Join our Newsletter — 33% off our NHI Course
Home› Glossary› Threats, Abuse & Incident Response› Nginx Loris Attack
Threats, Abuse & Incident Response

Nginx Loris Attack

← Back to Glossary
By NHI Mgmt Group Updated November 16, 2026 Domain: Threats, Abuse & Incident Response

A slow request exhaustion attack against a web server that keeps many connections open by sending partial HTTP requests very slowly. It targets connection handling, thread pools or worker processes, and can reduce availability without needing high traffic volume. The term is usually used for slowloris-style attacks against HTTP servers.

How Nginx Loris Attacks Work

A Nginx Loris attack is a low-rate exhaustion technique that abuses how a server holds sockets, parses HTTP requests, and assigns worker or thread resources. The attacker sends partial requests slowly enough to keep many connections open while consuming little bandwidth, which makes the attack especially efficient against availability.

The core problem is not volume, but resource retention. If a server keeps waiting for request completion, an attacker can monopolise connection slots, worker processes, or upstream capacity with only a modest number of long-lived connections. In practice, that can stall legitimate users long before traffic graphs look unusual.

Why Slow Request Exhaustion Is Effective

Slow request attacks exploit the gap between network activity and application progress. A connection that remains technically alive can still be functionally useless if the server is waiting on headers, body data, or request completion. That makes this class of attack effective against web stacks that do not aggressively bound read time, header size, or connection lifetime.

This is why the term is often associated with slowloris-style behaviour against HTTP servers. The attacker is not trying to crash the host with a flood, but to tie up enough concurrency that the server cannot serve normal clients promptly. The impact is often gradual degradation rather than an abrupt outage.

What Administrators Should Check

Exposure usually depends on how the reverse proxy or web server handles partial requests, idle reads, and client timeouts. Administrators should examine whether the service is internet-facing, whether it terminates many long-lived client connections, and whether upstream application servers inherit the same connection pressure.

It is also important to validate the actual deployment path, not just the software name. Nginx may sit in front of other proxies, application servers, or load balancers, and the weakest timeout or buffering setting in that path often determines whether the environment is exposed.

For incident triage, the useful signals are many open connections with little corresponding request progress, a rise in half-complete HTTP sessions, and worker saturation without proportional throughput. That pattern distinguishes slow request exhaustion from ordinary traffic spikes.

Mitigation, Detection, and Recovery

Defence starts with making partial requests expensive for the attacker and cheap for the server. Tightening read timeouts, limiting header and body processing windows, and enforcing sensible connection caps all reduce the attack surface. Buffering and proxy-layer controls can also help prevent one slow client from holding scarce application resources for too long.

Detection should look for connection-duration anomalies, unusual rates of incomplete requests, and saturation of worker pools or connection tables. After an event, review logs for sessions that were never completed and confirm that no legitimate clients were disproportionately dropped when protective thresholds were applied.

Recovery is usually about restoring fairness: clear stuck sessions, validate timeout settings, and confirm that upstream services are not still constrained by backlog or connection exhaustion. If the attack succeeded, also review whether monitoring missed the early signs because bandwidth usage remained low while concurrency was already under pressure.

Why This Matters for Availability

Slow request exhaustion is a reminder that availability attacks do not need high packet rates. A few carefully timed connections can create real service impact when the target treats incomplete work as a live reservation of capacity. That makes the attack particularly relevant for public web entry points and shared front-end infrastructure.

The operational lesson is that resilience depends on how long the server is willing to wait, not just how much traffic it can ingest. A platform that looks healthy under normal throughput can still be fragile if it cannot quickly shed slow or incomplete client activity.

Risk and Threat Considerations

Slow request exhaustion is dangerous because it creates a denial of service condition without the obvious signature of a traffic flood. The attacker benefits from tying up scarce concurrency, while defenders may initially see only a small number of connections and underestimate the impact.

Failure mechanism: The target retains sockets, workers, or upstream request capacity for incomplete HTTP sessions, so a limited number of slow connections can starve normal users of service.

Impact: Availability degrades, page loads stall, and the service can become unresponsive even when network bandwidth and host CPU do not appear heavily utilised.

Standards & Framework Alignment

This section maps relevant standards and security frameworks to the operational risks and controls described in this guidance.

NIST SP 800-53 Rev 5, NIST CSF 2.0, CIS Controls v8 and OWASP ASVS set the governance and control requirements practitioners need to meet.

FrameworkControl / ReferenceRelevance
NIST SP 800-53 Rev 5SC-5 — Denial of Service ProtectionAddresses service exhaustion and availability protection for web-facing systems
Recommendation — Limit connection lifetimes and resource consumption to resist slow request exhaustion.
NIST CSF 2.0PR.DS-4 — Data is Adequately ProtectedSupports resilience of public services against availability-impacting abuse
Recommendation — Apply rate and timeout safeguards that preserve service availability under abuse.
CIS Controls v8CIS-12 — Network Infrastructure ManagementCovers managing exposed network services and protective configurations
Recommendation — Harden front-end service settings to reduce exposure to slow connection abuse.
OWASP ASVSV12 — Secure CommunicationRelates to protecting HTTP service behaviour and connection handling
Recommendation — Validate timeout and connection handling controls for web-facing endpoints.

Practitioner Guidance

Why practitioners should care: Treat slow request exhaustion as a timeout and concurrency management problem, not just a network abuse problem. The strongest controls are the ones that force incomplete requests to expire quickly and prevent one client from monopolising worker capacity.

What to watch for: Tune and test your reverse proxy or web server under partial-request conditions, then verify that alerts capture long-lived connections, incomplete request patterns, and worker saturation before users feel the outage.

Free weekly newsletter

Subscribe to the NHI & AI Identity Journal

The latest on NHI and Agentic AI security – articles, research, breaches, news and events every week.

Bonus 33% off our NHI Course when you subscribe.

NHIMG Editorial Note
Reviewed and updated by the NHIMG editorial team on November 16, 2026.
NHI Mgmt Group — the #1 independent authority on Non-Human Identity, IAM, and Agentic AI security. nhimg.org