In OJS security, one of the most common mistakes is to judge an incident only by the visible state of the site. If the journal is still loading, many assume nothing serious happened. If a suspicious payload appears in the logs, others immediately conclude that the system has already been compromised.
Both interpretations can be wrong.
In a recent real-world case analyzed by OJS One, the starting point was a pattern that now appears frequently on servers hosting academic journals: increased suspicious traffic, automated requests, injection attempts against exposed endpoints, and systematic scanning of public-facing surfaces.
The most important finding was not simply the presence of malicious payloads. It was understanding which signals actually matter and which ones are still only failed attempts that should not be mistaken for confirmed compromise.
The Distinction That Prevents Costly Mistakes
In OJS, confusing reconnaissance with compromise leads to two equally dangerous operational errors:
- overreacting and declaring a major incident without sufficient evidence
- underestimating signals that do justify containment, privilege review, and deeper auditing
The purpose of this case is not just to describe hostile traffic. It is to show how to separate:
- automated noise
- a real exploitation attempt
- a strong indicator of compromise
That distinction determines the quality of the response.
Case Context
The case began on a server hosting more than one OJS installation in production, including at least one legacy deployment and another instance running on an older branch.
The initial observation was clear:
- unusual traffic growth in web logs
- multiple automated requests
- patterns consistent with mass scanning
- injection attempts against routes that should never receive that kind of input
This matters because many institutions still interpret logs in a binary way:
- if the site is still working, nothing happened
- if there was a
200 OK, we were compromised - if the firewall blocked something, the problem is over
All three conclusions can be misleading.
OJS security requires a more precise reading.
The Main Finding: SQLi Attempts Against an Exposed Endpoint
One of the main focuses of the analysis was a public endpoint related to bibliographic citation generation. Under legitimate use, the parameter being queried should contain a simple numeric identifier.
However, the evidence reviewed showed multiple requests in which that parameter contained SQL payloads instead of the expected value.
The payload families observed were associated with well-known SQL injection techniques, including:
EXTRACTVALUEUPDATEXMLGTID_SUBSETINFORMATION_SCHEMAORDSYS.ORD_DICOM.GETMAPPINGXPATHUNION- variants using
CAST,CONCAT, and different forms of obfuscation
From an operational standpoint, that means one thing: the endpoint was being actively tested by bots or automated actors looking for an exploitable weakness.
That does not automatically mean the exploitation succeeded.
That nuance is essential in any serious OJS security audit.
What It Really Means to See SQLi Payloads in OJS Logs
When SQLi strings appear in logs, the first mistake is to overreact. The second is to ignore them.
The correct interpretation is this:
- the actor has already identified a potentially interesting endpoint
- they are testing injection using multiple payload families
- they are measuring how the system responds
- it is probably part of a broader automated campaign
What this does not prove on its own is:
- confirmed database access
- confirmed payload execution
- privilege escalation
- effective server compromise
In other words:
a SQLi attempt is evidence of hostile activity, not automatic proof of successful exploitation.
That is why, in OJS, looking at the payload alone is not enough. You have to correlate it with other indicators.
Quick Interpretation: What Each Signal Means
| Observed signal | What it indicates | Operational severity |
|---|---|---|
| SQLi payload in logs | Reconnaissance or exploitation attempt | Medium |
Scan of /.env or /.git/config |
Search for exposed secrets | Medium |
200 OK on a tested endpoint |
Endpoint response, not confirmed compromise | Medium |
| New account with elevated privileges | Possible application compromise | High |
| Anomalous session on a critical account | Possible misuse of credentials | High |
| Unexpected changes in themes, plugins, or config | Possible active system alteration | High |
When Noise Stops Being Just Noise
In the time window analyzed, activity against that endpoint showed a pattern that could no longer be treated as casual traffic:
- more than two thousand requests to the endpoint
- a meaningful share of legitimate requests with numeric identifiers
- more than one thousand non-numeric requests or requests containing payloads
- several hundred malicious hits tied to specific SQLi families
Additional related patterns also appeared:
- SQLi attempts against the target endpoint
- scans for secrets such as
/.envand/.git/config - fingerprinting of CVEs, panels, and exposed surfaces
- probes aimed at WordPress, even though the server was not running WordPress
- repeated access to
robots.txtas part of reconnaissance activity
These data points matter for two reasons.
First, they show that the server was not receiving normal editorial traffic alone, but also automated reconnaissance and exploitation campaigns.
Second, they explain something many editorial teams fail to notice in time: attacks against OJS rarely arrive in isolation. What you usually see is a mix of fingerprinting, exposed-secret probing, route testing, SQLi attempts, and exploration of surfaces unrelated to the actual stack, simply because the automation is broad and opportunistic.
Other Attempts Observed Beyond the Main Endpoint
The citation endpoint was not the only signal. Other repeated patterns also appeared, consistent with real automated campaigns.
Exposed Secret Scanning
Requests to paths such as:
/.env/.git/config
These are relevant because they are attempts to discover credentials, sensitive variables, or misconfigured repositories accessible over the web.
Basic Reconnaissance
Repeated requests to:
/robots.txt
Although robots.txt is a legitimate file, in an attack context it is often used for fingerprinting and discovery of potentially useful routes.
Scanning of Surfaces Unrelated to the Actual Stack
There were also attempts against paths associated with other panels, devices, or technologies that were not even part of the real server environment.
That indicates something important: the host was receiving wide horizontal scanning, not just attacks designed specifically for OJS. The bot sweeps across multiple technologies, panels, and appliances until it finds a promising surface.
For any journal, that means it is not enough to simply have OJS up and running. You should assume that any Internet-exposed service is automatically in scope for these campaigns.
Which IOCs Really Matter in an OJS Incident
In OJS security, not all indicators carry the same weight. There are weak signals and strong signals.
Weak Signals
- SQLi payloads in logs
- scans of common routes
- attempts against
/.env,/.git/config, and generic surfaces 404,403, or even200responses without correlated impact
These matter, but on their own they do not confirm compromise.
Strong Signals
- appearance of new global administrator accounts without an authorized change
- appearance of new high-privilege journal-level accounts without authorization
- recent sessions on critical accounts from unusual IPs or user agents
- unexpected changes in configuration, themes, plugins, or templates
In OJS, these are the IOCs that truly change the severity of the incident.
Because the payload shows intent. But the privilege change, anomalous session, or code alteration brings you much closer to confirmed compromise.
What to Do First When You Detect This Pattern
If an OJS administrator detects SQLi payloads, repeated scans, and anomalous behavior in logs, the initial order of operations should be:
- contain the obvious noise with firewall rules, rate limiting, and automated blocking
- preserve basic evidence: web logs, application events, IPs, and time windows
- review privileged users, recent sessions, and administrative events inside OJS
- inspect unexpected changes in configuration, plugins, themes, and sensitive files
- decide whether the case is still a failed attempt or whether there are already real IOCs of compromise
This order matters because it avoids two common mistakes: cleaning too early and losing evidence, or spending all your time blocking IPs while the incident has already progressed at the application level.
What Was Verified in This Case
The operational review included controls aimed at separating noise from real damage:
- count of attacks by endpoint and pattern
- cross-checking attacking IPs against blocked IPs
- validation of
fail2banandufwstatus - review of events, sessions, and roles within the OJS database
- review of file changes outside cache-related paths
This is worth emphasizing. In many environments, the response stops at the firewall: IPs are blocked and the case is considered closed.
That is not enough.
In OJS, a serious response needs to examine at least three layers at the same time:
- web layer: logs, routes, patterns, and frequency
- application layer: users, roles, sessions, and events
- filesystem layer: changes in sensitive files, plugins, themes, and configuration
If you only review one layer, you may miss the signal that actually matters.
Outcome of the Critical IOC Review in This Case
Within the review window, there was no direct evidence of critical privilege escalation through the recent creation of new global administrator accounts.
That does not mean the incident should be minimized. It means something else: with the evidence available at that point, there was no direct proof that the observed attempts had led to that specific type of compromise.
That distinction is fundamental in a professional audit.
Saying there was an attack when there was an attack is correct. Saying there was confirmed compromise when there is not enough evidence is not.
Even so, this control should remain a permanent alarm point. In OJS, reviewing the creation of privileged accounts remains one of the most valuable checks available.
What an OJS Administrator Should Take Away From This Case
This case leaves several clear operational lessons.
First: a 200 OK in response to a SQLi payload does not, by itself, mean successful exploitation.
Second: repeated hostile attempts in logs do justify hardening controls, blocking repeat offenders, and reviewing correlations across the application and filesystem.
Third: in OJS, monitoring privileged roles and critical sessions is often more valuable than looking only at the firewall.
Because the firewall reduces noise. But privilege monitoring helps you determine whether the incident has crossed the line from attempt to compromise.
If you also detect unusual plugin changes, suspicious administrative activity, or artifacts that should not exist in the application tree, then the case should no longer be treated as simple noise. That usually means the incident has already moved beyond reconnaissance.
And if the investigation begins to point to persistence outside the webroot or suspicious activity in the server user environment, then the pattern is even deeper, as described in our separate security audit involving malware persistence outside the webroot.
Brief Checklist for OJS Administrators
- review recent creation of accounts with elevated privileges
- review recent sessions on critical accounts, including IP and user agent
- review unexpected changes in configuration, themes, and plugins
- apply automated blocking for repeated SQLi patterns
- enforce HTTPS and verify secure session cookie settings
- maintain a real upgrade plan to move off legacy branches
And if the server is still operating without verified backups, a WAF, or sufficient hardening, then the problem is no longer just monitoring. It is also architectural. For that, it is worth reviewing our post on off-site backups, OJS security, and production hardening.
Frequently Asked Questions
Does seeing a SQLi payload in OJS logs mean the database was exploited?
No. It means there was an exploitation attempt or automated probing against an endpoint. To speak of confirmed compromise, you need correlated evidence in the application, privileges, sessions, or filesystem.
Which IOC matters more in OJS: the payload or privilege changes?
Privilege changes, anomalous sessions, and unexpected alterations in sensitive files carry much more weight as confirmation of an incident. The payload shows aggression; the operational change shows possible real impact.
Is blocking malicious IPs at the firewall enough?
No. Blocking IPs helps reduce noise, but it does not replace the review of users, roles, sessions, sensitive files, and persistence mechanisms on the server.
What should an editor or technical OJS administrator review first?
Start with privileged access, recent sessions, and unexpected administrative changes. Then move to sensitive files and correlation with logs. The goal is to determine whether you are dealing with scanning only or whether there is already real operational impact.
Conclusion
Automated attacks against OJS are no longer an exception. They are a normal part of the operating environment for any journal exposed to the Internet.
The important question is not whether your installation receives scans, payloads, or exploitation attempts. The important question is whether your team knows how to distinguish between noise, a real attempt, and a strong indicator of compromise.
In editorial security, that distinction defines the quality of the response.
Because in OJS, the exploit gives you a warning. But the privilege change, anomalous session, or code alteration is what truly brings you closer to confirming the incident.
Is your current journal suffering from slowness or constant downtime?
Get in touch with our team and find out how enterprise-grade infrastructure can transform your editorial operations.
Request an assessmentOJS 360 Assessment
If you prefer to start with a diagnosis
The OJS 360 Assessment reviews the state of your journal before you decide.
OJS 360 Assessment