---
title: 'Struts 2.5 end of life: what to check before you decide'
description: 'Struts 2.5 end of life: Apache no longer ships fixes. Check the branch, runtime, server and workflows before choosing support or a code change.'
url: 'https://dfuture.co/insights/struts-2-5-end-of-life-application-checks/'
type: 'article'
typeLabel: 'Article'
language: 'en'
date: '2026-11-17'
categories: 'legacy-modernization'
---

# Struts 2.5 end of life: what to check before you decide

> Struts 2.5 end of life: Apache no longer ships fixes. Check the branch, runtime, server and workflows before choosing support or a code change.

Your application runs on Struts 2.5, and a decision is due: support, an environment change or a project. Struts 2.5 end of life means the branch no longer gets fixes from Apache; it does not say which route fits. The next step is a check of the branch and the stack.

## What Apache says about Struts 2.5

Apache [announced on 2023-10-30](https://struts.apache.org/struts25-eol-announcement) that Struts 2.5 would reach end of life in six months, with only security fixes until then. The last 2.5 release, 2.5.33, shipped on 2023-12-07 ([Apache announcements](https://struts.apache.org/announce-2023)), and Apache's [end-of-life page](https://struts.apache.org/eol-versions.html) records the branch under the announcement date, so the announcement and that page agree on 2023-10-30.

That page defines the status: "the Apache Struts Team no longer provides security patches, bug fixes, or updates for that branch". It dates the Struts 1 end of life to 2013-04-05 and the 2.3.x branch to 2019-09-12, so Struts 2 end of life has no single date.

Apache's [Download page](https://struts.apache.org/download.cgi) lists two lines that receive fixes, 7.x and the 6.x maintenance line, and calls every other version end of life. The 6.x maintenance line is the one with a Java 8 minimum; 7.x requires Java 17. Apache publishes no end date for 6.x.

### Confirm the branch the application ships

The version that counts is the one in the deployed archive. Ask the team to check on the production server:

1. Take the deployed archive from the server, not from the source repository.
2. List every Struts JAR in it: the core, XWork (separate on 2.3, inside the core on 2.5) and each plugin, because Apache bulletins name plugins one by one.
3. Read `Implementation-Version` from each JAR's manifest or from `pom.properties`. The file name is the weakest evidence: a renamed file keeps its manifest.
4. On WebSphere, check all three places a Struts JAR can sit: the application archive, a shared library bound to the application and a shared library bound to the server. IBM's shared library and class loader documentation explains why the loaded Struts copy may not come from the archive; read the manifest of the copy the class loader resolves.
5. Compare the result with the build. A dependency tree describes the source code, not the deployment.
6. Record any mismatch as an open question.

A scanner finding opens an inventory row; the manifest check closes it.

## Check the stack around Struts

The branch the application is on decides nothing about the runtime or the server; the target line sets the floor. Apache's [general availability announcements](https://struts.apache.org/announce-2026) declare the minimums: the 6.x maintenance line requires Servlet 3.1, JSP 2.1 and Java 8, and the 7.x line requires Java 17 and Jakarta EE. IBM documents WebSphere traditional 9.0.5 at [Java EE 7 with Servlet 3.1](https://www.ibm.com/docs/SSEQTP_9.0.5/com.ibm.websphere.base.doc/ae/rovr_specs_javaee7.html), with [IBM SDK 8 as default Java](https://www.ibm.com/docs/en/SSEQTP_9.0.5/was/ae/welc_newinrelease.html), and version 8.5 at [Java EE 6 with Servlet 3.0](https://www.ibm.com/docs/en/SSEQTP_8.5.5/com.ibm.websphere.base.doc/ae/tovr_migrating_javaee.html).

By those declared minimums, a move to 6.x fits WebSphere traditional 9.0.5 on IBM SDK 8 without a change of server or runtime. By the same minimums, a move to 7.x points to a change of environment: the cited IBM 9.0.5 pages name Java EE 7 and IBM SDK 8 and mention neither Java 17 nor Jakarta EE. On 8.5.5 the Servlet 3.1 minimum of 6.x is not met. On WebLogic, compare the server's servlet and Java levels with the target line's minimum.

Neither the runtime nor the server forces a date. Oracle offers Extended Support for Java SE 8 until December 2030 ([Oracle roadmap](https://www.oracle.com/java/technologies/java-se-support-roadmap.html)), and IBM states that there is no planned end-of-support date for WebSphere 8.5.5 and 9.0.5 ([IBM product lifecycle](https://www.ibm.com/support/pages/node/873142)). The Java 8 support channel and the WebSphere platform decision are separate questions outside this article.

**Limit of this method:** these are minimums declared by Apache and levels documented by IBM, not compatibility tested on your server and runtime.

The cited IBM 9.0.5 pages name IBM SDK 8 as the default Java SDK and do not mention Java 17, so a newer runtime on a given 9.0.5 server is a fact to read from that server. On WebSphere 8.5.5, IBM documents Servlet 3.0 against the 3.1 minimum of 6.x, so a 6.x move starts with the server level.

Record the stack in one row per application:

| Struts branch | Exact version | Java runtime and vendor | Application server | OS | Fix source | Affected workflow | Owner | Open question |
|---|---|---|---|---|---|---|---|---|
| 1.x, 2.3, 2.5 or later | From the deployed manifest | Version and vendor | Product, version, fix pack | Version | Apache line, support channel or none | Business process using it | Named role | What is unknown |

## What end of life does and does not prove

End of life establishes that fixes have stopped; it does not establish that the application is affected. Apache's [end-of-life page](https://struts.apache.org/eol-versions.html) states that no fixes ship for 2.5.

Apache's [security bulletins](https://cwiki.apache.org/confluence/display/WW/Security+Bulletins) still list affected 2.5 versions as end of life, with fixes only on the 6.x and 7.x lines.

| What end of life establishes | What it does not establish |
|---|---|
| Apache ships no new fixes for the branch | That the shipped version is in a bulletin's range |
| A new issue reaching 2.5 stays open there | That the bulletin's condition holds |
| An Apache fix means moving to a supported line | That no workaround exists |
| The project offers no support for the branch | That anyone is attacking the application |

Three questions decide whether a bulletin applies:

1. Is the shipped version inside the bulletin's range?
2. Does its condition hold: file upload in use, a particular action mapper, the named plugin?
3. Is there a workaround, or a fix on a supported line that the stack can take?

Record each bulletin in one of four states, with the reason:

- **Not affected**: out of range, or the condition does not hold.
- **Affected**: in range, and the condition holds.
- **Fixed**: a workaround or an upgrade removed the exposure.
- **Under investigation**: version or condition not yet confirmed.

**Limit of this method:** a "not affected" record on a branch without fixes holds until the next bulletin; on 2.5, each new bulletin reopens the review.

## Compare the routes for one application

![Three nodes: branch, stack and route. The arrow from stack to route is dashed until the first two nodes are filled; the route node lists five routes.](https://dfuture.co/images/insights/struts-2-5-end-of-life-application-checks-diagram-v4.webp)

The filled inventory decides the route; each route fits only under its conditions.

| Route | Fits when | Check first |
|---|---|---|
| Support or patch channel | A third-party extended support offer covers the applicable bulletins; Apache's [end-of-life page](https://struts.apache.org/eol-versions.html) lists such offers without endorsing them | Stated coverage against the inventory's plugins and bulletins |
| Environment change | The target line's minimums do not fit the current runtime or server | The levels the target line declares |
| Incumbent team | The team running the application knows the code and has room for the change | Whether it can read the deployed archive |
| Bounded code change | A fix on a supported line changes application code | Which actions, plugins and interceptors it reaches |
| No project yet | A bulletin workaround applies, or the bulletin's condition does not hold | Who reviews the next bulletin |

A supported line can also mean code work: one fix described in an Apache security bulletin for a file upload problem required a move to 6.4.0 and a rewrite of the upload actions.

**Limit of this method:** a support channel's coverage is the vendor's claim, to be checked against the bulletins that apply to the application; this article verifies no vendor's coverage.

Six questions to answer before deciding on a move from 2.5 to 6.x, from Apache's own material:

1. Which version does the deployed archive ship, and which version does the applicable bulletin name as the fix?
2. Does that fix change an API the application code uses? One bulletin states of its upload fix: "This change isn't backward compatible as you must rewrite your actions to start using the new Action File Upload mechanism."
3. Which plugins does the inventory list? Bulletins name plugins one by one, and some conditions sit in a single optional plugin such as REST.
4. Which interceptors and action mappers does the configuration use? Bulletin conditions name them, such as the file upload interceptor.
5. Do the Java and Servlet minimums of 6.x fit the current runtime and server?
6. Which entries of the Migration Guide that Apache's end-of-life announcement links, for the versions between the current one and the target, touch the application's actions, plugins, interceptors and configuration? The announcement calls migration to 6.x "smoother" than to 7.x.

### Work through an illustrative application

Illustrative application, assembled for this article; not a client system.

**Limit of this method:** DFG has no published Java case, so the example is synthetic and records no outcome.

An internal policy-servicing application whose build file says Struts 2.5:

| Struts branch | Exact version | Java runtime and vendor | Application server | OS | Fix source | Affected workflow | Owner | Open question |
|---|---|---|---|---|---|---|---|---|
| 2.5, per the build file | Not yet read from the deployed archive | IBM SDK 8 | WebSphere traditional 9.0.5 | AIX | None from Apache; support channel not checked | Policy document upload; file upload in use | Application owner (role) | Is the REST plugin deployed? Does a server shared library carry its own Struts copy? |

Decision record: `no decision yet`. Next step: read the manifest in the deployed archive and check the server shared library.

## Define what a checkable change must prove

A checkable change proves four things, and the production release stays a separate decision:

- a reproducible build that matches what will be deployed;
- named workflows that run end to end;
- known differences from the current behavior, listed;
- release to production as a separate step with its own approval.

A diagnosis is separate from that code proof and from the release. Data migration and ongoing support stay outside it.

## Record the branch, the stack and the open question

The record for one application holds:

- the inventory row, with the version read from the deployed archive;
- a state for each applicable bulletin, with its reason;
- one outcome: support or patch channel, environment change, bounded code change or no project yet;
- the open question that still blocks the decision, with its owner.

If one application blocks a platform step, bring the component, version and workflow to discuss the next justified result.

[Discuss one application](https://dfuture.co/enterprise-application-modernization/)

---
HTML: https://dfuture.co/insights/struts-2-5-end-of-life-application-checks/
Markdown: https://dfuture.co/insights/struts-2-5-end-of-life-application-checks.md
