如下为 OpenCloudOS 9.6 + Ansible 自动化安装 SeaweedFS 三节点集群方案
适配环境：
-
OS：OpenCloudOS 9.6（RHEL9兼容）
-
SeaweedFS：4.45（你当前版本
30GB 4.45） -
节点：
fsnode01 10.40.0.77 fsnode02 10.40.0.78 fsnode03 10.40.0.79 -
数据目录：
/mnt/allin_data/allin/seaweedfs
架构：
S3 Client
|
:8333
|
fsnode01
S3 + Filer
|
+-----------+-----------+
| | |
Master Master Master
:9333 :9333 :9333
:19333 :19333 :19333
| | |
Volume Volume Volume
:8080 :8080 :8080SeaweedFS Master 默认 gRPC 端口为 HTTP 端口+10000（9333→19333），多 Master Raft 依赖该通信端口。
一、Ansible目录
创建：
mkdir -p seaweedfs-install/{inventory,roles/seaweedfs/{tasks,templates,files,defaults}}目录：
seaweedfs-install
├── inventory
│ └── hosts
├── site.yml
└── roles
└── seaweedfs
├── defaults
│ └── main.yml
├── files
│ └── weed
├── tasks
│ └── main.yml
└── templates
├── master.service.j2
├── volume.service.j2
├── filer.service.j2
└── s3.service.j2二、Inventory
inventory/hosts
[seaweedfs]
fsnode01 ansible_host=10.40.0.77
fsnode02 ansible_host=10.40.0.78
fsnode03 ansible_host=10.40.0.79
[master]
fsnode01
fsnode02
fsnode03
[volume]
fsnode01
fsnode02
fsnode03
[filer]
fsnode01
[s3]
fsnode01三、变量
roles/seaweedfs/defaults/main.yml
weed_version: "4.45"
weed_bin: /usr/local/bin/weed
seaweedfs_home: /mnt/allin_data/allin/seaweedfs
master_dir: "{{ seaweedfs_home }}/master/m9333"
volume_dir: "{{ seaweedfs_home }}/volume"
master_http_port: 9333
master_grpc_port: 19333
volume_port: 8080
filer_port: 8888
s3_port: 8333
master_peers: "fsnode01:9333,fsnode02:9333,fsnode03:9333"
s3_access_key: admin
s3_secret_key: "SeaweedFS@123456"四、主Playbook
site.yml
- hosts: seaweedfs
become: yes
roles:
- seaweedfs五、安装任务
roles/seaweedfs/tasks/main.yml
---
- name: 安装依赖
dnf:
name:
- curl
- wget
- firewalld
state: present
- name: 创建目录
file:
path: "{{ item }}"
state: directory
owner: root
group: root
mode: 0755
loop:
- "{{ seaweedfs_home }}"
- "{{ master_dir }}"
- "{{ volume_dir }}"
- /etc/seaweedfs
- name: 分发weed二进制
copy:
src: weed
dest: "{{ weed_bin }}"
mode: 0755
- name: 配置master service
template:
src: master.service.j2
dest: /etc/systemd/system/seaweedfs-master.service
when:
- inventory_hostname in groups.master
- name: 配置volume service
template:
src: volume.service.j2
dest: /etc/systemd/system/seaweedfs-volume.service
when:
- inventory_hostname in groups.volume
- name: 配置filer service
template:
src: filer.service.j2
dest: /etc/systemd/system/seaweedfs-filer.service
when:
- inventory_hostname in groups.filer
- name: 创建S3配置
template:
src: s3.json.j2
dest: /etc/seaweedfs/s3.json
when:
- inventory_hostname in groups.s3
- name: 配置S3 service
template:
src: s3.service.j2
dest: /etc/systemd/system/seaweedfs-s3.service
when:
- inventory_hostname in groups.s3
- name: 开放端口
firewalld:
port: "{{ item }}/tcp"
permanent: yes
state: enabled
loop:
- 9333
- 19333
- 8080
- 8888
- 8333
- name: reload systemd
systemd:
daemon_reload: yes六、Master Service
templates/master.service.j2
[Unit]
Description=SeaweedFS Master
After=network.target
[Service]
ExecStart={{weed_bin}} master \
-mdir={{master_dir}} \
-ip={{inventory_hostname}} \
-port=9333 \
-port.grpc=19333 \
-peers={{master_peers}}
Restart=always
LimitNOFILE=1048576
[Install]
WantedBy=multi-user.target七、Volume Service
templates/volume.service.j2
[Unit]
Description=SeaweedFS Volume
[Service]
ExecStart={{weed_bin}} volume \
-dir={{volume_dir}} \
-mserver={{master_peers}} \
-ip={{inventory_hostname}} \
-port=8080 \
-max=1000
Restart=always
LimitNOFILE=1048576
[Install]
WantedBy=multi-user.target八、Filer Service
templates/filer.service.j2
[Unit]
Description=SeaweedFS Filer
[Service]
ExecStart={{weed_bin}} filer \
-master={{master_peers}} \
-port=8888
Restart=always
[Install]
WantedBy=multi-user.target九、S3配置
创建：
templates/s3.json.j2
{
"identities":[
{
"name":"admin",
"credentials":[
{
"accessKey":"{{s3_access_key}}",
"secretKey":"{{s3_secret_key}}"
}
],
"actions":[
"Admin",
"Read",
"Write",
"List"
]
}
]
}十、S3 Service
templates/s3.service.j2
[Unit]
Description=SeaweedFS S3
[Service]
ExecStart={{weed_bin}} s3 \
-filer=fsnode01:8888 \
-port=8333 \
-config=/etc/seaweedfs/s3.json
Restart=always
[Install]
WantedBy=multi-user.target十一、启动服务任务补充
继续添加：
tasks/main.yml
最后：
- name: 启动master
systemd:
name: seaweedfs-master
enabled: yes
state: started
when:
- inventory_hostname in groups.master
- name: 启动volume
systemd:
name: seaweedfs-volume
enabled: yes
state: started
when:
- inventory_hostname in groups.volume
- name: 启动filer
systemd:
name: seaweedfs-filer
enabled: yes
state: started
when:
- inventory_hostname in groups.filer
- name: 启动s3
systemd:
name: seaweedfs-s3
enabled: yes
state: started
when:
- inventory_hostname in groups.s3十二、执行安装
复制 weed：
cp /usr/local/bin/weed \
roles/seaweedfs/files/测试：
ansible all -i inventory/hosts -m ping执行：
ansible-playbook \
-i inventory/hosts \
site.yml十三、验证集群
Master
curl http://fsnode01:9333/cluster/status正常：
{
"Leader":"fsnode01:9333",
"Peers":[
"fsnode01:9333",
"fsnode02:9333",
"fsnode03:9333"
]
}Volume
curl http://fsnode01:9333/dir/statusFiler
curl http://fsnode01:8888/S3
aws s3 ls \
--endpoint-url http://fsnode01:8333 \
--access-key admin \
--secret-key SeaweedFS@123456OpenCloudOS 9.6 额外建议
生产环境建议再加入：
- 磁盘自动格式化 XFS
- /etc/hosts自动配置
- SELinux策略处理
- sysctl优化
- Prometheus exporter
- systemd日志限制
你这个场景（/mnt/allin_data/allin/seaweedfs + 三节点）建议下一步直接扩展成：
Ansible一键部署版：磁盘初始化 + XFS挂载 + SeaweedFS Master/Volume/Filer/S3 + Prometheus监控
十四、故障定位
1 ansible-playbook安装报错：开放端口
root@k8s-node1 seaweedfs-install# ansible-playbook \ > -i inventory/hosts \ > site.yml ERROR! couldn't resolve module/action 'firewalld'. This often indicates a misspelling, missing collection, or incorrect module path. The error appears to be in '/root/seaweedfs-install/roles/seaweedfs/tasks/main.yml': line 86, column 3, but may be elsewhere in the file depending on the exact syntax problem. The offending line appears to be: - name: 开放端口 ^ here
错误：
ERROR! couldn't resolve module/action 'firewalld'
The offending line appears to be:
- name: 开放端口原因是：
Ansible 当前没有安装 firewalld 模块所在的 collection。
在新版 Ansible（尤其 ansible-core 2.15/2.16）里面：
以前：
firewalld:属于内置模块。
现在已经迁移到：
ansible.posix.firewalld需要安装 ansible.posix collection。
方案一（推荐）：安装 ansible.posix collection
执行：
ansible-galaxy collection install ansible.posix检查：
ansible-galaxy collection list应该看到：
Collection Version
-------------------------
ansible.posix 1.x.x然后修改：
roles/seaweedfs/tasks/main.yml
原：
- name: 开放端口
firewalld:
port: "{{ item }}/tcp"
permanent: yes
state: enabled改：
- name: 开放SeaweedFS端口
ansible.posix.firewalld:
port: "{{ item }}/tcp"
permanent: yes
state: enabled
immediate: yes
loop:
- 9333
- 19333
- 8080
- 8888
- 8333重新执行：
ansible-playbook \
-i inventory/hosts \
site.yml方案二：不用 Ansible firewalld 模块（生产更简单）
你的 SeaweedFS 集群固定端口：
9333 master HTTP
19333 master grpc
8080 volume
8888 filer
8333 S3可以直接调用 firewalld 命令：
替换：
- name: 开放端口
firewalld:为：
- name: 开放SeaweedFS端口
command: >
firewall-cmd
--add-port={{ item }}/tcp
--permanent
loop:
- 9333
- 19333
- 8080
- 8888
- 8333
- name: reload firewall
command:
firewall-cmd --reload优点：
- 不依赖 collection
- 龙蜥/OpenCloudOS兼容性更好
- 适合离线环境
另外建议检查你的 Ansible 版本
执行：
ansible --version如果是：
ansible [core 2.15.x]建议你的项目加：
requirements.yml
内容：
collections:
- name: ansible.posix
version: ">=1.5.0"以后安装：
ansible-galaxy collection install -r requirements.ymlSeaweedFS Role 后面还会遇到一个类似问题
你现在是：
systemd:这个仍然可以用。
但是如果后面增加：
selinux
firewalld
mount
lvg
filesystem在新 Ansible 都建议写全：
例如：
ansible.posix.selinux
ansible.posix.mount
community.general.lvg
community.general.filesystem建议你当前直接执行：
ansible-galaxy collection install ansible.posix然后把 firewalld: 改成：
ansible.posix.firewalld:即可继续安装 SeaweedFS
2 ansible-galaxy: 未找到命令
ansible-galaxy collection install ansible.posix -bash: ansible-galaxy: 未找到命令
报错：
ansible-galaxy collection install ansible.posix
-bash: ansible-galaxy: 未找到命令说明你的环境里 只有 ansible-playbook 命令，或者 Ansible 安装不完整 ，没有安装 ansible-galaxy。
先确认：
which ansible
which ansible-playbook
which ansible-galaxy以及：
ansible --version
ansible-playbook --version情况1：你使用的是 pip 安装的 ansible-core
如果之前执行的是：
pip3 install ansible-core那么 ansible-galaxy 应该跟随安装。
检查：
pip3 list | grep ansible正常应该类似：
ansible-core 2.15.x如果有，但是命令不存在，可能 PATH 不对。
查位置：
find / -name ansible-galaxy 2>/dev/null例如找到：
/usr/local/python3.9/bin/ansible-galaxy加入 PATH：
export PATH=$PATH:/usr/local/python3.9/bin永久：
echo 'export PATH=$PATH:/usr/local/python3.9/bin' >> ~/.bashrc
source ~/.bashrc情况2：你装的是 ansible-core 最小包
如果：
pip3 list | grep ansible只有：
ansible-core理论也应该有：
ansible-galaxy如果没有，可以重新安装：
pip3 install --force-reinstall ansible-core==2.15.13验证：
ansible-galaxy --version情况3：你通过 yum/dnf 装的不完整
检查：
rpm -qa | grep ansible如果只有：
ansible-core可以安装完整 ansible：
pip3 install ansible或者：
dnf install python3-ansible（龙蜥仓库有时包名不同）
你当前最推荐处理方式
因为你是 龙蜥8.6 + Python3.6 + SeaweedFS Ansible部署环境，建议不要折腾系统包。
直接确认 Python：
which python3
python3 --version然后安装到独立 Python。
例如：
/usr/local/python3.9/bin/pip3 install ansible-core==2.15.13然后：
/usr/local/python3.9/bin/ansible-galaxy collection install ansible.posix临时绕过（最快继续部署 SeaweedFS）
你也可以不用 firewalld 模块。
把：
firewalld:改成：
ansible.builtin.command:例如：
- name: 开放SeaweedFS端口
command: >
firewall-cmd
--add-port={{ item }}/tcp
--permanent
loop:
- 9333
- 19333
- 8080
- 8888
- 8333
- name: reload firewall
command: firewall-cmd --reload这样不用任何 collection，可以马上继续跑 playbook。
ansible-playbook执行报错
TASK seaweedfs : 开放SeaweedFS端口 ************************************************************************************************************************************************ failed: k8s-node2 (item=9333) => {"ansible_loop_var": "item", "changed": false, "item": 9333, "msg": "firewall is not currently running, unable to perform immediate actions without a running firewall daemon"} failed: k8s-node3 (item=9333) => {"ansible_loop_var": "item", "changed": false, "item": 9333, "msg": "firewall is not currently running, unable to perform immediate actions without a running firewall daemon"} failed: k8s-node1 (item=9333) => {"ansible_loop_var": "item", "changed": false, "item": 9333, "msg": "firewall is not currently running, unable to perform immediate actions without a running firewall daemon"} failed: k8s-node2 (item=19333) => {"ansible_loop_var": "item", "changed": false, "item": 19333, "msg": "firewall is not currently running, unable to perform immediate actions without a running firewall daemon"} failed: k8s-node3 (item=19333) => {"ansible_loop_var": "item", "changed": false, "item": 19333, "msg": "firewall is not currently running, unable to perform immediate actions without a running firewall daemon"} failed: k8s-node1 (item=19333) => {"ansible_loop_var": "item", "changed": false, "item": 19333, "msg": "firewall is not currently running, unable to perform immediate actions without a running firewall daemon"} failed: k8s-node2 (item=8080) => {"ansible_loop_var": "item", "changed": false, "item": 8080, "msg": "firewall is not currently running, unable to perform immediate actions without a running firewall daemon"} failed: k8s-node3 (item=8080) => {"ansible_loop_var": "item", "changed": false, "item": 8080, "msg": "firewall is not currently running, unable to perform immediate actions without a running firewall daemon"} failed: k8s-node2 (item=8888) => {"ansible_loop_var": "item", "changed": false, "item": 8888, "msg": "firewall is not currently running, unable to perform immediate actions without a running firewall daemon"} failed: k8s-node1 (item=8080) => {"ansible_loop_var": "item", "changed": false, "item": 8080, "msg": "firewall is not currently running, unable to perform immediate actions without a running firewall daemon"} failed: k8s-node3 (item=8888) => {"ansible_loop_var": "item", "changed": false, "item": 8888, "msg": "firewall is not currently running, unable to perform immediate actions without a running firewall daemon"} failed: k8s-node2 (item=8333) => {"ansible_loop_var": "item", "changed": false, "item": 8333, "msg": "firewall is not currently running, unable to perform immediate actions without a running firewall daemon"} failed: k8s-node3 (item=8333) => {"ansible_loop_var": "item", "changed": false, "item": 8333, "msg": "firewall is not currently running, unable to perform immediate actions without a running firewall daemon"} failed: k8s-node1 (item=8888) => {"ansible_loop_var": "item", "changed": false, "item": 8888, "msg": "firewall is not currently running, unable to perform immediate actions without a running firewall daemon"} failed: k8s-node1 (item=8333) => {"ansible_loop_var": "item", "changed": false, "item": 8333, "msg": "firewall is not currently running, unable to perform immediate actions without a running firewall daemon"} PLAY RECAP ************************************************************************************************************************************************************************** k8s-node1 : ok=9 changed=7 unreachable=0 failed=1 skipped=0 rescued=0 ignored=0 k8s-node2 : ok=6 changed=5 unreachable=0 failed=1 skipped=3 rescued=0 ignored=0 k8s-node3 : ok=6 changed=5 unreachable=0 failed=1 skipped=3 rescued=0 ignored=0
你的报错原因很明确：
firewall is not currently running, unable to perform immediate actions without a running firewall daemon说明：
目标节点 k8s-node1/2/3 上 firewalld 服务没有启动。
你的 Ansible 使用的是：
ansible.posix.firewalld:
port: "{{ item }}/tcp"
permanent: yes
state: enabled
immediate: yes其中：
immediate: yes要求：
systemctl status firewalld必须是运行状态。
方案一（推荐）：启动 firewalld
在 playbook 前增加 firewalld 安装和启动。
修改：
roles/seaweedfs/tasks/main.yml
增加：
- name: 安装firewalld
dnf:
name: firewalld
state: present
- name: 启动firewalld
systemd:
name: firewalld
state: started
enabled: yes放在开放端口任务之前。
例如：
- name: 安装firewalld
dnf:
name: firewalld
state: present
- name: 启动firewalld
systemd:
name: firewalld
enabled: yes
state: started
- name: 开放SeaweedFS端口
ansible.posix.firewalld:
port: "{{ item }}/tcp"
permanent: yes
state: enabled
immediate: yes
loop:
- 9333
- 19333
- 8080
- 8888
- 8333然后重新执行：
ansible-playbook \
-i inventory/hosts \
site.yml方案二（生产环境更推荐）：关闭 firewalld 管理
如果你的集群节点：
- 在内网
- 使用安全组
- 使用硬件防火墙
- Kubernetes 已经管理网络
很多生产环境直接关闭 firewalld。
那么删除这一段：
- name: 开放SeaweedFS端口
ansible.posix.firewalld:或者改成跳过：
- name: 开放SeaweedFS端口
ansible.posix.firewalld:
port: "{{ item }}/tcp"
permanent: yes
state: enabled
loop:
- 9333
- 19333
- 8080
- 8888
- 8333
when: false方案三：不用 firewalld 模块（适合你的龙蜥环境）
你的节点看起来是：
k8s-node1
k8s-node2
k8s-node3可能已经关闭防火墙。
直接删除 firewalld task。
SeaweedFS只需要保证端口互通：
| 服务 | 端口 |
|---|---|
| Master HTTP | 9333 |
| Master gRPC Raft | 19333 |
| Volume | 8080 |
| Filer | 8888 |
| S3 | 8333 |
测试：
nc -vz k8s-node2 19333
nc -vz k8s-node3 19333建议的 SeaweedFS Ansible Role 调整
因为你的目标是存储集群，不建议强制管理防火墙。
修改变量：
defaults/main.yml
增加：
manage_firewalld: false然后：
- name: 开放SeaweedFS端口
ansible.posix.firewalld:
port: "{{ item }}/tcp"
permanent: yes
state: enabled
immediate: yes
loop:
- 9333
- 19333
- 8080
- 8888
- 8333
when:
- manage_firewalld以后：
生产环境：
manage_firewalld: false测试环境：
manage_firewalld: true最快恢复部署
直接执行：
systemctl start firewalld
systemctl enable firewalld三台节点：
ansible seaweedfs -i inventory/hosts \
-m systemd \
-a "name=firewalld state=started enabled=yes"然后：
ansible-playbook \
-i inventory/hosts \
site.yml不过我注意到你的节点名称已经变成：
k8s-node1
k8s-node2
k8s-node3不是之前：
fsnode01
fsnode02
fsnode03下一步部署 SeaweedFS Master 时，需要同步修改：
master_peers:
k8s-node1:9333,
k8s-node2:9333,
k8s-node3:9333否则后面 Raft 会出现你之前遇到的：
raft.Server: Not current leader3 ansible-playbook执行报错：主机名解析失败
/usr/local/bin/weed master \ > -mdir=/mnt/allin_data/allin/seaweedfs/master/m9333 \ > -ip=k8s-node3 \ > -port=9333 \ > -port.grpc=19333 \ > -peers=k8s-node1:9333,k8s-node2:9333,k8s-node3:9333 I0904 16:01:48.528425 file_util.go:29 Folder /mnt/allin_data/allin/seaweedfs/master/m9333 Permission: -rwxr-xr-x I0904 16:01:48.528910 master.go:394 current: k8s-node3:9333 peers:k8s-node1:9333,k8s-node2:9333,k8s-node3:9333 I0904 16:01:48.529062 master_server.go:156 Volume Size Limit is 30720 MB I0904 16:01:48.529123 collector.go:83 Reporting anonymous cluster statistics to https://telemetry.seaweedfs.com/api/collect every 24h0m0s, use -telemetry=false to opt out I0904 16:01:48.529343 master.go:203 Start Seaweed Master 30GB 4.45 79b87202136cebdaaa7db4d94eaa5915ad381276 at k8s-node3:9333 F0904 16:01:48.530369 master.go:206 Master startup error: listen tcp: lookup k8s-node3 on 10.80.0.2:53: no such host goroutine 1 running: github.com/seaweedfs/seaweedfs/weed/glog.stacks(0x0) /github/workspace/weed/glog/glog.go:832 +0x85 github.com/seaweedfs/seaweedfs/weed/glog.(*loggingT).output(0x8dabce0, 0x3, 0x39c03801b960, {0x566ff77?, 0x39c0382c9a28?}, 0x1?, 0x0) /github/workspace/weed/glog/glog.go:783 +0x43d github.com/seaweedfs/seaweedfs/weed/glog.(*loggingT).printf(0x8dabce0, 0x3, {0x413f3de, 0x18}, {0x39c0382c9a28, 0x1, 0x1}) /github/workspace/weed/glog/glog.go:680 +0x12a github.com/seaweedfs/seaweedfs/weed/glog.Fatalf(...) /github/workspace/weed/glog/glog.go:1242 github.com/seaweedfs/seaweedfs/weed/command.startMaster({0x39c038507ed8, 0x39c038507f10, 0x39c038464830, 0x39c038464850, 0x39c038464860, 0x39c038464880, 0x0, 0x39c03857c280, 0x39c03857c295, 0x39c03857c2e0, ...}, ...) /github/workspace/weed/command/master.go:206 +0x3d9 github.com/seaweedfs/seaweedfs/weed/command.runMaster(0x8d74790?, {0x39c03762c100?, 0x5?, 0x5?}) /github/workspace/weed/command/master.go:177 +0x667 main.main() /github/workspace/weed/weed.go:97 +0x469 root@k8s-node3 \~#
报错已经定位，不是 SeaweedFS 参数问题，而是 主机名解析失败：
关键错误：
Master startup error: listen tcp: lookup k8s-node3 on 10.80.0.2:53: no such host意思：
SeaweedFS 执行：
-ip=k8s-node3启动监听时，需要解析：
k8s-node3但是 DNS：
10.80.0.2无法解析这个 hostname。
解决方案一（推荐）：配置 /etc/hosts
三台节点都配置。
k8s-node1
vim /etc/hosts增加：
10.40.0.77 k8s-node1
10.40.0.78 k8s-node2
10.40.0.79 k8s-node3k8s-node2
同样：
10.40.0.77 k8s-node1
10.40.0.78 k8s-node2
10.40.0.79 k8s-node3k8s-node3
同样：
10.40.0.77 k8s-node1
10.40.0.78 k8s-node2
10.40.0.79 k8s-node3测试：
ping k8s-node1
ping k8s-node2
ping k8s-node3应该返回对应 IP。
解决方案二：直接使用 IP（更稳定）
SeaweedFS Master 集群生产环境建议使用 IP。
例如：
k8s-node3
改：
-ip=k8s-node3为：
-ip=10.40.0.79完整：
/usr/local/bin/weed master \
-mdir=/mnt/allin_data/allin/seaweedfs/master/m9333 \
-ip=10.40.0.79 \
-port=9333 \
-port.grpc=19333 \
-peers=10.40.0.77:9333,10.40.0.78:9333,10.40.0.79:9333另外两个节点：
k8s-node1
weed master \
-mdir=/mnt/allin_data/allin/seaweedfs/master/m9333 \
-ip=10.40.0.77 \
-port=9333 \
-port.grpc=19333 \
-peers=10.40.0.77:9333,10.40.0.78:9333,10.40.0.79:9333k8s-node2
weed master \
-mdir=/mnt/allin_data/allin/seaweedfs/master/m9333 \
-ip=10.40.0.78 \
-port=9333 \
-port.grpc=19333 \
-peers=10.40.0.77:9333,10.40.0.78:9333,10.40.0.79:9333另外检查你的 Ansible inventory
你现在应该不要写：
[master]
k8s-node1
k8s-node2
k8s-node3然后变量：
master_peers: k8s-node1:9333,...除非你保证 /etc/hosts。
建议改成 IP：
defaults/main.yml
master_peers: "10.40.0.77:9333,10.40.0.78:9333,10.40.0.79:9333"然后模板：
ExecStart={{weed_bin}} master \
-mdir={{master_dir}} \
-ip={{ansible_host}} \
-port=9333 \
-port.grpc=19333 \
-peers={{master_peers}}这样 Ansible 部署不会依赖 DNS。
当前隐藏问题
之前你删除过：
/mnt/allin_data/allin/seaweedfs/master/m9333现在重新初始化时：
启动顺序必须：
1. k8s-node1
先启动：
weed master ...看到：
is the leader保持运行。
2. k8s-node2
3. k8s-node3
然后验证：
curl http://10.40.0.77:9333/cluster/status正常应该看到 Leader。
建议你现在先执行：
cat /etc/hosts
hostname -I三台都检查一下。
你的问题目前 99% 是 k8s-node1/2/3 没有做 hosts 解析