feat(backup): log everything stdout/err, implement lockfile for both …

…backup/restore

backup/pvc/VERSION.txt

@@ -1 +1 @@
-v0.3.0
+v0.4.0

backup/pvc/bin/backup.sh

@@ -1,39 +1,64 @@
 #!/usr/bin/env bash
 
 set -eo pipefail
+source "$(dirname "$0")/utils.sh"
+set -x
 
-[[ ! $# -eq 1 ]] && echo "Usage: $0 backup_number" && exit 1;
-[[ -z "${BACKUP_DIR}" ]] && echo "Required 'BACKUP_DIR' env not set" && exit 1;
-[[ -z "${JENKINS_HOME}" ]] && echo "Required 'JENKINS_HOME' env not set" && exit 1;
+[[ ! $# -eq 1 ]] && _log "ERROR" "Usage: $0 BACKUP_NUMBER" && exit 1
+[[ -z "${BACKUP_DIR}" ]] && _log "ERROR" "Required 'BACKUP_DIR' env not set" && exit 1
+[[ -z "${JENKINS_HOME}" ]] && _log "ERROR" "Required 'JENKINS_HOME' env not set" && exit 1
+RETRY_COUNT=${RETRY_COUNT:-3}
+RETRY_INTERVAL=${RETRY_INTERVAL:-60}
+BACKUP_NUMBER=$1
+TRAP_FILE="${BACKUP_DIR}/_backup_${BACKUP_NUMBER}_is_running"
 
+# --> Check if another backup process is running (operator restart/crash)
+for ((i=0; i<RETRY_COUNT; i++)); do
+    [[ ! -f "${TRAP_FILE}" ]] && _log "INFO" "Backup: no other backup process are running" && break
+    _log "INFO" "Backup is already running. Waiting for ${RETRY_INTERVAL} seconds..."
+    sleep "${RETRY_INTERVAL}"
+done
+[[ -f "${TRAP_FILE}" ]] && { _log "ERROR" "Backup is stil running after waiting ${RETRY_COUNT} time ${RETRY_INTERVAL}s. Exiting."; exit 1; }
+# --< Done
+
+_log "INFO" "Running backup ${BACKUP_NUMBER}"
+touch "${TRAP_FILE}"
 # create temp dir on the same filesystem with a BACKUP_DIR to be able use atomic mv enstead of copy
-BACKUP_TMP_DIR=$(mktemp -d --tmpdir=${BACKUP_DIR})
-trap "test -d "${BACKUP_TMP_DIR}" && rm -fr "${BACKUP_TMP_DIR}"" EXIT SIGINT SIGTERM
+BACKUP_TMP_DIR=$(mktemp -d --tmpdir="${BACKUP_DIR}")
+
+_clean(){
+    test -d "${BACKUP_TMP_DIR}" && rm -fr "${BACKUP_TMP_DIR}"
+    test -f "${TRAP_FILE}" && rm -f "${TRAP_FILE}"
+}
+
+_trap(){
+    _clean
+    _log "ERROR" "Backup: something wrong happened, check the logs"
+}
 
-backup_number=$1
-echo "Running backup"
+trap '_trap' SIGQUIT SIGINT SIGTERM
 
 # config.xml in a job directory is a config file that shouldn't be backed up
 # config.xml in child directories is state that should. For example-
 # branches/myorg/branches/myrepo/branches/master/config.xml should be retained while
 # branches/myorg/config.xml should not
-tar --zstd -C "${JENKINS_HOME}" -cf "${BACKUP_TMP_DIR}/${backup_number}.tar.zstd" \
+tar --zstd -C "${JENKINS_HOME}" -cf "${BACKUP_TMP_DIR}/${BACKUP_NUMBER}.tar.zstd" \
     --exclude jobs/*/workspace* \
     --no-wildcards-match-slash --anchored \
     --ignore-failed-read \
     --exclude jobs/*/config.xml -c jobs || ret=$?
 
 if [[ "$ret" -eq 0 ]]; then
-  echo "Backup was completed without warnings"
+  _log "INFO" "Backup ${BACKUP_NUMBER} was completed without warnings"
 elif [[ "$ret" -eq 1 ]]; then
-  echo "Backup was completed with some warnings"
+  _log "INFO" "Backup ${BACKUP_NUMBER} was completed with some warnings"
 fi
 
-# atomically create a backup file
-mv "${BACKUP_TMP_DIR}/${backup_number}.tar.zstd" "${BACKUP_DIR}/${backup_number}.tar.zstd"
+mv "${BACKUP_TMP_DIR}/${BACKUP_NUMBER}.tar.zstd" "${BACKUP_DIR}/${BACKUP_NUMBER}.tar.zstd"
 
-rm -rf "${BACKUP_TMP_DIR}"
-[[ ! -s ${BACKUP_DIR}/${backup_number}.tar.zstd ]] && echo "backup file '${BACKUP_DIR}/${backup_number}.tar.zstd' is empty" && exit 1;
+_log "INFO" "Cleaning ${BACKUP_TMP_DIR} and trap file ${TRAP_FILE}"
+_clean
+[[ ! -s ${BACKUP_DIR}/${BACKUP_NUMBER}.tar.zstd ]] && _log "ERROR" "Backup file '${BACKUP_DIR}/${BACKUP_NUMBER}.tar.zstd' is empty" && exit 1
 
-echo Done
+_log "INFO" "Backup ${BACKUP_NUMBER} done"
 exit 0

backup/pvc/bin/get-latest.sh

@@ -1,30 +1,26 @@
 #!/usr/bin/env bash
 
 set -eo pipefail
+source "$(dirname "$0")/utils.sh"
 
 is_backup_not_exist() {
     local backup_dir="$1"
     # Save the current value of 'set -e'
     local previous_e
     previous_e=$(set +e; :; echo $?)
-
     # Temporarily turn off 'set -e'
     set +e
-
     # Run ls command to check if any files matching the pattern exist
     ls "${backup_dir}"/*.tar.* 1> /dev/null 2>&1
-
     # Store the exit status of the ls command
     local ls_exit_status=$?
-
     # Restore the previous value of 'set -e'
     [ "$previous_e" = "0" ] && set -e
-
     # Return true if ls command succeeded (no files found), otherwise return false
     [ $ls_exit_status -ne 0 ]
 }
 
-[[ -z "${BACKUP_DIR}" ]] && { echo "Required 'BACKUP_DIR' env not set"; exit 1; }
+[[ -z "${BACKUP_DIR}" ]] && { _log "ERROR" "Required 'BACKUP_DIR' env not set"; exit 1; }
 
 # Check if we have any backup
 if is_backup_not_exist "${BACKUP_DIR}"; then

backup/pvc/bin/restore.sh

@@ -1,29 +1,45 @@
 #!/usr/bin/env bash
 
 set -eo pipefail
+source "$(dirname "$0")/utils.sh"
 
-[[ ! $# -eq 1 ]] && echo "Usage: $0 backup_number" && exit 1
-[[ -z "${BACKUP_DIR}" ]] && echo "Required 'BACKUP_DIR' env not set" && exit 1;
-[[ -z "${JENKINS_HOME}" ]] && echo "Required 'JENKINS_HOME' env not set" && exit 1;
+[[ ! $# -eq 1 ]] && _log "ERROR" "Usage: $0 <backup number>" && exit 1
+[[ -z "${BACKUP_DIR}" ]] && _log "ERROR" "Required 'BACKUP_DIR' env not set" && exit 1
+[[ -z "${JENKINS_HOME}" ]] && _log "ERROR" "Required 'JENKINS_HOME' env not set" && exit 1
+BACKUP_NUMBER=$1
+RETRY_COUNT=${RETRY_COUNT:-3}
+RETRY_INTERVAL=${RETRY_INTERVAL:-60}
 
-backup_number=$1
-backup_file="${BACKUP_DIR}/${backup_number}"
-echo "Running restore backup with backup number #${backup_number}"
+# --> Check if another restore process is running (operator restart/crash)
+TRAP_FILE="${BACKUP_DIR}/_restore_${BACKUP_NUMBER}_is_running"
+trap "rm -f ${TRAP_FILE}" SIGINT SIGTERM
 
-if [[ -f "$backup_file.tar.gz" ]]; then
-    echo "Old format tar.gz found, restoring it"
+for ((i=0; i<RETRY_COUNT; i++)); do
+    [[ ! -f "${TRAP_FILE}" ]] && _log "INFO" "Restore: No other process are running, restoring" && break
+    _log "INFO" "Restore is already running. Waiting for ${RETRY_INTERVAL} seconds..."
+    sleep "${RETRY_INTERVAL}"
+done
+[[ -f "${TRAP_FILE}" ]] && { _log "ERROR" "Restore is stil running after waiting ${RETRY_COUNT} time ${RETRY_INTERVAL}s. Exiting."; exit 1; }
+# --< Done
+
+_log "INFO" "Running restore backup with backup number #${BACKUP_NUMBER}"
+touch "${TRAP_FILE}"
+BACKUP_FILE="${BACKUP_DIR}/${BACKUP_NUMBER}"
+
+if [[ -f "$BACKUP_FILE.tar.gz" ]]; then
+    _log "INFO" "Restore: ld format tar.gz found, restoring it"
     OPTS=""
     EXT="tar.gz"
-elif [[ -f "$backup_file.tar.zstd" ]]; then
-    echo "Backup file found, proceeding"
+elif [[ -f "$BACKUP_FILE.tar.zstd" ]]; then
+    _log "INFO" "Restore: Backup file found, proceeding"
     OPTS="--zstd"
     EXT="tar.zstd"
 else
-  echo "ERR: Backup file not found: $backup_file"
+  _log "ERROR" "Restore: Backup file not found: $BACKUP_FILE"
   exit 1
 fi
 
-tar $OPTS -C "${JENKINS_HOME}" -xf "${BACKUP_DIR}/${backup_number}.${EXT}"
+tar $OPTS -C "${JENKINS_HOME}" -xf "${BACKUP_DIR}/${BACKUP_NUMBER}.${EXT}"
 
-echo Done
+_log "INFO" "Restore: ${BACKUP_NUMBER} Done"
 exit 0

backup/pvc/bin/run.sh

@@ -1,14 +1,15 @@
 #!/usr/bin/env bash
 
 set -eo pipefail
+source "$(dirname "$0")/utils.sh"
 
 # Use 60 as default in case BACKUP_CLEANUP_INTERVAL did not set
 BACKUP_CLEANUP_INTERVAL=${BACKUP_CLEANUP_INTERVAL:=60}
 
 # Ensure required environment variables are set
 check_env_var() {
     if [[ -z "${!1}" ]]; then
-        echo "Required '$1' environment variable is not set"
+        _log "ERROR" "Required '$1' environment variable is not set"
         exit 1
     fi
 }
@@ -41,7 +42,7 @@ find_exceeding_backups() {
     local backup_count="$2"
     # Check if we have any backup
     if is_backup_not_exist "${backup_dir}"; then
-        echo "backups not found in ${backup_dir}" >&2
+        _log "ERROR" "backups not found in ${backup_dir}"
         return
     fi
     find "${backup_dir}"/*.tar.zstd -maxdepth 0 -exec basename {} \; | sort -gr | tail -n +$((backup_count +1))
@@ -51,9 +52,9 @@ check_env_var "BACKUP_DIR"
 check_env_var "JENKINS_HOME"
 
 if [[ -z "${BACKUP_COUNT}" ]]; then
-    echo "ATTENTION! No BACKUP_COUNT set, it means you MUST delete old backups manually or by custom script"
+    _log "WARNING" "No BACKUP_COUNT set, it means you MUST delete old backups manually or by custom script"
 else
-    echo "Retaining only the ${BACKUP_COUNT} most recent backups, cleanup occurs every ${BACKUP_CLEANUP_INTERVAL} seconds"
+    _log "INFO" "Retaining only the ${BACKUP_COUNT} most recent backups, cleanup occurs every ${BACKUP_CLEANUP_INTERVAL} seconds"
 fi
 
 while true;
@@ -62,7 +63,7 @@ do
     if [[ -n "${BACKUP_COUNT}" ]]; then
         exceeding_backups=$(find_exceeding_backups "${BACKUP_DIR}" "${BACKUP_COUNT}")
         if [[ -n "$exceeding_backups" ]]; then
-            echo "Removing backups: $(echo "$exceeding_backups" | tr '\n' ', ' | sed 's/,$//')"
+            _log "INFO" "Removing backups: $(echo "$exceeding_backups" | tr '\n' ', ' | sed 's/,$//')"
             echo "$exceeding_backups" | while read -r file; do
                 rm "${BACKUP_DIR}/${file}"
             done

backup/pvc/bin/utils.sh

@@ -0,0 +1,13 @@
+#!/usr/bin/env bash
+# Common utils
+
+_log() {
+    local level="$1"
+    local message="$2"
+    local timestamp=$(date '+%Y-%m-%d %H:%M:%S')
+    if [[ "$level" =~ ^(ERROR|ERR|error|err)$ ]]; then
+        echo "${timestamp} [${level}] ${message}" > /proc/1/fd/2
+    else
+        echo "${timestamp} [${level}] ${message}" > /proc/1/fd/1
+    fi
+}