Since the Xcode Cloud outage on 2026-11-17 until this morning (2026-11-20), every Xcode Cloud build of ours has failed from some kind of network timeout on the runner. We did not change our network setup or our CI scripts at that time. This morning, 5 of 7 builds failed, and the 2 that passed still had significant network latency (see below).
### 1. The clone/setup step is slow
This is the time from the start of the run to the first line of `ci_post_clone.log`. None of our code runs in this step.
- Before 2026-11-17: 1–3 minutes.
- Since 2026-11-17: 28–94 minutes.
- This morning: 28 and 59 minutes on the 2 builds that passed, and 54 and 85 minutes on 2 that failed.
### 2. Run times
- Before 2026-11-17: our test workflow took a median of 18 minutes, and our archive workflows took 25–40 minutes.
- Since 2026-11-17: every archive run stops at the 120-minute limit with "Build has exceeded the maximum allotted time".
- This morning: the 2 test runs that passed took 47 and 77 minutes.
### 3. Large downloads fail with "No route to host"
- **Test:** Swift Package Manager cannot download binary packages that are hosted as github.com release assets: "One or more binary package dependencies could not be downloaded. This may be a transient infrastructure issue."
- **Archive:** Gradle cannot download `kotlin-native-prebuilt` from `dl.google.com`. It stops after 3 attempts.
### 4. Small requests work
In the same failed runs, API calls, webhooks, RubyGems installs and a JDK download succeed.
### Questions
1. Did the network configuration of the Xcode Cloud runners change during or after the 2026-11-17 outage, for example IPv6, egress rules or a proxy?
2. Does anyone else see slow clones or failed large downloads since that date?
### We filed a Feedback Assistant report: FB25658313.
1
6
95