Dockerfile: drop cap_sys_nice file caps — they break exec on unprivileged-LXC hosts (lab net-sim crash-looped)

Dockerfile

@@ -71,7 +71,6 @@ RUN apt-get update && apt-get install -y --no-install-recommends \
         libbsd0 libgps28 libasound2 libjack-jackd2-0 libpulse0 \
         direwolf \
         ca-certificates \
-        libcap2-bin \
     && rm -rf /var/lib/apt/lists/* /var/cache/apt/* /var/log/*
 
 # binaries
@@ -89,16 +88,15 @@ EXPOSE 8080 8001 8002
 # A non-root user for the running process.
 RUN useradd --system --no-create-home --shell /usr/sbin/nologin sim
 
-# -rt-priority renices the router and its TNC children, which needs
-# CAP_SYS_NICE — but the container runs as the non-root `sim` user, and
-# docker's --cap-add only populates the *bounding* set (a non-root process
-# doesn't inherit it). File capabilities on the binaries grant it to the
-# process directly; still inert unless the container ALSO gets
-# --cap-add SYS_NICE (file caps can't exceed the bounding set), and the
-# binaries only use it when -rt-priority is passed.
-RUN setcap cap_sys_nice+ep /usr/local/bin/sim-web \
-    && setcap cap_sys_nice+ep /usr/local/bin/sim-router
-
+# NOTE on -rt-priority: it renices the router + TNC children, which needs
+# CAP_SYS_NICE. Grant it at RUNTIME with `--cap-add SYS_NICE` (and run the
+# container as root, or with that cap in the user-namespace) — NOT with file
+# capabilities on the binaries. A `setcap cap_sys_nice+ep` here makes the
+# binary REFUSE TO exec (EPERM "operation not permitted") on any host whose
+# bounding set lacks the cap — e.g. an unprivileged LXC — breaking the image
+# for everyone who isn't using -rt-priority. -rt-priority already degrades
+# gracefully (logs a one-line warning and runs at normal priority) where the
+# platform won't grant the cap, so the binary must stay capability-free.
 USER sim
 
 ENTRYPOINT ["/usr/local/bin/sim-web", "-addr", ":8080", "-config", "/etc/sim/network.yaml"]

README.md

@@ -110,8 +110,9 @@ services:
     cap_add: [SYS_NICE]
 ```
 
+The capability is granted at **runtime** (`--cap-add` / `cap_add`), deliberately — the image carries no file capabilities on its binaries. (An earlier build set `cap_sys_nice+ep` on `sim-web`/`sim-router`; that made them refuse to `exec` with *"operation not permitted"* on any host whose bounding set lacks the cap — e.g. an unprivileged LXC — breaking the image even for runs that never pass `-rt-priority`. Runtime cap-add is the correct mechanism.)
 
-> **Nested-container hosts:** on a host that is itself an unprivileged container (e.g. Docker inside an unprivileged Proxmox LXC), the kernel checks CAP_SYS_NICE against the *init* user namespace, so negative nice is unavailable to anything inside — even container root. The flag then logs its one-line warning and the sim runs at normal priority; everything else is unaffected.
+> **Nested-container hosts:** on a host that is itself an unprivileged container (e.g. Docker inside an unprivileged Proxmox LXC), the kernel checks CAP_SYS_NICE against the *init* user namespace, so negative nice is unavailable to anything inside — even container root, even with `cap_add`. The flag then logs its one-line warning and the sim runs at normal priority; everything else is unaffected.
 
 ## Quick install (curl | sudo bash)
 

internal/dockerfile/dockerfile_test.go

@@ -0,0 +1,52 @@
+package dockerfile
+
+// Guards against reintroducing file capabilities on the shipped binaries.
+//
+// A `setcap cap_sys_nice+ep` on sim-web/sim-router makes them refuse to exec
+// ("operation not permitted") on any host whose capability bounding set lacks
+// CAP_SYS_NICE — e.g. an unprivileged LXC — which broke the image for every
+// run, including those that never pass -rt-priority. The capability must be
+// granted at runtime (--cap-add SYS_NICE) instead, so the binaries stay
+// capability-free. This caught a real production breakage (packethacking/net-sim
+// — the lab net-sim crash-looped after the setcap landed).
+
+import (
+	"os"
+	"path/filepath"
+	"strings"
+	"testing"
+)
+
+func TestDockerfileHasNoSetcapOnBinaries(t *testing.T) {
+	// Walk up to the repo root (where the Dockerfile lives).
+	wd, err := os.Getwd()
+	if err != nil {
+		t.Fatal(err)
+	}
+	var dockerfile string
+	for d := wd; d != "/" && d != "."; d = filepath.Dir(d) {
+		p := filepath.Join(d, "Dockerfile")
+		if _, err := os.Stat(p); err == nil {
+			dockerfile = p
+			break
+		}
+	}
+	if dockerfile == "" {
+		t.Fatal("could not locate Dockerfile from " + wd)
+	}
+	b, err := os.ReadFile(dockerfile)
+	if err != nil {
+		t.Fatal(err)
+	}
+	// Allow the word inside a comment explaining why we don't do it; only fail
+	// on an actual RUN setcap ... line.
+	for _, line := range strings.Split(string(b), "\n") {
+		trimmed := strings.TrimSpace(line)
+		if strings.HasPrefix(trimmed, "#") {
+			continue
+		}
+		if strings.Contains(trimmed, "setcap") {
+			t.Fatalf("Dockerfile reintroduced file capabilities (%q) — these break exec on hosts lacking the cap in their bounding set; grant CAP_SYS_NICE at runtime via --cap-add instead", trimmed)
+		}
+	}
+}