fix(deploy): 补齐集群备份兼容与切换前门禁
阿里云 OSS 的 S3 兼容层要求 virtual-hosted addressing,且 boto3 需要使用 Signature V2。本提交为 Barman 的归档、备份、保留与恢复统一挂载 AWS 配置,并将实际 WAL 归档和近期全备设为切换硬门禁。 同时补充 CNPG/etcdutl 固定版本安装、显式主库切换、三节点逐台停机、OSS 快照下载校验与远端临时 Secret 清理。已通过 ShellCheck、Kustomize、服务端 dry-run、迁移测试、发布脚本测试、敏感信息扫描及生产 precutover 验收。
This commit is contained in:
@@ -6,29 +6,55 @@ script_dir=$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)
|
||||
source "$script_dir/common.sh"
|
||||
load_cluster_env
|
||||
|
||||
cluster_scp "$script_dir/oss-object.mjs" \
|
||||
"$CLUSTER_NINGBO_HOST:/root/easyai-oss-object.mjs"
|
||||
cluster_ssh "$CLUSTER_NINGBO_HOST" 'bash -s' <<'REMOTE'
|
||||
set -euo pipefail
|
||||
working_directory=$(mktemp -d)
|
||||
cleanup() {
|
||||
find "$working_directory" -type f -delete
|
||||
find "$working_directory" -depth -type d -empty -delete
|
||||
[[ ! -e /root/easyai-oss-object.mjs && ! -L /root/easyai-oss-object.mjs ]] ||
|
||||
unlink /root/easyai-oss-object.mjs
|
||||
}
|
||||
trap cleanup EXIT HUP INT TERM
|
||||
command -v etcdutl >/dev/null
|
||||
command -v unzip >/dev/null
|
||||
command -v node >/dev/null
|
||||
|
||||
snapshot_name=acceptance-$(date -u '+%Y%m%dT%H%M%SZ')
|
||||
k3s etcd-snapshot save --name "$snapshot_name" >/dev/null
|
||||
k3s etcd-snapshot save --name "$snapshot_name" >/dev/null 2>&1
|
||||
snapshot=$(find /var/lib/rancher/k3s/server/db/snapshots -maxdepth 1 -type f \
|
||||
-name "$snapshot_name-*" -printf '%T@ %p\n' | sort -rn | awk 'NR == 1 { sub(/^[^ ]+ /, ""); print }')
|
||||
[[ -n $snapshot && -s $snapshot ]]
|
||||
etcdutl=$(find /var/lib/rancher/k3s/data -type f -path '*/bin/etcdutl' -print -quit)
|
||||
[[ -x $etcdutl ]]
|
||||
working_directory=$(mktemp -d)
|
||||
trap 'rm -rf -- "$working_directory"' EXIT
|
||||
case $snapshot in
|
||||
*.zip)
|
||||
if command -v unzip >/dev/null 2>&1 && unzip -t "$snapshot" >/dev/null 2>&1; then
|
||||
unzip -p "$snapshot" >"$working_directory/snapshot.db"
|
||||
else
|
||||
gzip -dc "$snapshot" >"$working_directory/snapshot.db"
|
||||
fi
|
||||
;;
|
||||
*) cp "$snapshot" "$working_directory/snapshot.db" ;;
|
||||
esac
|
||||
"$etcdutl" snapshot status "$working_directory/snapshot.db" -w json |
|
||||
snapshot_file=$(basename "$snapshot")
|
||||
s3_location=$(k3s etcd-snapshot ls -o json 2>/dev/null |
|
||||
jq -r --arg name "s3-$snapshot_file" \
|
||||
'.items[] | select(.metadata.name == $name and .status.readyToUse == true) |
|
||||
.spec.location')
|
||||
|
||||
export ALI_KEY ALI_SECRET ALI_REGION ALI_BUCKET
|
||||
ALI_KEY=$(k3s kubectl get secret easyai-etcd-s3 -n kube-system \
|
||||
-o jsonpath='{.data.etcd-s3-access-key}' | base64 -d)
|
||||
ALI_SECRET=$(k3s kubectl get secret easyai-etcd-s3 -n kube-system \
|
||||
-o jsonpath='{.data.etcd-s3-secret-key}' | base64 -d)
|
||||
ALI_REGION=$(k3s kubectl get secret easyai-etcd-s3 -n kube-system \
|
||||
-o jsonpath='{.data.etcd-s3-region}' | base64 -d)
|
||||
ALI_BUCKET=$(k3s kubectl get secret easyai-etcd-s3 -n kube-system \
|
||||
-o jsonpath='{.data.etcd-s3-bucket}' | base64 -d)
|
||||
[[ $s3_location == \
|
||||
"s3://$ALI_BUCKET/easyai-ai-gateway/production/k3s-etcd/$snapshot_file" ]]
|
||||
object_key=${s3_location#s3://"$ALI_BUCKET"/}
|
||||
node /root/easyai-oss-object.mjs get "$object_key" \
|
||||
"$working_directory/s3-snapshot.zip" >/dev/null
|
||||
[[ $(sha256sum "$snapshot" | awk '{print $1}') == \
|
||||
$(sha256sum "$working_directory/s3-snapshot.zip" | awk '{print $1}') ]]
|
||||
|
||||
archive_entry=$(unzip -Z1 "$working_directory/s3-snapshot.zip")
|
||||
[[ -n $archive_entry && $archive_entry != *$'\n'* ]]
|
||||
unzip -p "$working_directory/s3-snapshot.zip" "$archive_entry" \
|
||||
>"$working_directory/snapshot.db"
|
||||
etcdutl snapshot status "$working_directory/snapshot.db" -w json |
|
||||
jq -e '.hash != null and .totalKey > 0 and .totalSize > 0' >/dev/null
|
||||
k3s etcd-snapshot ls --s3 | grep -F "$snapshot_name" >/dev/null
|
||||
echo "etcd_backup_verification=PASS snapshot=$snapshot_name local=true oss=true"
|
||||
echo "etcd_backup_verification=PASS snapshot=$snapshot_file local=true oss_download=true status=true"
|
||||
REMOTE
|
||||
|
||||
Reference in New Issue
Block a user