Practical Guide to BandwagonHost and DMIT OS Reinstallation, Custom ISO Mounting, and Emergency Rescue Mode

During the cloud server operations lifecycle, firewall misconfiguration can block SSH, kernel upgrades can cause a Kernel Panic, and accidental edits to /etc/fstab causing boot hangs, and complete rebuilds necessitated by a cluttered, inconsistent system environment, are scenarios every operations engineer and website operator is likely to encounter.

For the two major infrastructure providers already integrated into VPSMap:BandwagonHost（ BandwagonHost ） and DMIT, the two differ significantly in virtualization consoles, credential delivery, out-of-band recovery, and ISO mounting support. This article examines both platforms' underlying toolsets and provides standardized reinstallation procedures, practical offline Chroot recovery commands, and Netboot deployment methods.

Platform Operations Capabilities at a Glance#

One. Fully Automated OS Reinstallation (Reinstall / Rebuild)#

OS reinstallation formats the primary virtual disk and writes a base image to it. Before proceeding, understand that this is irreversible and learn the two providers' different rules for initializing security credentials.

1. BandwagonHost KiwiVM Automated Reinstallation Process#

BandwagonHost centralizes all infrastructure-level operations in its proprietary KiwiVM Management Panel. Please note:The KiwiVM Password, BandwagonHost Client Area Login Password, and Virtual Machine's root Password Are All Independent。

1. 登录 KiwiVM 面板，在左侧菜单栏选择 Install new OS。
2. Select the target distribution. For production, prioritize a stable, maintained release such as Debian 12 or Ubuntu 24.04 LTS.
3. Select the checkbox confirming data erasure (I understand that all existing data will be lost）。
4. After confirming reinstallation, the VM automatically formats and writes the system, usually taking between 60 and 180 seconds.
5. Record the Credentials Displayed on Screen：安装完成后，页面会输出一次性提示信息，包含：
   • randomly generated strong root initial password;
   • The SSH port on which the instance actually listens (to prevent automated brute-force attacks, BandwagonHost does not open the standard port 22 by default; instead, it maps SSH to a random port above 20000).

Security Tips: securely save the output shown on this screen. If you close the page without recording it, you cannot view the password again; you must reinstall through KiwiVM or use the root password reset feature to issue a new one.

2. Rebuilding a DMIT Instance from a Template#

DMIT 的安全策略与现代化 cloud services 规范对齐，其 instance 系统在初始化阶段默认Disable Remote root Password Authentication, fully adopting public-key-based SSH Key login.

1. Log into the DMIT client console and open the relevant VM's management page.
2. Click Rebuild(shown in some interfaces as Reinstall）。
3. Select the Required Operating System Image Version.
4. Associate an SSH Key: In the credential options, select the SSH Public Key you previously added to the DMIT key repository. If it has not been imported, first go to the account's SSH Keys menu to enter your local public key (id_ed25519.pub or id_rsa.pub）。
5. After you confirm the rebuild, the system rewrites the disk in the background. There is no need to wait for a randomly generated password; once the rebuild is ready, connect directly using your local private key root@<IP> -p 22 and that is all.

Credential Change Rules: if you later use the DMIT instance panel's Access tab to change the root password or replace an SSH Key, the cloud platform injects the configuration through the underlying initialization agent (Cloud-Init / QEMU Guest Agent),Perform a Hard Reboot (Reboot / Power Cycle) as Instructed by the Panel, only then will the new key or password take effect.

Two. Out-of-Band Emergency Console: The First Troubleshooting Channel When Network Access Is Lost#

When you have misconfigured iptables、nftables、ufw blocks all incoming ports, or the network configuration file (/etc/network/interfaces or Netplan YAML) syntax errors prevent the instance from obtaining an IP, ordinary SSH tools become unusable. You must intervene through an Out-of-Band Console.

常规远程路径:  [本地终端] --- (公网 TCP: SSH 端口) ---> [虚拟网卡 eth0] ---> (sshd 进程) ❌ 网络阻断即失联
带外控制台:    [浏览器] --- (HTTPS 隧道) ---> [宿主机 Hypervisor VNC] ---> (虚拟显卡/串口 tty) ✔ 绕过网络层

1. BandwagonHost Interactive Console#

• 访问入口: In the KiwiVM navigation bar, click Interactive Console。
• Operating Characteristics: This feature connects directly to the VNC/serial interface exposed by the virtualization layer. Even without an assigned IP, with all networking down, or with a firewall denying everything, you can obtain a browser terminal as long as the underlying VM is Running.
• Typical Uses: enter the username root and password to log in, then promptly stop the malfunctioning firewall daemon (such as systemctl stop ufw), or correct the network adapter binding.

2. DMIT Web Console#

• 访问入口: On the instance overview page, click Console button.
• Operating Characteristics: Built with HTML5 Canvas, it provides complete character-terminal emulation. For users whose root password login is disabled by default and whose private key no longer works, if previously configured in Access menu and restarted the instance to apply it, you can log in directly to the Console as root using that password to recover the system.

Three. Hands-On Emergency Rescue Mode and Data Recovery#

When a kernel update crashes the system into a Kernel Panic, a damaged system disk is mounted as a Read-Only Filesystem, or critical boot modules are missing, even the ordinary out-of-band console can only display the errors. You must then boot the virtual machine intoMinimal In-Memory System (Rescue Live System), enabling offline disk repairs or off-site data export.

              ┌────────────────────────────────────────────────────────┐
              │           Rescue Mode (纯内存运行的 Live 环境)          │
              │         RAM Disk: Alpine Linux / Debian Live           │
              └───────────────────────────┬────────────────────────────┘
                                          │ 挂载原有受损磁盘
                                          ▼
              ┌────────────────────────────────────────────────────────┐
              │          物理宿主机直通磁盘设备 (/dev/vda)             │
              │  /dev/vda1 (原主分区: 存放 /etc, /var/www, 数据库文件) │
              └────────────────────────────────────────────────────────┘

1. Prepare the Environment to Enter Rescue Mode#

• DMIT Approach：在控制面板点击 Rescue Mode to start. The system forces a shutdown and boots from a predefined read-only recovery image into an independent miniature OS in RAM. The panel displays a temporary rescue IP, SSH port, and random access credentials. Connect to this temporary environment over SSH to begin troubleshooting.
• BandwagonHost Approach: when KiwiVM does not have a one-click Rescue button, the standard alternative is to open Mount ISO menu and mount the built-in SystemRescueCD or Debian Live image, set CD-ROM as the first boot device, then start the instance and open Interactive Console to enter the fully featured rescue terminal.

2. The Standard Four-Step Low-Level Troubleshooting and Repair Method#

Perform all the following operations in the terminal after entering the rescue environment.

步骤 1：排查并识别目标 Storage 块设备#

First identify the virtual disk device names and partition layout assigned by the host:

# 列出系统识别到的所有块设备及分区
lsblk -f

The primary disk is usually identified as /dev/vda or /dev/sda，业务系统根分区多为 /dev/vda1(if a UEFI partition table is used, it may include an EFI boot partition /dev/vda1 and the root partition /dev/vda2）。

Before mounting, run a filesystem consistency check on the damaged partition, assuming an ext4 filesystem:

# 强制检查并修复分区逻辑错误
e2fsck -f -y /dev/vda1

Step 2: Mount the Damaged Filesystem in Read/Write Mode#

Create a temporary mount target in the RAM-based rescue system:

# 创建挂载点
mkdir -p /mnt/sysroot

# 以读写模式挂载损坏的原系统分区
mount -o rw /dev/vda1 /mnt/sysroot

# 核对目录完整性
ls -la /mnt/sysroot/
# 确认可以看到原系统的 bin, etc, home, usr, var 等核心目录

Step 3: Access Core Subsystems and Reconstruct the Chroot Environment#

To seamlessly run binaries from the original system, such as changing passwords, regenerating initramfs, or installing or removing kernels, you must map the kernel's underlying virtual filesystems into the mount point:

# 依次绑定宿主内核的虚拟文件系统
mount --bind /dev /mnt/sysroot/dev
mount --bind /dev/pts /mnt/sysroot/dev/pts
mount --bind /proc /mnt/sysroot/proc
mount --bind /sys /mnt/sysroot/sys
mount --bind /run /mnt/sysroot/run

# 将 DNS 解析配置文件拷贝进入环境，确保修复期间网络畅通
cp -L /etc/resolv.conf /mnt/sysroot/etc/resolv.conf

# 切换上下文根目录至原系统
chroot /mnt/sysroot /bin/bash

After running the commands above, the terminal context has entered the original system's actual environment. You can now perform high-risk repairs:

Scenario A: Reset a Completely Lost root Password#

# 直接重写 root 凭据
passwd root

场景 B：修复因错误配置导致的 SSH 无法启动#

# 检查 SSH 配置文件是否存在语法阻断
sshd -t

# 修复错误的 PAM 限制或允许 root 登录（按需调整）
sed -i 's/^PermitRootLogin.*/PermitRootLogin yes/' /etc/ssh/sshd_config

# 若此前卸载了核心软件包，可触发自动修复
apt update && apt --fix-broken install -y

Scenario C: Repair the Following Cause of Boot Hangs: /etc/fstab Mount Entries#

If you recently attached an external disk, formatted a partition, or enabled an unresponsive NFS/SMB share, boot-time timeouts may leave the system stuck on the Maintenance screen:

# 检查原系统的挂载映射表
nano /etc/fstab
# 将失效的分区挂载行注释（行首添加 #），或纠正错误的 UUID

Exit and Cleanup#

After repairs are complete, clean up the environment in the correct reverse order to avoid further file corruption from caches that have not been flushed to disk:

# 退出 chroot 虚环境
exit

# 卸载绑定的虚拟文件系统及物理分区
umount -R /mnt/sysroot

# 将写缓存强制刷新至物理磁盘
sync

Then return to the console, exit Rescue Mode or unmount the ISO, and reboot the instance into the original operating system.

3. Off-Site Data Export After an Irrecoverable System Crash#

如果 Operating system 内核组件严重损坏、甚至关键动态链接库已被破坏，强行修复耗时过高，应果断在救援模式下提取核心静态资产与 Databases 目录。

After entering the rescue system (having completed mount /dev/vda1 /mnt/sysroot), use a network pipeline to archive and transfer the data directly to a secure remote server or a local administration relay machine:

Option A: Incremental Export with rsync#

# 将原系统的网站目录与数据库持久化目录推送至异地灾备服务器
rsync -avz -e 'ssh -p 22' /mnt/sysroot/var/www/ [email protected]:/data/emergency/www/
rsync -avz -e 'ssh -p 22' /mnt/sysroot/var/lib/mysql/ [email protected]:/data/emergency/mysql/

Method B: Pipelined Compression and Concurrent Transfer (For Low-Specification or Low-Memory Servers)#

Start a listener on your local machine or a relay server to receive the streamed archive without consuming space on the rescue RAM disk:

# 在救援机执行：将目标目录打包并直通输出到 SSH 隧道，在远端直接落盘
tar -czvf - -C /mnt/sysroot/var/www . | ssh -p 22 [email protected] "cat > /data/emergency/www_backup.tar.gz"

Once the data has been safely archived offline, you can confidently run Reinstall / Rebuild in the provider's panel and then restore your application assets.

Four. Custom ISO Mounting and Automated Network Reinstallation (Netboot.xyz)#

Providers 官方提供的 Operating system image 通常为 Standard 定制模版。部分专业场景需要安装极致轻量的 Alpine Linux、BSD 系统，或者全新纯净的自定义内核系统。

1. Native ISO Mounting Guidelines for BandwagonHost KiwiVM#

BandwagonHost has deeply integrated virtual optical-drive mounting support into KiwiVM:

1. Open the following in KiwiVM's left navigation menu: Mount ISO。
2. The page lists commonly used third-party and open-source installation media cached by the platform, including mainstream versions of Alpine, Debian Netinst, FreeBSD, and Ubuntu Server.
3. Select the appropriate image file and click Mount. The virtualization layer maps the ISO as a virtual CD-ROM and places it first in the BIOS boot order.
4. force the following operation once in the panel: Reboot, and quickly open Interactive Console。
5. The console now displays the distribution's native interactive installation wizard. Customize the partition layout, Swap size, and network parameters as needed.
6. A Crucial Final Step： Operating system 安装完毕并提示重启时，First Return to KiwiVM and Run Unmount ISO. If the image is not unmounted, the instance will boot from CD-ROM again after restarting and repeatedly return to the installer.

2. Chainload Netboot.xyz by Injecting GRUB#

当 Providers Not provided 直观的 ISO 自定义上传接口（如 DMIT 默认面板未放开任意本地 ISO 直传），或者 BandwagonHost 官方列表中不包含所需冷门发行版时，最优雅的解决方案是Inject a Generic iPXE Boot Kernel Through the Current System's GRUB Bootloader。

This approach requires only a basic Linux environment and public DNS resolution on the current instance:

[开机 BIOS/UEFI]
       │
       ▼
[磁盘 MBR/GPT] ---> [本地 GRUB2 引导菜单]
                            │
                            ├── (默认) 启动原有系统内核
                            └── (自定义) 链式加载 /boot/netboot.xyz.lkrn
                                              │
                                              ▼
                                   [进入内存 iPXE 环境]
                                              │ 从官方 CDN 拉取
                                              ▼
                                   [加载任意 OS 安装引导器]

Hands-On Deployment#

步骤 1：下载 Netboot.xyz 内核介质#

In the terminal of the currently running Linux instance, execute with root privileges:

# 切换至引导目录并拉取 iPXE 预编译内核
cd /boot
wget -O netboot.xyz.lkrn https://boot.netboot.xyz/ipxe/netboot.xyz.lkrn

Step 2: Add a Custom Boot Entry to GRUB#

Edit /etc/grub.d/40_custom file, and add a custom menu entry:

cat << 'EOF' >> /etc/grub.d/40_custom

menuentry "Network Boot - netboot.xyz" {
    set root=(hd0,1)
    linux16 /boot/netboot.xyz.lkrn
}
EOF

Note: if the current system's /boot is a separate partition, or the disk uses EFI partitions (with the main root partition on /dev/vda2），需相应将 (hd0,1) adjust it to the actual boot partition number.

步骤 3：刷新引导配置#

# Debian / Ubuntu 环境
update-grub

# AlmaLinux / Rocky Linux / CentOS 环境
grub2-mkconfig -o /boot/grub2/grub.cfg

Step 4: Intercept Boot Through the Console and Configure Networking#

1. Open the provider's out-of-band console in your browser beforehand, using KiwiVM Interactive Console or DMIT Web Console.
2. In the original system's terminal, enter reboot Reboot the server.
3. Watch the console output closely. During the boot countdown, in the brief few seconds when the GRUB menu is displayed, repeatedly press ↑ / ↓ use the arrow keys to interrupt the default boot process, then move the cursor to select Network Boot - netboot.xyz and press Enter.
4. The server loads the iPXE core locally, automatically obtains a public IP through DHCP, and connects to the network. Netboot.xyz's text-based menu then appears, offering Linux Network Installs, Live CDs, BSD, and hardware diagnostic tools.
5. Select the target operating system. The boot kernel is automatically downloaded from a remote CDN into memory, and the full installation begins.

Five. Related Lifecycle Considerations and Operational Boundaries#

During OS reinstallation, migration, and data recovery, infrastructure-level actions may interact with the provider's billing model and resource lifecycle. Follow these operational boundaries and rules:

1. BandwagonHost Operations and Account Policy Limits#

• Credential Boundaries：请始终区分三套凭据——客户中心用于工单与财务结算，KiwiVM 拥有独立的控制台授权秘钥， instance 内部拥有 root 权限。KiwiVM 密码切勿与系统密码混淆。
• Billing and Renewal Logic: under BandwagonHost's default configurationDoes Not Automatically Link a Credit Card or PayPal Account for Silent Charges. If renewal is enabled, the system automatically generates an invoice 7 days before expiration. If sufficient prepaid credit is available, you can use it to pay. During intensive maintenance or experimentation, keep track of expiration dates to avoid suspension for overdue billing.
• Refund and Data Deletion Boundaries：新购 instance 在 30 days内且完全满足服务条款前提下方可申请退款，且要求 Monthly transfer 消耗不得超过配额的 10%。需要明确的是，一旦退款申请获得批准，对应服务 instance 将被直接销毁，虚拟磁盘及其关联备份数据将立即被永久 Clear 。
• IP Status and Migration Restrictions: Before KiwiVM starts “Migrate to another DC,” it automatically checks the health of the currently assigned IP. If the IP is blacklisted, migration is restricted and cannot be used to bypass network restrictions.

2. Key DMIT Operations and Lifecycle Guidelines#

• Renewal Invoice Schedule and Grace Period: DMIT monthly renewal invoices are usually generated 7 days before the service expires. If payment is not completed on time, the server enters Suspended status.Instances Suspended for Nonpayment Usually Receive Only About a 3-Day Retention Grace Period; once the grace period expires, the system irreversibly reclaims the resources. The original virtual disk data is completely erased and cannot be recovered on request.
• Pay Attention to the Traffic Billing Model：DMIT 针对不同机房和 Networking 产品线（如 Premium 优质优化 Networks 、Eyeball 普通优化 Networks 及 Storage 大盘鸡等）配置了不同的流量超额处理规则，包含超额停机、超额降速以及 Usage-based 增购等模式。在实施大型异地数据救援备份（使用 rsync / tar exporting large volumes of files at full bandwidth), check the plan's current monthly traffic usage and overage policy on the instance page to avoid a forced network disconnection caused by rapidly exhausting the allowance.
• 退款时效与用量约束: DMIT offers refunds for eligible newly purchased instances, but with stricter requirements: within 3 days of purchase, a full refund may be requested under the terms if total usage across the entire period does not exceed 30GB. Within 30 days, a separate prorated refund mechanism calculates the remaining value. All details are governed by the latest official terms of service. If recovery fails and you need to cancel to limit losses, immediately check whether your data usage has reached the threshold.