Skip to content

seaworld008/centos7-lighthouse / skills/centos7-almalinux-upgrade

centos7-almalinux-upgrade

Flagged

把老旧 CentOS 7 / CentOS 8(及 RHEL7/8 系)服务器原地升级到最新 AlmaLinux 10(经 ELevate/Leapp 7→8→9→10,CentOS 8 先用 almalinux-deploy 转换),并保证业务、容器、nginx、kubectl 等升级后照常工作。凡是用户提到 CentOS 7 / CentOS 停服 / EOL / 老系统升级 / 迁移到 AlmaLinux / Rocky / CentOS 继任者 / ELevate / leapp / 系统漏洞要升级系统、批量升级阿里云/华为云/AWS 上的 CentOS 服务器,都要使用本技能——即使用户没说"升级"而只说"系统太老了""换个新系统但业务不能停"。In-place upgrade CentOS 7 to AlmaLinux 10 safely, single server or whole fleet.

FileKindSize
SKILL.mdskill md14.7 KB
assets/fleet-tracker.csvother617 B
assets/report-template.mdmarkdown850 B
references/app-compat.mdmarkdown14.5 KB
references/cloud-providers.mdmarkdown5.7 KB
references/fleet.mdmarkdown2.6 KB
references/kubernetes.mdmarkdown7.6 KB
references/pitfalls.mdmarkdown51.0 KB
references/recovery.mdmarkdown8.3 KB
scripts/backup.shexecutablescript5.0 KB
scripts/convert-el8.shexecutablescript7.6 KB
scripts/finalize.shexecutablescript12.5 KB
scripts/fleet_summary.pyexecutablescript7.9 KB
scripts/hop.shexecutablescript32.6 KB
scripts/k8s.shscript16.2 KB
scripts/known-hosts-fill.shscript2.7 KB
scripts/mysql-backup.shscript9.2 KB
scripts/nginx-static-build.shexecutablescript5.1 KB
scripts/post-audit.shexecutablescript12.4 KB
scripts/prep-el7.shexecutablescript13.5 KB
scripts/remote.pyexecutablescript9.7 KB
scripts/rescue-grubenv.shscript3.2 KB
scripts/state.shexecutablescript5.2 KB
scripts/survey.shexecutablescript28.6 KB
scripts/swap-openssh.shexecutablescript3.6 KB
scripts/verify.shexecutablescript5.3 KB

scripts/finalize.sh

View on GitHub
#!/bin/bash
# finalize.sh - 到达 EL10 后的收尾/现代化工具(按需选用,每个子命令独立)
#
#   bash finalize.sh compose-wrapper          docker-compose v1 → 转发到 "docker compose" 的包装脚本(原文件备份)
#   bash finalize.sh kubectl [auto|1.30]      kubectl 换成 pkgs.k8s.io 源、与集群同 minor 版本(auto=按 kubeconfig 集群版本)
#   bash finalize.sh helm                     helm 升级到最新 v3(sha256 校验)
#   bash finalize.sh postfix-lmdb             postfix 的 hash:/btree: 表改 lmdb:(EL10 无 libdb)并重建
#   bash finalize.sh nginx-repo               添加 nginx.org 官方 stable 源并安装 nginx(不改配置、不启动)
#   bash finalize.sh cgroupv1-shim <compose目录>  给需要 cgroup v1 内存文件的老应用挂只读兼容文件并重建容器
#   bash finalize.sh iptables-legacy         加载并持久化 legacy iptables 表模块(容器内用 iptables-legacy 的镜像,如 wg-easy)
#   bash finalize.sh redis-valkey            redis → valkey(EL10 无 redis):迁移 dump/aof、启用 valkey、redis-cli 兼容链接
#   bash finalize.sh manual-procs <DIR>      把升级前手工启动的进程(backup.sh 记录在 <BACKUP>/manual-procs)生成 systemd 服务
#   bash finalize.sh sysv-units              为仍靠 SysV init 脚本启动的服务生成原生 systemd unit(防止将来 systemd 去掉 sysv 支持)
#   bash finalize.sh oom-policy              设了内存上限的 unit:MemoryLimit→MemoryMax,加 OOMPolicy=continue + Restart=on-failure(EL9+ 默认 OOMPolicy=stop)
export LC_ALL=C
PATH=/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin
K=/root/os-upgrade-kit/bak; mkdir -p $K
MIR=${MIRROR:-https://mirrors.aliyun.com}   # 环境变量 MIRROR=upstream 表示用官方源

case "$1" in
compose-wrapper)
  f=$(command -v docker-compose)
  if [ -n "$f" ] && ! grep -q "exec docker compose" "$f" 2>/dev/null; then
    v=$("$f" version --short 2>/dev/null); case "$v" in 1.*) cp -a "$f" "$K/docker-compose-$v"; echo "备份 v$v → $K";; *) echo "docker-compose 不是 v1($v),不处理"; exit 0;; esac
  else f=/usr/local/bin/docker-compose; fi
  docker compose version >/dev/null 2>&1 || { has_dnf=$(command -v dnf || command -v yum); $has_dnf -y install docker-compose-plugin >/dev/null 2>&1; }
  # 没有 compose v2 插件时绝不能替换(真实踩坑:EL7 上 docker 19.03 没有插件,替换后 docker-compose 直接不可用)
  if ! docker compose version >/dev/null 2>&1; then
    echo "!! 'docker compose'(v2 插件)不可用:保留原 docker-compose,不做替换。升级到带插件的 docker-ce 后再执行本命令"; exit 1
  fi
  printf '#!/bin/sh\n# docker-compose v1 与 Docker>=25 不兼容(KeyError: ContainerConfig),转发到 Compose v2 插件\nexec docker compose "$@"\n' > "$f"; chmod 755 "$f"
  docker-compose version ;;

kubectl)
  want=$2
  if [ -z "$want" ] || [ "$want" = auto ]; then
    want=$(timeout 20 kubectl version -o json 2>/dev/null | grep -A3 '"serverVersion"' | grep -oE '"minor": *"[0-9]+' | grep -oE '[0-9]+$')
    [ -n "$want" ] && want=1.$want || { echo "取不到集群版本,请指定如: finalize.sh kubectl 1.30"; exit 1; }
  fi
  if [ "$MIR" = upstream ]; then base=https://pkgs.k8s.io/core:/stable:/v$want/rpm; else base=$MIR/kubernetes-new/core/stable/v$want/rpm; fi
  curl -sfI -m 15 "$base/repodata/repomd.xml" >/dev/null || base=https://pkgs.k8s.io/core:/stable:/v$want/rpm
  printf "[kubernetes]\nname=Kubernetes v%s\nbaseurl=%s/\nenabled=1\ngpgcheck=1\ngpgkey=%s/repodata/repomd.xml.key\n" "$want" "$base" "$base" > /etc/yum.repos.d/kubernetes.repo
  cp -a "$(command -v kubectl)" "$K/kubectl.old" 2>/dev/null
  rpm -q kubectl >/dev/null 2>&1 && ! rpm -q --qf '%{VERSION}' kubectl | grep -q "^$want\." && rpm -e kubectl
  [ -f /usr/local/bin/kubectl ] && mv /usr/local/bin/kubectl "$K/kubectl.usr-local"
  dnf -y install kubectl 2>&1 | tail -1; hash -r
  kubectl version --client 2>&1 | head -1; timeout 20 kubectl get nodes 2>&1 | head -5
  rpm -q bash-completion >/dev/null || dnf -y install bash-completion >/dev/null 2>&1 ;;

helm)
  v=$(curl -s -m 20 "https://api.github.com/repos/helm/helm/releases?per_page=30" | grep -oE '"tag_name": "v3\.[0-9]+\.[0-9]+"' | head -1 | grep -oE 'v3[0-9.]+')
  [ -z "$v" ] && { echo "取不到 helm 最新版本"; exit 1; }
  d=$(mktemp -d); cd "$d" || exit 1
  curl -fsSL -m 300 -o h.tgz "https://get.helm.sh/helm-$v-linux-amd64.tar.gz" && curl -fsSL -m 60 -o h.sha "https://get.helm.sh/helm-$v-linux-amd64.tar.gz.sha256sum"
  echo "$(cut -d' ' -f1 h.sha)  h.tgz" | sha256sum -c - || exit 1
  tar xzf h.tgz; old=$(command -v helm); [ -n "$old" ] && cp -a "$old" "$K/helm.old"
  install -m 755 linux-amd64/helm "${old:-/usr/local/bin/helm}"; cd / && rm -rf "$d"
  helm version --short ;;

postfix-lmdb)
  postconf -n | grep -E "(hash|btree):" | sed 's/^/转换前: /'
  for k in $(postconf -n | awk -F' = ' '$2 ~ /(hash|btree):/{print $1}'); do
    v=$(postconf -h "$k" | sed -E 's/(hash|btree):/lmdb:/g'); postconf -e "$k = $v"
  done
  postconf -e "default_database_type = lmdb" 2>/dev/null
  newaliases; for f in $(postconf -n | grep -oE "lmdb:[^ ,]+" | sed 's/lmdb://' | sort -u); do [ "$f" != /etc/aliases ] && [ -f "$f" ] && postmap "lmdb:$f"; done
  # 禁用了 IPv6 时 inet_protocols=all 会让 postfix 报 "no local interface found for ::1"(真实踩坑)
  ip -6 addr show lo 2>/dev/null | grep -q "::1" || { [ "$(postconf -h inet_protocols)" = all ] && postconf -e inet_protocols=ipv4 && echo "IPv6 已禁用 → inet_protocols=ipv4"; }
  systemctl restart postfix; systemctl is-active postfix
  echo test | sendmail root && sleep 2 && journalctl -u postfix --since -1min --no-pager -o cat | grep -E "status=" | tail -1 ;;

nginx-repo)
  cat > /etc/yum.repos.d/nginx.repo <<'EOF'
[nginx-stable]
name=nginx stable repo
baseurl=https://nginx.org/packages/rhel/$releasever_major/$basearch/
gpgcheck=1
enabled=1
gpgkey=https://nginx.org/keys/nginx_signing.key
module_hotfixes=true
EOF
  dnf -q makecache --repo nginx-stable 2>/dev/null || sed -i 's#\$releasever_major#10#' /etc/yum.repos.d/nginx.repo
  cp -a /etc/nginx "$K/etc-nginx.before-pkg" 2>/dev/null
  dnf -y install nginx --repo nginx-stable 2>&1 | grep -E "nginx|Complete|Error" | head -3
  /usr/sbin/nginx -v
  echo "已安装(未启动)。按 references/app-compat.md 的 nginx 迁移步骤写 /etc/nginx/conf.d 后再切换" ;;

cgroupv1-shim)
  d=${2:?需要 compose 目录}; cd "$d" || exit 1
  f=$(ls docker-compose.y*ml compose.y*ml 2>/dev/null | head -1); [ -n "$f" ] || { echo "没有 compose 文件"; exit 1; }
  cp -a "$f" "$f.bak-before-cgroupv2"
  mkdir -p cgroupv1-shim/memory; echo 9223372036854771712 > cgroupv1-shim/memory/memory.limit_in_bytes; echo 0 > cgroupv1-shim/memory/memory.usage_in_bytes
  echo "在 $f 的目标 service 下加入(若已有 volumes 就追加一行):"
  echo "    volumes:"; echo "      - ./cgroupv1-shim:/sys/fs/cgroup:ro"
  echo "然后: docker compose up -d   (不要用 docker-compose v1)" ;;

sysv-units)
  for s in $(ls /etc/rc.d/init.d/ 2>/dev/null | grep -vE "^(functions|README|netconsole|network)$"); do
    [ -f /etc/systemd/system/$s.service ] && continue
    systemctl is-enabled "$s" 2>/dev/null | grep -q enabled || { echo "跳过未启用的 $s"; continue; }
    cat > /etc/systemd/system/$s.service <<EOF
[Unit]
Description=$s (wrapped SysV script /etc/rc.d/init.d/$s)
After=network-online.target
Wants=network-online.target
[Service]
Type=forking
RemainAfterExit=yes
TimeoutStartSec=300
ExecStart=/etc/rc.d/init.d/$s start
ExecStop=/etc/rc.d/init.d/$s stop
[Install]
WantedBy=multi-user.target
EOF
    echo "生成 /etc/systemd/system/$s.service"
  done
  systemctl daemon-reload; echo "检查后执行 systemctl enable <name>;当前正在运行的不要 restart,下次开机生效" ;;

iptables-legacy) # 容器里用 iptables-legacy 的镜像(如 wg-easy):EL10 默认不加载 legacy 表 → "can't initialize iptables table nat"(真实踩坑)
  printf 'ip_tables\niptable_nat\niptable_filter\niptable_mangle\nip6_tables\nip6table_nat\nip6table_filter\n' > /etc/modules-load.d/iptables-legacy-for-containers.conf
  while read -r m; do modprobe "$m" 2>/dev/null; done < /etc/modules-load.d/iptables-legacy-for-containers.conf
  echo "已加载并持久化 legacy iptables 模块:$(lsmod | grep -cE '^(ip|ip6)table_') 个表模块;重启相关容器(docker restart <name>)" ;;

redis-valkey) # EL10 没有 redis 包,leapp 换成 valkey 但不迁移数据/不启用(真实踩坑:redis.service 消失)
  rpm -q valkey >/dev/null 2>&1 || dnf -y install valkey 2>&1 | tail -1
  old=$(ls /etc/redis/redis.conf.rpmsave /etc/redis.conf.rpmsave /etc/redis/redis.conf /etc/redis.conf 2>/dev/null | head -1)
  [ -n "$old" ] && { echo "旧配置 $old 的关键项(核对后按需写进 /etc/valkey/valkey.conf):"; grep -E "^(bind|port|requirepass|maxmemory|appendonly|save|dir|dbfilename) " "$old" | sed -E 's/^(requirepass ).*/\1***/'; }
  for f in /var/lib/redis/dump.rdb /var/lib/redis/appendonly.aof; do [ -f "$f" ] && [ ! -f "/var/lib/valkey/$(basename $f)" ] && install -o valkey -g valkey -m 640 "$f" /var/lib/valkey/ && echo "数据 $f → /var/lib/valkey/"; done
  grep -qE "^requirepass " "$old" 2>/dev/null && echo "!! 旧配置有 requirepass:请把同样的密码写进 /etc/valkey/valkey.conf 再启动"
  for b in cli server benchmark; do [ -e /usr/local/bin/redis-$b ] || ln -s /usr/bin/valkey-$b /usr/local/bin/redis-$b; done
  systemctl enable --now valkey 2>&1 | tail -1; sleep 2; redis-cli ping 2>&1 | head -1 ;;

manual-procs) # 升级前手工从 SSH 会话启动(nohup ... &)的进程,重启后不会自己起来:按 backup.sh 记录的 cmdline/cwd/环境变量生成 systemd 服务
  D=${2:?用法: finalize.sh manual-procs <BACKUP_DIR>/manual-procs}
  for c in "$D"/*.cmdline; do
    [ -e "$c" ] || continue
    p=${c%.cmdline}; cwd=$(cat "$p.cwd"); exe=$(cat "$p.exe" 2>/dev/null)
    name=osup-$(echo "$cwd" | sed -E 's#^/(data|opt|home|root)/##; s#/bin$##; s#/.*/#-#; s#[^A-Za-z0-9_.]+#-#g')
    [ -e "/etc/systemd/system/$name.service" ] && { echo "已存在 $name,跳过"; continue; }
    first=$(head -1 "$c"); case "$first" in /*) ;; *) [ -n "$exe" ] && first=$exe;; esac
    args=$(tail -n +2 "$c" | sed 's/%/%%/g' | awk '{gsub(/"/,"\\\""); printf " \"%s\"", $0}')
    envs=$(grep -E '^(PATH|JAVA_HOME|LANG|LC_ALL|CLASSPATH)=' "$p.environ" 2>/dev/null | sed 's/^/Environment="/; s/$/"/')
    printf '[Unit]\nDescription=Restored manual process in %s (was started by hand before the OS upgrade)\nAfter=network-online.target docker.service\nWants=network-online.target\n\n[Service]\nType=simple\nWorkingDirectory=%s\n%s\nExecStart=%s%s\nSuccessExitStatus=143\nTimeoutStopSec=60\n\n[Install]\nWantedBy=multi-user.target\n' \
      "$cwd" "$cwd" "$envs" "$first" "$args" > "/etc/systemd/system/$name.service"
    echo "生成 $name: $first $(tail -n +2 "$c" | tr '\n' ' ' | cut -c1-80)"
  done
  systemctl daemon-reload
  echo "检查 /etc/systemd/system/osup-*.service 后:systemctl enable --now <name>(LANG=zh_CN.GB18030 等需要 glibc-langpack-zh)" ;;

oom-policy) # EL7 上内存超限只杀一个进程(mysqld_safe 等守护会拉起来);EL9+ 默认 OOMPolicy=stop 停掉整个 unit,SysV 生成的 unit 又是 Restart=no(真实踩坑:MySQL 停了 10 小时)
  D=${SYSTEMD_DIR:-/etc/systemd/system}; n=0
  for f in "$D"/*.service "$D"/*.service.d/*.conf; do
    [ -e "$f" ] || continue
    grep -qE '^(MemoryLimit|MemoryMax)=' "$f" || continue
    u=${f#"$D"/}; u=${u%%.d/*}
    grep -qsE '^OOMPolicy=' "$D/$u" "$D/$u".d/*.conf && continue
    cp -a "$f" "$K/$(echo "$f" | tr / _).bak-oom-policy"
    sed -i -E 's/^MemoryLimit=/MemoryMax=/' "$f"     # cgroup v2 下 MemoryLimit= 已废弃
    mkdir -p "$D/$u.d"; o="$D/$u.d/90-oom-policy.conf"
    printf '# finalize.sh oom-policy: EL9+ 默认 OOMPolicy=stop,OOM 杀一个进程就停掉整个 unit\n[Service]\nOOMPolicy=continue\n' > "$o"
    # 没设 Restart 的(SysV 生成的 unit 默认 Restart=no)补上 on-failure
    grep -qsE '^Restart=' "$D/$u" "$D/$u".d/*.conf "/usr/lib/systemd/system/$u" || printf 'Restart=on-failure\nRestartSec=10\n' >> "$o"
    echo "已修改 $u:$(cat "$f" "$o" | grep -E '^(MemoryMax|OOMPolicy|Restart)=' | tr '\n' ' ')(原文件备份在 $K)"; n=$((n+1))
  done
  [ $n -gt 0 ] && { systemctl daemon-reload 2>/dev/null; echo "已 daemon-reload;内存上限按实际占用调整(例如 MySQL:buffer pool + 连接数 × 每连接缓冲,留余量)"; }
  [ $n = 0 ] && echo "没有需要修改的 unit" ;;

*) sed -n '2,16p' "$0";;
esac
writing-skillsobra/superpowersUse when creating new skills, editing existing skills, or verifying skills work before deploymentStars: 322kUpdated Claude CodeCodexVerifiedInstall
xlsxanthropics/skillsUse this skill any time a spreadsheet file is the primary input or output. This means any task where the user wants to: open, read, edit, or fix an existing .xlsx, .xlsm, .xltx, .csv, or .tsv file (e.g., adding columns, computing formulas, formatting, charting, cleaning messy data); create a new spreadsheet from scratch or from other data sources; or convert between tabular file formats. Trigger especially when the user references a spreadsheet file by name or path — even casually (like "the xlsx in my downloads") — and wants something done to it or produced from it. Also trigger for cleaning or restructuring messy tabular data files (malformed rows, misplaced headers, junk data) into proper spreadsheets. The deliverable must be a spreadsheet file. Do NOT trigger when the primary deliverable is a Word document, HTML report, standalone Python script, database pipeline, or Google Sheets API integration, even if tabular data is involved.Stars: 198kUpdated Claude CodeCodexVerifiedInstall
shipping-and-launchaddyosmani/agent-skillsPrepares production launches. Use when preparing to deploy to production, or when asking what needs to be in place before shipping. Use when you need a pre-launch checklist, when setting up monitoring, when planning a staged rollout, or when you need a rollback strategy.Stars: 119kUpdated Claude CodeCodexVerifiedInstall
childhoodthedotmack/claude-memBabysit a shipped release after merge, publication, or deployment. Verify published packages, production behavior, installed plugins, and late CI or user reports until the observation window is clean; use babysit for PR readiness before release.Stars: 118kUpdated Claude CodeCodexVerifiedInstall