NetWorker: Il backup del client NetWorker del contenitore Docker ha esito negativo con errore savefs: fail

Il software client NetWorker è installato all'interno di un container docker.
Il backup basato su client del client docker ha esito negativo con il seguente errore:

<client name>:savefs failed.
<client name>:savefs See the file '/nsr/logs/policy/<policy>/<workflow/backup_<jobid>_logs/<jobid>.log' for command output.
Stopped processing of the save set <cleint name>:<save set> because the savefs job job terminated with an error.
Job <jobid> host: <client name> savepoint: <save set> had ERROR indication(s) at completion
<client name>:<save set> abandoned.
Client '<client name>' is being skipped because no savesets of this client have been backed up as part of the backup action.
Action backup traditional 'backup' with job id <job id> is exiting with status 'failed', exit code 1

Attualmente non è supportato alcun metodo per eseguire il backup dei container Docker come client NetWorker. Se si desidera che questa funzionalità sia disponibile in NetWorker, contattare il proprio account Dell o il responsabile vendite in merito a una richiesta di miglioramento (RFE).

Per impedire a un client NetWorker di eseguire il backup dei dati Docker, è necessario usare le direttive skip sul client NetWorker per la struttura di file/cartelle Docker.

Esempio (modificare i percorsi e i processi in base alla configurazione di sistema e Docker):

<< /var/lib/docker >>
skip: *
<< /var/lib/containers >>
skip: *
<< /run >>
skip: docker docker.sock containers podman crun

Ulteriori informazioni sulle direttive NetWorker sono disponibili nella Guida all'amministrazione di NetWorker. La documentazione di NetWorker è disponibile tramite: Supporto per NetWorker | Manuali e documenti 

Per eseguire il backup dei dati del container Docker, utilizzare una delle seguenti soluzioni alternative:

Soluzione alternativa uno: 

1. Accedere all'host docker tramite SSH.
2. Dalla sessione SSH, connettersi al contenitore docker. La sintassi esatta utilizzata dipende dalla piattaforma del container:

• Docker: docker exec -it CONTAINER_NAME sh
• Podman: podman exec -it CONTAINER_NAME sh

3. Creare un fstab utilizzando il metodo mtab del server NetWorker Management Console (NMC): cp /etc/mtab /etc/fstab
4. Uscire dal container: exit
5. Eseguire il backup del client container da NetWorker.

Soluzione alternativa due:

Creare "archivi" dei dati del container docker sull'host Docker, eseguire il backup dei dati come saveset del file system dall'host Docker. Questo approccio può richiedere alcune attività di scripting e l'utilizzo di cron Lavori.

1. Creare uno script bash che esegua le seguenti operazioni:
   1. Arrestare temporaneamente il contenitore Docker. In questo modo è possibile garantire la coerenza del database nel container Docker.
   2. Comprimere il contenuto della cartella del contenitore Docker.
   3. Avviare il contenitore docker
   4. (opzionale) pulire gli archivi più vecchi di X giorni.

#!/usr/bin/env bash
set -euo pipefail
# ===== Config (tunable) =====
CONTAINER_TOOL="podman"                  # container platform (podman or docker)
CONTAINER_NAME="npm"                     # the container name to verify after restart
COMPOSE_DIR="/root/docker/nginx"         # directory containing docker-compose.yaml/podman-compose.yaml
BACKUP_ROOT="/root/docker/backups"       # where backups are stored on the host
RETENTION_DAYS=3                         # delete archives older than N days
TIMESTAMP="$(date +%F_%H%M%S)"
ARCHIVE="${BACKUP_ROOT}/${CONTAINER_NAME}-backup-${TIMESTAMP}.tar.gz"
LOCK_FILE="/var/lock/${CONTAINER_NAME}_backup.lock"
# ========================================
mkdir -p "$BACKUP_ROOT" /var/lock
# --- Handle stale lock files (if previous run crashed) ---
if [[ -f "$LOCK_FILE" ]]; then
  # Try to find an owning process; if none, remove the stale lock
  LOCK_FD_OWNER="$(lsof -t -- "$LOCK_FILE" 2>/dev/null || true)"
  if [[ -z "$LOCK_FD_OWNER" || ! -d "/proc/$LOCK_FD_OWNER" ]]; then
    echo "[INFO] Removing stale lock file: $LOCK_FILE"
    rm -f "$LOCK_FILE"
  fi
fi
# --- Simple lock to avoid overlapping runs ---
exec 9>"$LOCK_FILE"
if ! flock -n 9; then
  echo "[INFO] Another backup is running. Exiting."
  exit 0
fi
# --- Safety checks ---
if [[ ! -d "$COMPOSE_DIR" ]]; then
  echo "[ERROR] Compose directory not found: $COMPOSE_DIR" >&2
  exit 1
fi
if [[ ! -f "${COMPOSE_DIR}/docker-compose.yaml" && ! -f "${COMPOSE_DIR}/podman-compose.yaml" ]]; then
  echo "[ERROR] No compose file found in ${COMPOSE_DIR}" >&2
  exit 1
fi
# --- Functions ---
prune_backups() {
  local root="$1" days="$2"
  echo "[INFO] Pruning backups older than ${days} days in ${root}..."
  find "$root" -maxdepth 1 -type f -name "${CONTAINER_NAME}-backup-*.tar.gz" -mtime +"$days" -print -delete || true
}
verify_running() {
  # Expect container named "${CONTAINER_NAME}" to be running again
  if ${CONTAINER_TOOL} ps --format '{{.Names}}' | grep -qx "${CONTAINER_NAME}"; then
    echo "[INFO] Container '${CONTAINER_NAME}' is running."
  else
    echo "[WARN] Container '${CONTAINER_NAME}' not found in '${CONTAINER_TOOL} ps'. Check '${CONTAINER_TOOL} ps -a' and logs."
  fi
}
# --- Stop stack cleanly using *-compose in its directory ---
echo "[INFO] Bringing down stack (${CONTAINER_TOOL}-compose down) in ${COMPOSE_DIR}..."
(
  cd "$COMPOSE_DIR"
  # Do NOT pass -p; let the project default to the directory name
  ${CONTAINER_TOOL}-compose down || {
    echo "[WARN] ${CONTAINER_TOOL}-compose down returned non-zero. Continuing (stack may not have been up)."
  }
)
# --- Create archive (compose dir: data, letsencrypt, compose file, etc.) ---
echo "[INFO] Archiving ${COMPOSE_DIR} to ${ARCHIVE}..."
# Preserve SELinux/xattrs and ownership; exclude the backups dir if it's under same parent
tar -C "$(dirname "$COMPOSE_DIR")" \
  --xattrs --selinux --same-owner --numeric-owner \
  --exclude "$(basename "$BACKUP_ROOT")" \
  -czf "$ARCHIVE" "$(basename "$COMPOSE_DIR")"
# Optional integrity checksum
sha256sum "$ARCHIVE" > "${ARCHIVE}.sha256"
# --- Start stack back up ---
echo "[INFO] Starting stack (${CONTAINER_TOOL}-compose up -d)..."
(
  cd "$COMPOSE_DIR"
  ${CONTAINER_TOOL}-compose up -d
)
# --- Verify expected container is running ---
verify_running
# --- Prune according to retention ---
prune_backups "$BACKUP_ROOT" "$RETENTION_DAYS"
echo "[OK] Backup completed: ${ARCHIVE}"

[root@rhel-client01 scripts]# ./backup_npm.sh
[INFO] Removing stale lock file: /var/lock/npm_backup.lock
[INFO] Bringing down stack (podman-compose down) in /root/docker/nginx...
npm
npm
14dc3c3871a7a9716142a1011f1039258d330be1ddb49543eb7a8e257256239a
nginx_default
[INFO] Archiving /root/docker/nginx to /root/docker/backups/npm-backup-2026-02-10_155107.tar.gz...
[INFO] Starting stack (podman-compose up -d)...
be491bf43aa17f1a767075ed8b2cd55d93cdf4581ad04abe66a58f3bce0b5e74
e677dbedd5fdd92a440896dc4cae22165cb34ce704ec3ed8f7620ce243183585
npm
[INFO] Container 'npm' is running.
[INFO] Pruning backups older than 3 days in /root/docker/backups...
[OK] Backup completed: /root/docker/backups/npm-backup-2026-02-10_155107.tar.gz
[root@rhel-client01 scripts]#
[root@rhel-client01 scripts]# ls -ltr /root/docker/backups/
total 8
-rw-r--r--. 1 root root 638 Feb 10 15:51 npm-backup-2026-02-10_155107.tar.gz
-rw-r--r--. 1 root root 123 Feb 10 15:51 npm-backup-2026-02-10_155107.tar.gz.sha256

[root@rhel-client01 scripts]# ln -s /root/scripts/backup_npm.sh /etc/cron.daily/backup_npm
[root@rhel-client01 scripts]#

2. Creare un client NetWorker per il sistema che ospita i container docker. Il "Save set" è la directory utilizzata nella directory che contiene gli archivi del tarball docker:

3. Eseguire i backup del client. Il documento NetWorker Administration Guide fornisce ulteriori informazioni sulla creazione di policy di backup e sulla pianificazione. La documentazione di NetWorker è disponibile tramite: Supporto per NetWorker | Manuali e documenti 

Il backup del client contiene i dati del tarball di archivio docker:

[root@nsr ~]# mminfo -avot -q client=rhel-client01.amer.lan -r "nsavetime,savetime(20),name,sumflags,ssretent"
  save time    date     time   name                           fl  retent
...
 1770757763    02/10/2026 16:09 /root/docker/backups          cb 03/10/2026

[root@nsr ~]# nsrinfo -t 1770757763 rhel-client01.amer.lan
scanning client `rhel-client01.amer.lan' for savetime 1770757763(Tue 10 Feb 2026 04:09:23 PM EST) from the backup namespace
/root/docker/backups/npm-backup-2026-02-10_155107.tar.gz
/root/docker/backups/npm-backup-2026-02-10_155107.tar.gz.sha256
/root/docker/backups/
/root/docker/
/root/
/
6 objects found