decisions. Loop stored no glucose between 22:01Z (18.9 mmol/L / 340 mg/dL) and 02:06Z (5.1 mmol/L / 92 mg/dL).
• The Dexcom G7 app on the same phone kept receiving readings throughout; glucose fell to ~4.0 mmol/L (72 mg/dL).
• A 3.3 U correction bolus had been given at 21:54Z. With Loop not running, the pod reverted to scheduled basal (1.0 U/hr) and nothing could suspend delivery as glucose fell.
• Loop Failure notifications were delivered: 22:21Z and 22:41Z (timeSensitive), 23:01Z and 00:01Z (critical).
• 02:04:57Z – a new process (pid 1594) started in the background; APP FOREGROUND at 02:05:05Z when I opened Loop. Looping resumed immediately.

This is recurring, not a one-off. Across 29–30 Sep the Loop process changed about 10 times. Most processes end exactly at APP BACKGROUND, and the next process starts only when I open the app (first lifecycle event is APP FOREGROUND within <1 s of process start). Process IDs and time ranges (UTC):

• 29 Sep: 583 (→05:30:28), 2687 (05:39:28–11:37:04), 4193 (11:37:24–16:21:21), 5408 (16:42:01–16:57:48), 5569 (17:07:53–17:07:54), 5578 (17:08:29–19:51:22)
• iPhone restarted between 19:51Z and 19:58Z (PID reset)
• 29–30 Sep: 386 (19:58:41–22:01:50), 1594 (02:04:57–12:06:36), 3731 (12:06:38–13:16:29), 4018 (13:28:30–15:42:23), 4301 (15:42:56–)

[I did not swipe Loop away in the app switcher.]

Attach an Issue Report
Issue Report attached, together with the Critical Event Log export (Export-20260930T210025Z.zip). [Add any Loop-*.ips crash reports or panic-full logs from 29–30 Sep if found in Settings → Privacy & Security → Analytics Data.]

To Reproduce
Not deterministic. It happens several times a day.

1. Run Loop in Closed Loop (Temp Basal Only) with Dexcom G7 and Omnipod 5 on iPhone 18 Pro / iOS 27.0.
2. Send Loop to the background (lock the phone or switch apps).
3. Leave the phone locked.
4. Loop sometimes stops logging immediately. It is not relaunched by BLE events from the G7 or pod, and Loop Failure notifications follow until the app is opened manually.

Expected behavior
Loop keeps running in the background, or is relaunched by iOS via Bluetooth state restoration on the next G7 reading, so automated dosing continues without the app needing to be opened.

Screenshots
N/A. The logs are attached.

Phone

• Hardware: iPhone 18 Pro (iPhone19,2)
• OS Version: iOS 27.0 (24A437)

Loop Version

• Version Number: 3.14.8 (2)
• Repo: LoopKit/LoopWorkspace main, Browser Build (fork: gsg-apps/LoopWorkspace)

CGM

• Device: Dexcom G7
• Manager app: Dexcom G7 app

Pump

• Manufacturer: Insulet
• Model: Omnipod 5
• Firmware version: [Loop → pod screen → Pod Details]

Additional context

• iPhone 18 Pro on iOS 27.0 is not yet a recommended combination. I previously saw a crash on background launch before first unlock after a reboot (EXC_BREAKPOINT/SIGTRAP, wasUnlockedSinceBoot = 0), which may be related.
• Separately, the DeviceLog shows intermittent G7 BLE errors ("Failed to encrypt the connection" CBErrorDomain 15; "connection has timed out unexpectedly" code 6; "Error enabling notification for authentication: unknownCharacteristic"). These cause 20–25 minute gaps with Loop Failure alerts, e.g. 30 Sep 06:06–06:42Z, 09:01–09:24Z and 17:51–18:29Z. Happy to open a separate issue for this if preferred.

Export-20260930T210025Z.zip

Archive.zip