Files
scylladb/scripts/open-coredump.sh
Petr Hála a04dbac369 open-coredump: Change to use new backtrace
* This is a breaking change, which removes compatibility with the old backtrace
    - See https://staging.backtrace.scylladb.com/api/docs#/default/search_by_build_id_search_build_id_post for the APIDoc
* Add timestamp field to log
* Tested locally

Closes scylladb/scylladb#28325
2026-02-05 11:50:47 +02:00

444 lines
14 KiB
Bash
Executable File

#!/bin/bash
# -*- coding: utf-8 -*-
#
# Copyright (C) 2022-present ScyllaDB
#
#
# SPDX-License-Identifier: LicenseRef-ScyllaDB-Source-Available-1.0
#
#
# Requirements:
# * eu-unstrip
# * curl
# * jq
# * git
set -e
trap 'echo "error $? in $0 line $LINENO"' ERR
SCRIPT_NAME=$(basename $0)
SCYLLA_S3_RELOC_SERVER_DEFAULT_URL=https://api.backtrace.scylladb.com
function print_usage {
cat << EOF
Usage: ./${SCRIPT_NAME} [options] COREFILE
Script for "one-click" opening of coredumps.
It extracts the build-id from the coredump, retrieves metadata for that
build, downloads the binary package, the source code and finally
launches the dbuild container, with everything ready to load the
coredump.
Sometimes the metadata retrieved from the build lacks the package URL.
In this case, the scylla package can be downloaded manually and extracted
into the artifact directory (see below) using the "scylla.package" name.
When this directory exists, the script will not attempt to extract the
package URL from the metadata and download the package itself, instead it
will use the provided package.
The script is idempotent: running it after the prepartory steps will
reuse what is already downloaded. It is *strongly* recommended to run
this from an empty directory, with nothing but the core-file present.
Options:
--help,-h
Print this help message and exit.
--quiet,-q
Only print errors (if any).
--verbose,-v
Print progress information when downloading the package and cloning
the git repo.
--ci
Designate the coredump as one coming from CI. These coredumps are
produced by a merge commit between the main branch (master or
enterprise) and the tested branch and this commit will not be
present in origin. Using this flag will force the main branch to
be checked out, instead of the commit hash obtained from the
ScyllaDB vesrsion.
--artifact-dir,-d ARTIFACT_DIR
Directory where the script will store all downloaded artifacts. If
the directory doesn't exist, it will be created.
It is strongly recommended that this is an empty directory, not used
for anything else, to avoid interference with the script.
To reset the state of the script, just delete this directory.
This can be the same as the directory the coredump is already in, or
some other path.
By default, a directory called ${SCRIPT_NAME}.dir will be created in
the script's working directory.
Can be also provided via env variable ARTIFACT_DIR.
If both are provided, command line has precedence.
--scylla-package-url,-p SCYLLA_PACKAGE_URL
Instead of querying the s3-reloc-server (see below) and downloading the
package from the obtained URL, download it directly from the provided URL
and extract the scylla version metadata from the package.
If the package is already downloaded, it is not downloaded again.
--scylla-s3-reloc-server-url,-u SCYLLA_S3_RELOC_SERVER_URL
The URL of the s3 reloc server to connect to. Needed to fetch the
build info based on the build-id.
Defaults to ${SCYLLA_S3_RELOC_SERVER_DEFAULT_URL}
Can be also provided via env variable SCYLLA_S3_RELOC_SERVER_URL.
If both are provided, command line has precedence.
--scylla-repo-path,-r SCYLLA_REPO_PATH
Path to an existing scylla repository to use. The repo is expected to have
the branch and commit the packages were built from.
If not provided, the appropriate git repository (either scylladb.git or
scylla-enterprise.git) will be cloned from https://github.com/scylladb.
The script assumes you have the appropriate access rights set up.
Can be also provided via env variable SCYLLA_REPO_PATH.
If both are provided, command line has precedence.
--scylla-gdb-py-source,-s repo|package|none
Where to get the scylla-gdb.py script from:
* repo - download the latest one from the appropriate main branch;
* package - use the one that comes with the relocatable package;
* none - don't try to obtain scylla-gdb.py;
Defaults to repo.
Unlike other artifacts, scylla-gdb.py is placed in the same directory
where the coredump is located. If one is already present, it is not
overwritten.
Can be also provided via env variable SCYLLA_GDB_PY_SOURCE.
If both are provided, command line has precedence.
EOF
}
function log {
if [[ $VERBOSE_LEVEL -ge 1 ]]
then
echo $@
fi
}
function get_json_field {
local json_obj=$1
local field_name=$2
local optional=${3:-0}
local field_val=$(jq -r ".${field_name}" <<< "$json_obj")
if [[ -z $field_val ]] && [[ $optional -eq 0 ]]
then
echo "error: failed to get field '$field_name' from: $json_obj" >&2
exit 1
fi
echo $field_val
}
function get_repo_name() {
local branch=$1
local sha1=$2
local repo_name
local remote_repos
pushd ${SCYLLA_REPO_PATH} > /dev/null
# First, try to find the repo from local information
repo_name=$(git branch -r --contains "$sha1" 2>/dev/null | \
sed -n 's|^ *\(.*\)/.*$|\1|p' | \
head -n 1)
if [ -n "$repo_name" ]; then
echo "$repo_name"
return 0
fi
# If not found locally, try fetching from each remote
remote_repos=$(git remote -v | awk "/scylladb\/${PRODUCT}.*fetch/{ print \$1 }")
remote_name=$(while read -r remote; do
git fetch -q "$remote" "$branch" 2>/dev/null
if git branch -r --contains "$sha1" | grep -q "$remote"; then
echo "$remote"
break
fi
done <<< "$remote_repos")
popd > /dev/null
if [ -n "$remote_name" ]; then
echo "$remote_name"
else
echo "error: could not determine remote repository for SHA1 $sha1" >& 2
return 1
fi
}
for required in eu-unstrip jq curl git; do
if ! type $required >& /dev/null; then
echo "error: missing required program $required, please install first" >&2
exit 1
fi
done
VERBOSE_LEVEL=1
CORE_FROM_CI=0
PACKAGE_URL="${SCYLLA_PACKAGE_URL}"
SCYLLA_S3_RELOC_SERVER_URL="${SCYLLA_S3_RELOC_SERVER_URL:-$SCYLLA_S3_RELOC_SERVER_DEFAULT_URL}"
SCYLLA_REPO_PATH="${SCYLLA_REPO_PATH}"
SCYLLA_BUILD_ID="${SCYLLA_BUILD_ID}"
SCYLLA_GDB_PY_SOURCE="${SCYLLA_GDB_PY_SOURCE:-repo}"
ARTIFACT_DIR=${ARTIFACT_DIR:-${SCRIPT_NAME}.dir}
if [[ $# -ge 1 && ( "$1" == "--help" || "$1" == "-h" ) ]]; then
print_usage
exit 0
fi
while [[ $# -gt 1 ]]
do
case $1 in
"--verbose"|"-v")
VERBOSE_LEVEL=2
shift 1
;;
"--quiet"|"-q")
VERBOSE_LEVEL=0
shift 1
;;
"--ci")
CORE_FROM_CI=1
shift 1
;;
"--artifact-dir"|"-d")
ARTIFACT_DIR=$2
shift 2
;;
"--scylla-package-url"|"-p")
PACKAGE_URL=$2
shift 2
;;
"--scylla-s3-reloc-server-url"|"-u")
SCYLLA_S3_RELOC_SERVER_URL=$2
shift 2
;;
"--scylla-repo-path"|"-r")
SCYLLA_REPO_PATH=$2
shift 2
;;
"--scylla-gdb-py-source"|"-s")
SCYLLA_GDB_PY_SOURCE=$2
shift 2
;;
"--scylla-build-id"|"-b")
SCYLLA_BUILD_ID=$2
shift 2
;;
*)
echo "error: unrecognized option: $1, see $0 -h for usage" >&2
exit 1
;;
esac
done
COREFILE=$1
shift 1
if ! [[ -f $COREFILE ]]
then
echo "error: ${COREFILE} is not a valid path to a core file" >&2
exit 1
fi
COREDIR=$(dirname ${COREFILE})
ARTIFACT_DIR=$(realpath ${ARTIFACT_DIR})
mkdir -p $ARTIFACT_DIR
if [[ -z "${SCYLLA_REPO_PATH}" ]]
then
SCYLLA_REPO_PATH="${ARTIFACT_DIR}/scylla.repo"
fi
SCYLLA_REPO_PATH=$(realpath ${SCYLLA_REPO_PATH})
if [[ ${VERBOSE_LEVEL} -lt 2 ]]
then
CURL_QUIET_FLAG="-s"
GIT_QUIET_FLAG="-q"
else
CURL_QUIET_FLAG=""
GIT_QUIET_FLAG=""
fi
case "${SCYLLA_GDB_PY_SOURCE}" in
("repo"|"package"|"none")
;;
*)
echo "error: invalid value for option SCYLLA_GDB_PY_SOURCE, has to be one of repo, package or none, got: ${SCYLLA_GDB_PY_SOURCE}" >&2
exit 1
;;
esac
if [[ -z "${PACKAGE_URL}" ]]
then
BUILD_ID="${SCYLLA_BUILD_ID}"
if [[ -z "${BUILD_ID}" ]]; then
BUILD_ID=$(eu-unstrip -n --core ${COREFILE} | grep 'scylla$' | cut -f2 -d' ' | cut -f1 -d@)
fi
log "Build id: ${BUILD_ID}"
# https://api.backtrace.scylladb.com/api/docs#/default/search_by_build_id_search_build_id_get
BUILD=$(curl "${SCYLLA_S3_RELOC_SERVER_URL}/api/search/build_id?build_id=${BUILD_ID}" -H 'accept: application/json')
if [[ -z "$BUILD" ]]
then
echo "error: failed to retrieve build information from ${SCYLLA_S3_RELOC_SERVER_URL}" >&2
exit 1
fi
RESPONSE_BUILD_ID=$(get_json_field "$BUILD" "build_id")
BUILD_MODE=$(get_json_field "$BUILD" "build_type")
PACKAGE_URL=$(get_json_field "$BUILD" "unstripped_url")
BUILD_DATA=$(get_json_field "$BUILD" "build_data")
VERSION=$(get_json_field "$BUILD_DATA" "version")
PRODUCT=$(get_json_field "$BUILD_DATA" "product")
RELEASE=$(get_json_field "$BUILD_DATA" "release")
ARCH=$(get_json_field "$BUILD_DATA" "platform")
TIMESTAMP=$(get_json_field "$BUILD_DATA" "timestamp")
if [[ "$RESPONSE_BUILD_ID" != "$BUILD_ID" ]]
then
echo "error: mismatching build id: requested ${BUILD_ID} but got ${RESPONSE_BUILD_ID}" >&2
exit 1
fi
log "Matching build is ${PRODUCT}-${VERSION} ${RELEASE} ${BUILD_MODE}-${ARCH} from ${TIMESTAMP}"
fi
if ! [[ -d ${ARTIFACT_DIR}/scylla.package ]]
then
if [[ -z $PACKAGE_URL ]]
then
echo "error: no package_url in build object: ${BUILD}" >&2
echo "" >&2
echo "The package can be provided manually by placing it (unpacked) to "${ARTIFACT_DIR}/scylla.package" to work around this problem." >&2
exit 1
fi
PACKAGE_FILE=$(basename ${PACKAGE_URL})
if ! [[ -f ${ARTIFACT_DIR}/${PACKAGE_FILE} ]]
then
log "Downloading relocatable package from ${PACKAGE_URL}"
curl -L ${CURL_QUIET_FLAG} --output ${ARTIFACT_DIR}/${PACKAGE_FILE} ${PACKAGE_URL}
else
log "Relocatable package ${PACKAGE_URL} already downloaded"
fi
log "Extracting package ${PACKAGE_FILE}"
pushd ${ARTIFACT_DIR} > /dev/null
tar -zxf $PACKAGE_FILE
mv scylla scylla.package
popd > /dev/null
else
log "Relocatable package ${PACKAGE_URL} already downloaded and extracted"
fi
# If the package was provided directly we bypassed talking to the S3 server and
# the version metadata has to be loaded from the package itself.
if [[ -z "$VERSION" ]]
then
VERSION=$(cat ${ARTIFACT_DIR}/scylla.package/SCYLLA-VERSION-FILE)
PRODUCT=$(cat ${ARTIFACT_DIR}/scylla.package/SCYLLA-PRODUCT-FILE)
RELEASE=$(cat ${ARTIFACT_DIR}/scylla.package/SCYLLA-RELEASE-FILE)
fi
if [[ "${PRODUCT}" == "scylla-enterprise" ]]
then
MAIN_BRANCH=enterprise
else
MAIN_BRANCH=master
fi
COMMIT_HASH=$(cut -f3 -d. <<< $RELEASE)
if [ $CORE_FROM_CI -eq 1 ]
then
COMMIT_HASH=${MAIN_BRANCH}
fi
if [ "$(grep -o ~dev <<< $VERSION)" == "~dev" ]
then
BRANCH=${MAIN_BRANCH}
JENKINS_WORKSPACE_DIR=/jenkins/workspace/scylla-${MAIN_BRANCH}/next
else
BASE_VERSION=$(grep -o "^[0-9]\+\.[0-9]\+" <<< $VERSION)
BRANCH=branch-${BASE_VERSION}
JENKINS_WORKSPACE_DIR=/jenkins/workspace/scylla-${BASE_VERSION}/build
fi
if ! [[ -d ${SCYLLA_REPO_PATH} ]]
then
log "Cloning ${PRODUCT}.git"
git clone ${GIT_QUIET_FLAG} -b ${BRANCH} git@github.com:scylladb/${PRODUCT}.git ${SCYLLA_REPO_PATH}
REMOTE_REPO_NAME=origin
else
log "${PRODUCT}.git already cloned"
REMOTE_REPO_NAME=$(get_repo_name ${BRANCH} ${COMMIT_HASH})
fi
# We do the checkout unconditionally, it is cheap anyway.
pushd ${SCYLLA_REPO_PATH} > /dev/null
# we should already have the commit in the local repo, so no need to fetch it
git checkout -q ${COMMIT_HASH}
# Skip the other submodules, they are not needed for debugging
git submodule -q sync
git submodule update -q --depth=1 --init seastar
popd > /dev/null
if ! [[ -f ${COREDIR}/scylla-gdb.py ]]
then
if [[ "${SCYLLA_GDB_PY_SOURCE}" == "repo" ]]
then
WORKDIR=$(pwd)
cd ${SCYLLA_REPO_PATH}
git checkout -q $MAIN_BRANCH
git pull -q --no-recurse-submodules ${REMOTE_REPO_NAME} $MAIN_BRANCH
log "Copying scylla-gdb.py from ${SCYLLA_REPO_PATH}"
cp scylla-gdb.py ${WORKDIR}/scylla-gdb.py
git checkout -q ${COMMIT_HASH}
cd $WORKDIR
elif [[ "${SCYLLA_GDB_PY_SOURCE}" == "package" ]]
then
log "Copying scylla-gdb.py from package"
cp -n ${ARTIFACT_DIR}/scylla.package/scylla-gdb.py .
else
log "scylla-gdb.py was not requested"
fi
else
log "scylla-gdb.py already present"
fi
if [[ ${VERBOSE_LEVEL} -ge 1 ]]
then
cat << EOF
Launching dbuild container.
To examine the coredump with gdb:
$ gdb -x scylla-gdb.py -ex 'set directories /src/scylla' \
-iex 'set substitute-path ${JENKINS_WORKSPACE_DIR}/scylla /src/scylla' \
-iex 'set solib-search-path /opt/scylladb/libreloc/' \
--core ${COREFILE} /opt/scylladb/libexec/scylla
See https://github.com/scylladb/scylladb/blob/master/docs/dev/debugging.md for more information on how to debug scylla.
Good luck!
EOF
fi
TOP_SRCDIR="$(dirname "$(dirname "$(readlink -f "$0")")")"
IMAGE="${SCYLLA_REPO_PATH}/tools/toolchain/image"
exec ${TOP_SRCDIR}/tools/toolchain/dbuild \
--image "$(<"$IMAGE")" \
-it \
-v $(pwd):/workdir \
-v ${SCYLLA_REPO_PATH}:/src/scylla \
-v ${ARTIFACT_DIR}/scylla.package:/opt/scylladb \
-w /workdir -- bash -l