Submit an Effective Support Ticket: Make the Issue Reproducible

For Unmanaged VPS hosting, the technical support team's core responsibility is to keep the underlying host hardware, virtualization layer (KVM), host network access, and client-area billing system functioning. Software compilation, firewall rules, kernel tuning, and application configuration inside the operating system are generally the user's responsibility.

当服务发生异常时，提交一份低信噪比的工单（例如仅描述“服务器连不上”或“IP 无法访问”）往往会导致客服要求补充基础信息，延长故障恢复时间。撰写有效工单的关键，在于Clarify Support Responsibilities Before Submitting, Use the Provider's Out-of-Band Tools to Collect Key Evidence, and Present the Problem in a Structured, Independently Reproducible Format。

Defining Responsibilities: Which Issues Are Covered by Support Tickets#

Technical support at major international VPS providers is usually organized into different response tiers. Clarify the nature of the problem before submitting a ticket so it can be routed accurately to billing, networking, or underlying hardware support.

Out-of-Band Checks and Key Evidence Collection Before Submission#

The most direct way to demonstrate that a fault is in the infrastructure rather than inside the system is to provide troubleshooting records from the provider's own console. Major providers differ in console architecture and access mechanisms, so collect evidence accordingly.

1. Use the Out-of-Band Console to Confirm the Instance Is Alive#

When an SSH connection cannot be established, do not immediately assume that the network is blocked or the host is down. Even if external connectivity is interrupted, the out-of-band console can still connect directly to the instance terminal through an internal-network tunnel.

• BandwagonHost（ BandwagonHost ）：
  • 登录客户中心后进入 KiwiVM Control Panel。
  • Check the following on the left: Interactive Console or Root shell-interactive。
  • Distinguishing Credentials: note the KiwiVM management panel password, the BandwagonHost client area login password, and the Linux system's root 密码是三套独立的认证凭据。 Passed 控制台登录系统排查时，使用的是 Linux 自身的 root 凭据。
  • If Interactive Console shows a login prompt, the virtualization layer is working and the problem lies in networking, the firewall, or the SSH daemon. If the console is blank or stuck on a Kernel Panic and KiwiVM cannot send a reboot signal, the underlying virtualization or boot image has failed. Capture the console error logs in that case.
• DMIT：
  • Log in to the client area, open the instance details, and use Console(VNC entry point) for emergency access.
  • 认证特性: For security, DMIT's official templates disable remote root password login by default and require SSH key pairs. If access is lost because the private key is missing or permissions are incorrect, use the panel's Access tab to reset the password or inject a new public key.
  • Requirements for Taking Effect: After changing the password or updating SSH keys under Access in the DMIT panel,Restart the Instance as Instructed by the Panel, allowing the underlying script to write the new credentials into the VM image. Only if Console login still fails with the new credentials after reboot should you record the panel operation time and request technical assistance.

2. Networking 中断与路由劣化的取证规范#

If the instance runs normally internally but is completely inaccessible externally, investigate the network path rather than giving technical support an undiagnosable conclusion such as “Ping doesn't work.”

• Two-Way Connection Diagnostics: One-way ping cannot identify where the routing failure occurs. Use a tool that supports route tracing and packet-loss statistics, such as mtr or traceroute), trace at least 50 packets from the local client to the VPS IP, while running a reverse MTR from the VPS console to the local IP or a public DNS server.
• Troubleshooting IP Availability and Data Center Migration：
  • If you try to migrate a BandwagonHost instance through KiwiVM's “Migrate to another DC” because its IP is blocked, migration is generally restricted if the system detects the IP on a blocklist. For network-unavailability tickets, include results from third-party multi-location probes and state whether the problem is regional routing blockage or globally blocked ports.
  • 在 DMIT 环境中，不同路由层级（如 Premium 优质 CN2 GIA/CMIN2 优化路由与普通 Eyeball/Lite Networks ）的承载能力和保障条款不同。若反馈路由异常，必须提供包含 ASN 节点跃点信息的 MTR 报告，证明流量未走合同注明的优化 Gateway 。

3. Preliminary Checks Before Billing and Refund Disputes#

Billing support tickets must align precisely with the provider's established Terms of Service (TOS) to avoid immediate rejection because the request falls outside the service agreement.

• 续费时效问题：
  • BandwagonHost renewal invoices are generally generated 7 days before expiration. The system does not silently initiate automatic charges against credit cards or third-party payment accounts. If you have sufficient prepaid account credit, actively confirm that applying the balance has been selected to avoid overdue suspension.
  • DMIT monthly renewal invoices are also generated 7 days before expiration. If unpaid by the due date, instances enter suspension and generally receive only about a 3-day grace period, after which automated destruction physically erases the disk data. If renewal payment succeeds but the instance does not power on automatically, immediately submit a ticket with the transaction ID.
• Refund Compliance Check：
  • When requesting a refund from BandwagonHost, verify that the purchase was made within 30 days and that cumulative traffic usage during the billing cycle is strictly below 10% of the quota. Also confirm that the account has not abused the refund policy. When the refund is processed, the data is destroyed at the infrastructure level as well.
  • When requesting a DMIT refund, distinguish between full and prorated refunds. New instances generally qualify for a full refund only within 3 days and with total traffic usage no greater than 30GB. After that window but within 30 days, fees and used service value are deducted under the Terms of Service to calculate the remaining refund. Include the traffic usage panel reading directly in the ticket.

Writing a Standardized Support Ticket: Structure and Language#

Support teams route large volumes of tickets daily. A consistently structured ticket can move directly into a second-tier engineer's troubleshooting workflow, avoiding delays from repeated requests for basic information.

1. 标题命名规范#

The subject must include these key elements:Affected Resource + Key Error / Main Symptom。

• Poor Example：
  • Server is down please help(No Asset Identifier or Specific Symptoms)
  • Why my VPS not working(Lacks Technical Details)
  • Billing error(It Is Unclear Whether This Concerns an Extra Charge, a Missing Payment, or a Refund Request)
• Properly Formatted Example：
  • [Hardware/Kernel] VM 204928 hung on boot with "kernel panic - not syncing: VFS"
  • [Network] Dropping 100% packets at upstream gateway 198.51.100.1 (IP: 198.51.100.24)
  • [Billing] Invoice #84920 paid via PayPal, transaction completed but status unpaid

2. Message Body Structure#

Avoid lengthy narratives in the ticket body. Use structured sections designed for technical support tickets:

Service / Asset ID:
[填写服务 ID、主 IP 地址、机房编号，如 HKG / LAX DC6 / TYO]

Incident Timestamp:
[发生时间，明确标注时区，如 2026-09-26 14:20:00 UTC 或 CST / GMT+8]

Symptoms & Error Messages:
[详细描述异常现象，附上终端中返回的精确报错文本或控制台错误码]

Diagnostic Steps Taken:
[说明已经完成的自主排障操作，避免技术支持给出重复建议]
- Accessed via Out-of-band Console (KiwiVM / Web Console): [结果]
- Verified internal network interfaces and routing table: [结果]
- Verified local firewall (iptables / nftables) status: [结果]
- Ran bidirectional MTR tracing: [附带结果概要]

Reproducibility / Impact:
[故障是持续发生还是间歇性发生？当前是否已彻底不可用？]

Specific Request:
[明确需要技术人员执行的操作，例如：检查宿主机虚拟化宿主状态 / 重置底层网卡绑定 / 校验财务到账记录]

3. 英文工单的表达要点#

Although many overseas providers support multilingual initial triage, engineers handling core infrastructure and data center networks communicate in English. Clear English can significantly speed up routing and resolution:

• Preserve the Original Error Message: Whether it is kernel output, a web service status code, or a control panel error, never use a translation tool to turn it into natural-language wording. Preserve the original text exactly (such as No route to host、Connection refused、I/O error）。
• 陈述事实，避免猜测: describe the system state objectively. Write MTR shows 100% packet loss starting from the second hop inside the DC, which is better than I think your network is broken。
• Redact Sensitive Information: Never include system login private keys, plaintext administrator passwords, full bank card numbers, or other highly confidential information in support tickets. If technicians need temporary access for troubleshooting, require authorization through the client area's protected encrypted channel, or rotate passwords and keys through the console immediately afterward.

Guidelines for Preparing and Attaching Evidence#

Attachments Are the Most Important Supporting Evidence in a Ticket, but Unfiltered Material Can Disrupt Troubleshooting.

1. Handling Text Logs#

• Capture the Relevant Section: Do not take tens of thousands of lines of /var/log/syslog or /var/log/messages upload it in full. Extract only the relevant entries from 2 to 5 minutes before and after the error timestamp.
• Code Block Formatting: in ticket interfaces that support Markdown, enclose long logs in fenced code blocks. If only plain text is supported, preserve clear line breaks and label the source file path.

2. Screenshot Capture Guidelines#

• 完整保留上下文: when taking a screenshot of an out-of-band console error, retain the console window's title and status bars, including the virtual machine ID or status indicator, so support can confirm it is the correct server.
• Exclude Irrelevant Windows: crop out your browser's personal bookmarks bar, chat windows, and desktop areas containing personal information so support staff can focus on the error itself.

工单生命周期管理与协同原则#

工单提交后，处理过程同样需要遵循规范，以确保支持请求高效推进。

1. Handling Automated Initial Responses (Auto-Responder)#

Many large providers use automated ticket classification and recommendation systems. When you submit a networking or boot issue, the system may immediately send an automated reply containing knowledge-base troubleshooting links.

• Do Not Ignore It: If the system requires user confirmation, a prolonged lack of response may cause the ticket to be automatically marked “Resolved” or “Closed.”
• Respond to the Specific Issue：阅读自动化建议中的检查项，在原工单内逐条回应对照结果（例如：“I have verified step 1 to 3 provided by the automated system; the internal SSH daemon is running normally on port 22, but the external gateway remains unreachable. Please escalate to a network engineer.”), prompting escalation of the ticket to the manual handling queue.

2. Avoid Repeated Follow-Ups and Duplicate Tickets (Ticket Bumping & Duplicate Tickets)#

• Never Open Multiple Tickets for the Same Issue: Opening duplicate tickets with different departments, such as Sales and Technical Support, does not speed up handling. Instead, it can cause automatic thread merging, disrupt engineers' queues, or even trigger anti-abuse controls.
• Follow Up on Tickets with Care: provider ticket queues are usually sorted by “Last Updated.” Repeatedly sending “Any updates?” before a technician replies may reset your waiting position on some ticketing platforms, pushing you backward in the queue.

3. Confirmation Before High-Risk Operations#

When technical support recommends actions such as “OS Reload,” a forced low-level filesystem check (fsck), or restoring the initial snapshot:

• Verify Data Safety：非托管型 VPS 的底层 Reset 通常伴随宿主磁盘数据的 All 重抹。在授权技术人员执行破坏性 Details 前，必须在工单中明确确认自己是否已有离线备份，或要求对方先确认能否挂载救援模式（Rescue Mode / LiveCD）导出重要数据。
• 记录 Details 授权: State authorization clearly and unambiguously in the ticket, for example: “I confirm that all critical data has been backed up off-site. You are authorized to proceed with the OS re-installation.”

故障恢复后的验收与归档#

After support reports that the issue is fixed, run tests in both directions along the original failing path. Do not agree to close the ticket without verification.

1. 路径复测：
   • Networking 类故障: rerun MTR probes from the machine used for the initial tests, checking whether round-trip latency, packet loss, and routing hops have returned to normal levels.
   • System and Hardware Failures: Inspect system load in the console, confirm that disk reads and writes are normal, and run reboot 测试系统能否脱离人工干预顺利完成引导。
   • 计费类故障: Open the client area and check whether the invoice is marked “Paid” and whether the service's “Next Due Date” has been correctly extended according to the billing period.
2. Final Confirmation and Closure：
   • After verification succeeds, reply to the ticket confirming recovery and thanking support, then let the technician or system close it.
   • If the failure was caused by instability in the underlying host, promptly establish scheduled local backups after the ticket is closed and synchronize critical data to another availability zone to reduce the ongoing impact of single-point hardware failures.