Merge pull request #46 from scality/improvement/archive-cores

Adds archive core action.

archive-cores/action.yaml

@@ -0,0 +1,31 @@
+name: "archive cores"
+description: 'Archive all cores corresponding to a given set of core_patterns'
+
+inputs:
+  dest:
+    description: |
+      Destination directory to archive cores to. A subdirectory
+      'cores' will be created under that destination directory
+    required: true
+
+  core_patterns:
+    description: |
+      List of paths where the cores might be searched. Glob paths
+      are supported to filter core files from the others.
+    required: true
+
+  ignore_filters:
+    description: |
+      Filters to ignore a found core (space separated, globs supported) applied
+      against the full backtrace.
+
+      This can for example be the function name that will always be present
+      in the relevant call stack.
+    default: ""
+
+runs:
+  using: "composite"
+  steps:
+    - name: Archive found cores
+      shell: bash
+      run: ${GITHUB_ACTION_PATH}/archive-cores.sh "${{ inputs.dest }}" "${{ inputs.ignore_filters }}" ${{ inputs.core_patterns }}

archive-cores/archive-cores.sh

@@ -0,0 +1,134 @@
+#!/bin/bash
+set -euo pipefail
+
+# Archive all cores corresponding to a given set of core_patterns to
+# $dest_directory/cores.
+#
+# When possible, the binary that generated the core is also archived, and
+# backtraces are generated and archived.
+#
+# Ignore filters are applied against each found backtrace to determine if the
+# core should be archived or not.
+#
+# Example:
+# ./archive_cores.sh $ARTIFACT_DIR "daemontest" ./core.* /tmp/core.*
+
+if [ "$#" -lt 2 ]; then
+    echo "usage: $0 <dest_directory> <ignore_filters> <core_patterns...>"
+    exit 1
+fi
+
+DEST="$1"
+IGNORE_FILTERS="$2"
+shift
+shift
+CORE_PATTERNS="$@"
+
+OUTPUT_DIR="${DEST}/cores"
+
+# Piping into xargs is sometime necessary when globs are single quoted
+# by the caller / environment.
+CORES="$(echo $CORE_PATTERNS | xargs ls -dUN || /bin/true 2> /dev/null)"
+
+is_expected_crash()
+{
+    local bt_full="$1"
+    local bin="$2"
+
+    for match in $FILTER_PATTERNS; do
+        if grep -q "$match" "$bt_full"; then
+            echo "Known and expected crash of $bin: $match, skipping"
+            return 0
+        fi
+    done
+
+    # Ignores externally sent SIGSEGV signals. This might be used to test behavior
+    # of the process when encountering this exception.
+    if grep '^#3  <signal handler called>' -A2 "$bt_full" | grep -q '^#4 .*\(epoll_wait\|libpthread\)'; then
+        echo "Known and expected crash of $bin: daemon externally killed with SIGSEGV, skipping"
+        return 0
+    fi
+
+    return 1
+}
+
+# Retrieve the path of the binary from a core
+get_bin_path()
+{
+    local core="$1"
+    local bin=
+
+    # First try: get the core name with file
+    bin="$(file "$core" | grep 'execfn: ' | sed "s/.*execfn: '\([^']\+\)'.*/\1/")"
+    test -n "$bin" && bin="$(command -v "$bin")"
+
+    # This may fail if there are too many section headers in the core, newer
+    # versions of `file` can deal with this
+    if [ ! -e "$bin" ]; then
+        bin="$(file -Pelf_phnum=20000 "$core" 2>/dev/null \
+               | grep 'execfn: ' \
+               | sed "s/.*execfn: '\([^']\+\)'.*/\1/")"
+        test -n "$bin" && bin="$(command -v "$bin")"
+    fi
+
+    # Last try with gdb
+    if [ ! -e "$bin" ]; then
+        bin="$(gdb -n --batch "$core" 2>/dev/null \
+               | grep "^Core was generated by" \
+               | sed "s/^Core was generated by \`\([^ \`']\+\).*/\1/")"
+        test -n "$bin" && bin="$(command -v "$bin")"
+    fi
+
+    if [ -e "$bin" ]; then
+        echo "$bin"
+        return 0
+    else
+        return 1
+    fi
+}
+
+if [ -z "$CORES" ]; then
+    echo "no core to collect"
+    exit 0
+fi
+
+mkdir -p "${OUTPUT_DIR}"
+for core in $CORES; do
+    if ! file "$core" | grep -q 'core file'; then
+        echo "$core is not a core file"
+        file "$core"
+        continue
+    fi
+
+    core_name="$(basename "$core")"
+    bin="$(get_bin_path "$core")"
+
+    echo "archive core '$core' for binary '$bin' to $OUTPUT_DIR"
+
+    if [ ! -e "$bin" ]; then
+        echo "failed to parse binary path or does not exist, got: '$bin'"
+        file "$core"
+    else
+        # Generate backtraces
+        core_name="$(basename "$bin")-$(basename "$core")"
+        prefix="${OUTPUT_DIR}/$core_name-backtrace"
+        bt_full="$prefix-full.txt"
+        bt_all="$prefix-all.txt"
+
+        gdb -batch -ex 'bt full' "$bin" "$core" > "$bt_full"
+
+        # Known and expected crashes
+        if is_expected_crash "$bt_full" "$bin"; then
+            rm "$bt_full"
+            continue
+        fi
+
+        gdb -batch -ex 'thread apply all bt' "$bin" "$core" > "$bt_all"
+
+        # Archive binary
+        cp "$bin" "$OUTPUT_DIR"
+    fi
+
+    # Archive core
+    tar -czSf "${OUTPUT_DIR}/$core_name.tar.gz" "$core"
+done