An email with the subject "Abuse Notice" has arrived, containing a ticket number and a 24-hour deadline. The server is still up, the site loads, but the notice text references a specific clause of the terms of use and asks you to respond before it expires. This moment is where the most wrong decisions get made: some people immediately shut down the service, others ignore the notice and wait to "see what happens." Both are wrong.
The abuse policy is a set of rules that defines what behavior is not permitted on the infrastructure, what evidence is accepted, what deadline exists for responding, and at what stage the service is stopped. This text is exactly what the notice refers to; not a moral recommendation, but a process with defined stages and timing.
What counts as an abuse policy violation
The list isn't short, but its categories are clear. The most important cases that in practice lead to a notice:
- Sending bulk email without opt-in from your server. This is the most frequent reason for suspension. If the server IP lands on a blacklist, the notice usually comes from the recipient's side, not from us.
- Port scanning and penetration testing on a server you don't own. Even a simple
nmap -sSagainst a neighboring IP range is a violation. - Hosting phishing or malware, usually through an uploaded script in the
wp-content/uploadsfolder that no one has checked for months. - Illegal content in the legal sense, including copyright infringement with a valid DMCA report.
- Malicious resource consumption: cryptomining, or a script that eats all the CPU with
while(true)and affects neighbors.
There's an important distinction that's often confused in responses: a "resource" violation is not the same as a "behavior" violation. High CPU usage, if caused by real traffic, is a capacity issue and is solved by upgrading the plan. But the same usage, if it comes from a backdoor, is a security issue and follows a completely different handling path.
The handling process from the moment of report to case closure
A case opens with a report. The report can come from another user, from an external organization like Spamhaus, or from our own monitoring system. The first thing the relevant team does is verify the evidence: SMTP logs, web access logs, or tcpdump output. Up to this stage, no action is taken on the service.
If the evidence is sufficient, the notice is sent with these components: case ID, the violated clause, a summary of the evidence, and the response deadline. The default deadline for high-risk cases like active phishing or ongoing spam sending is short and sometimes amounts to a few hours. For lower-risk cases, a one-day or longer window is usually considered. The exact number is always written in the notice itself; rely on that, not on what you read in a forum.
What your response should contain
A response of "Hi, I'll look into it" doesn't close the case. A response that works has three things: acknowledgment of receipt, a root-cause explanation of the problem, and proof of its resolution. For example:
Ticket: AB-2024-1187
Root cause: outdated plugin "contact-form-x" v2.1 allowed arbitrary file upload.
Action taken: plugin removed, uploads directory cleaned, all admin passwords rotated.
Evidence: grep -r "eval(base64" /home/user/public_html → no results
mail queue flushed; postfix log shows no outbound since 14:20 UTC
This structure gets the case closed at the first stage. A vague response pushes the case to the next stage.
Response deadline and what happens after expiry
After the deadline passes without a response, the first action is usually temporary restriction, not data deletion. That means ports 80 and 25 are closed but the files remain in place and you can access them via SSH or the control panel. This separation is deliberate: the goal is to stop the behavior, not to destroy your business.
Full suspension happens when either the behavior continues, or the case is active phishing that is currently victimizing people, or there's a history of repetition. In this state, network access is cut off but the data is retained for a specified period so you can take a backup.
| Violation level | Typical response deadline | Action in case of silence |
|---|---|---|
| Resource consumption, without malicious evidence | Longer, with possibility of extension | Notification and upgrade suggestion |
| Outbound spam, IP on blacklist | Short | Closing port 25 and the mail queue |
| Active phishing or malware | A few hours | Immediate service suspension |
| Repeat violation after resolution | No deadline | Suspension and review of continued cooperation |
Here's where people make a mistake: many assume "since the service is still up, the problem isn't serious" and postpone the notice. The sign we see in practice is this: the site was working at noon, by evening port 25 is closed and transactional emails aren't leaving the queue, while no second notice has been sent. The first deadline is the last warning.
When you are the victim, not the violator
A significant portion of cases involve a server that has itself become a victim: an old WordPress script has been hacked and is being used to send email. In this state you are both accused and affected, and the response must cover both aspects.
The first step is to find the actual source. Sort the logs by request count:
awk '{print $1}' /var/log/nginx/access.log | sort | uniq -c | sort -rn | head -20
grep -c "POST /wp-login.php" /var/log/nginx/access.log
If a specific IP appears with thousands of requests per minute, that's likely the source. For blocking, the guide on blocking malicious IPs covers control panel, htaccess, and firewall paths. But blocking an IP masks the symptom; if the infected file remains on disk, the problem returns. Also check file permissions: directories should be 755 and files 644, and no file should be owned by the web server user with write access to executable scripts. The exact details are in setting secure file permissions in Linux.
If the attack is volumetric and the server goes down under unwanted traffic load, this is no longer a policy violation and relates to the protection layer; DDoS protection is designed exactly for this case and is better configured before it happens, not in the middle of an attack.
Responding to the notice: what to write and what not to write
Three sentences that prolong a case: "We didn't do anything," "Our server hasn't been hacked," and "Please send the evidence." The third isn't bad if the notice genuinely lacked sufficient evidence, but as a first response to a documented notice, it only buys time.
Instead, follow this order:
- Respond within the first 24 hours, even if the fix isn't complete. State that you're investigating and give an approximate resolution time.
- Name the root cause, not the symptom. "Plugin X version 2.1 was vulnerable" is better than "the site was hacked."
- Send evidence. Command output, a log line, or a screenshot from the panel.
- If you need more time, request an extension before expiry. Extensions exist; silence doesn't.
One practical note: if you have several sites on the same server, know that suspension is usually applied at the account level, not the domain. That means an old plugin on a test site can also cut off access to your main site. Separating the test environment onto an independent account costs little and reduces risk significantly.
Prevention: things that don't open a case at all
Most of the notices we see come from three things: a plugin that hasn't been updated in two years, a password reused in several places, and a contact form whose upload field isn't restricted. None of these are complicated.
Enable two-factor authentication on the admin panel and SSH; the guide on enabling two-factor authentication shows the WordPress and control panel paths. Configure security headers so the browser itself deflects part of the attacks. And if you have an upload form, define allowed extensions explicitly; php, phtml, and phar should none of them be on the list.
For transactional email services, monitor the outbound queue. A simple cron that checks the number of queued messages every 15 minutes gives warning far earlier than a blacklist. If the number passes a few hundred, something is sending that shouldn't be.
And finally, if you're not sure whether your service is currently in a normal state, before doing anything check the real-time service status; sometimes what looks like a violation is an infrastructure disruption the technical team is already working on.
Frequently asked questions
What happens if I don't respond to the abuse policy notice?
After the deadline passes, the service is usually temporarily restricted: web and email ports are closed but data isn't deleted and you can access the files via SSH or the control panel. If the behavior continues or the case is active phishing, full suspension is applied.
A request to extend the deadline, before expiry, is usually accepted. Silence is not accepted.
How do I know my server is being used to send spam?
The main sign is a sudden growth in the mail queue and outbound traffic on port 25. Use mailq or postqueue -p to see the number of pending messages and grep "sasl_username" /var/log/mail.log to see which account is sending.
If your IP is listed on Spamhaus or Barracuda, the notice usually comes from the recipient's side and may arrive several hours after sending begins.
Does suspension for a policy violation delete my data?
Not at the first stage. The goal is to stop the behavior, not delete data. Even in full suspension, data is retained for a specified period so you can take a backup. But this period isn't unlimited, and it's better to have an off-server backup from the start rather than relying on it.
What's the difference between an abuse policy violation and a resource consumption issue?
High resource usage, if caused by real traffic, is a capacity issue and is solved by upgrading the plan or optimizing. But if the same usage comes from a backdoor, cryptominer, or malicious script, it's a policy violation and follows a security handling path. Distinguishing these two is the first thing the handling team does.
If you have a notice on your desk right now, the next step is this: take the ticket number, find the root cause in the logs, and respond before the deadline expires. Even if the fix isn't complete.