fix(linux): reduce RPM build disk use

.github/workflows/linux-packages.yml

@@ -220,12 +220,15 @@ jobs:
           dist="$(find _packages/sources -name 'tablepro-*.tar.xz' ! -name '*-vendor.tar.xz')"
           vendor="$(find _packages/sources -name 'tablepro-*-vendor.tar.xz')"
           linux/build-aux/packaging/build-rpm.sh "$dist" "$vendor" _packages/rpm
-      - name: Keep the Meson test log
+      - name: Keep the RPM failure logs
         if: failure()
         uses: actions/upload-artifact@cf430e030ddbb5b0abf93d22962f4752f3646cd9 # v7.0.2
         with:
           name: linux-rpm-testlog-${{ runner.arch }}
-          path: _packages/rpm/testlog.txt
+          path: |
+            _packages/rpm/testlog.txt
+            _packages/rpm/disk-usage.tsv
+            _packages/rpm/disk-usage.txt
           if-no-files-found: ignore
           retention-days: 7
       - uses: actions/upload-artifact@cf430e030ddbb5b0abf93d22962f4752f3646cd9 # v7.0.2

linux/build-aux/packaging/build-rpm.sh

@@ -15,7 +15,55 @@ out="$(cd "$3" && pwd)"
 
 name="$(basename "$1" .tar.xz)"
 top="$(mktemp -d)"
-trap 'rm -rf "$top"' EXIT
+disk_samples="$out/disk-usage.tsv"
+disk_summary="$out/disk-usage.txt"
+disk_monitor=""
+
+sample_disk() {
+    df -Pk "$top" | awk -v timestamp="$(date +%s)" '
+        NR == 2 { printf "%s\t%s\t%s\t%s\n", timestamp, $2, $3, $4 }
+        END { if (NR != 2) exit 1 }
+    ' >> "$disk_samples"
+}
+
+cleanup() {
+    build_status=$?
+    trap - EXIT HUP INT TERM
+    if [ -n "$disk_monitor" ]; then
+        kill "$disk_monitor" 2>/dev/null || true
+        wait "$disk_monitor" 2>/dev/null || echo "RPM disk sampler stopped early; measurements may be incomplete." >&2
+    fi
+    # The archive and debug packages can peak before the build tree is removed.
+    sample_disk || true
+    if awk '
+        NR == 2 { start = $4; used = $3; peak = $3; minimum = $4 }
+        NR > 1 { if ($3 > peak) peak = $3; if ($4 < minimum) minimum = $4 }
+        END {
+            if (NR < 2) exit 1
+            print "RPM filesystem disk usage (sampled every 1 second, KiB):"
+            printf "start_available\t%.0f\nminimum_available\t%.0f\n", start, minimum
+            printf "peak_used\t%.0f\npeak_increase\t%.0f\n", peak, peak - used
+        }
+    ' "$disk_samples" > "$disk_summary"; then
+        cat "$disk_summary" || true
+    fi
+    rm -rf "$top" || true
+    exit "$build_status"
+}
+trap cleanup EXIT
+trap 'exit 129' HUP
+trap 'exit 130' INT
+trap 'exit 143' TERM
+printf 'unix_seconds\tsize_kib\tused_kib\tavailable_kib\n' > "$disk_samples"
+sample_disk
+(
+    trap - EXIT
+    trap 'exit 0' HUP INT TERM
+    while sleep 1; do
+        sample_disk
+    done
+) &
+disk_monitor=$!
 mkdir "$top/SOURCES" "$top/SPECS"
 cp "$1" "$2" "$top/SOURCES/"
 
@@ -35,9 +83,12 @@ dnf builddep -y --setopt=install_weak_deps=False "$srpm"
 # reads the vendored crates from the config %cargo_prep writes.
 # Meson prints only the end of a failed suite's log, so the whole log is
 # kept beside the packages.
-if ! CARGO_NET_OFFLINE=true rpmbuild --define "_topdir $top" --rebuild "$srpm"; then
-    find "$top/BUILD" -path '*/meson-logs/testlog.txt' -exec cp {} "$out/" \;
-    exit 1
+if CARGO_NET_OFFLINE=true rpmbuild --define "_topdir $top" --rebuild "$srpm"; then
+    :
+else
+    build_status=$?
+    find "$top/BUILD" -path '*/meson-logs/testlog.txt' -exec cp {} "$out/" \; || true
+    exit "$build_status"
 fi
 
 cp "$srpm" "$top"/RPMS/*/*.rpm "$out/"

linux/build-aux/rpm/tablepro.spec

@@ -2,6 +2,9 @@
 # Full debuginfo adds to what each rustc of the app and the final link
 # need, and an 8 GB builder ran out of memory with it.
 %global rustflags_debuginfo 1
+# Native objects and their archive coexist; full DuckDB DWARF can fill the builder before linking.
+%global _pkg_extra_cflags -g1
+%global _pkg_extra_cxxflags -g1
 # Fedora's one codegen unit puts the whole app crate through LLVM as one
 # module, and its unit test build passed 7 GB that way. Cargo's release
 # default of 16 splits the work, and thin LTO still optimizes across it.

linux/build-aux/tests/test_rpm_disk_usage.py

@@ -0,0 +1,77 @@
+import csv
+import io
+import os
+import subprocess
+import tempfile
+import tarfile
+import unittest
+from pathlib import Path
+
+SCRIPT = Path(__file__).resolve().parent.parent / "packaging" / "build-rpm.sh"
+
+
+class RpmDiskUsageTest(unittest.TestCase):
+    def test_a_failed_build_keeps_its_exit_code_and_stops_the_sampler(self):
+        with tempfile.TemporaryDirectory() as directory:
+            work = Path(directory)
+            tools = work / "bin"
+            tools.mkdir()
+            dnf = tools / "dnf"
+            dnf.write_text("#!/bin/sh\nexit 0\n")
+            dnf.chmod(0o755)
+            rpmbuild = tools / "rpmbuild"
+            rpmbuild.write_text("""#!/bin/sh
+set -eu
+build_top=${2#_topdir }
+if [ "$3" = -bs ]; then
+    mkdir -p "$build_top/SRPMS" "$build_top/BUILD/failed/meson-logs"
+    touch "$build_top/SRPMS/tablepro.src.rpm"
+else
+    printf 'package test failed\\n' > "$build_top/BUILD/failed/meson-logs/testlog.txt"
+    sleep 2
+    exit 37
+fi
+""")
+            rpmbuild.chmod(0o755)
+            temporary = work / "temporary"
+            temporary.mkdir()
+            source = work / "tablepro-0.1.0.tar.xz"
+            vendor = work / "tablepro-0.1.0-vendor.tar.xz"
+            with tarfile.open(source, "w:xz") as archive:
+                spec = b"Name: tablepro\n"
+                entry = tarfile.TarInfo("tablepro-0.1.0/linux/build-aux/rpm/tablepro.spec")
+                entry.size = len(spec)
+                archive.addfile(entry, io.BytesIO(spec))
+            vendor.touch()
+            output = work / "output"
+            environment = dict(os.environ, PATH=f"{tools}:{os.environ['PATH']}", TMPDIR=str(temporary))
+
+            # An orphan sampler holds the captured pipes open after the build exits.
+            result = subprocess.run(
+                ["sh", str(SCRIPT), str(source), str(vendor), str(output)],
+                env=environment,
+                capture_output=True,
+                text=True,
+                timeout=10,
+                check=False,
+            )
+
+            self.assertEqual(result.returncode, 37, result.stderr)
+            self.assertEqual(list(temporary.iterdir()), [])
+            self.assertEqual((output / "testlog.txt").read_text(), "package test failed\n")
+            with (output / "disk-usage.tsv").open() as samples:
+                rows = list(csv.DictReader(samples, delimiter="\t"))
+            self.assertGreaterEqual(len(rows), 3)
+            self.assertTrue(all(int(row["size_kib"]) > 0 for row in rows))
+            self.assertTrue(all(int(row["available_kib"]) >= 0 for row in rows))
+            summary = (output / "disk-usage.txt").read_text()
+            self.assertIn("sampled every 1 second", summary)
+            self.assertIn(summary, result.stdout)
+            measurements = dict(line.split("\t") for line in summary.splitlines()[1:])
+            self.assertEqual(int(measurements["start_available"]), int(rows[0]["available_kib"]))
+            self.assertEqual(int(measurements["minimum_available"]), min(int(row["available_kib"]) for row in rows))
+            self.assertEqual(int(measurements["peak_used"]), max(int(row["used_kib"]) for row in rows))
+
+
+if __name__ == "__main__":
+    unittest.main()