# Admin page shows errors after upgrading to v3.5.3
**URL:** https://meta.discourse.org/t/v3-5-3/398634
**Category:** Self-hosting
**Tags:** unsupported-install
**Created:** [March 17, 2026, 1:02pm UTC](https://meta.discourse.org/t/v3-5-3/398634 "2026-03-17T13:02:41Z")
**Posts on this page:** 14
**Page:** 1
### Author:  [@xwj](https://meta.discourse.org/u/xwj)
#### Post date: [March 17, 2026, 1:02pm UTC](https://meta.discourse.org/t/v3-5-3/398634/1 "2026-03-17T13:02:41Z")
The admin page is showing errors after upgrading from version v3.5.2 to v3.5.3
And reports the following error:
> loader.js:247 Uncaught (in promise) Error: Could not find module `discourse/admin/models/admin-plugin` imported from `discourse/plugins/docker_manager/discourse/models/repo`
> at loader.js:247:1
> at a (loader.js:258:1)
> at s.findDeps (loader.js:168:1)
> at a (loader.js:262:1)
> at s.findDeps (loader.js:168:1)
> at a (loader.js:262:1)
> at requireModule (loader.js:24:1)
> at n.i [as getRoute] (index.ts:121:18)
> at p.\_getQPMeta (index.ts:101:20)
> (anonymous) @ loader.js:247
> a @ loader.js:258
> (anonymous) @ loader.js:168
> a @ loader.js:262
> (anonymous) @ loader.js:168
> a @ loader.js:262
> requireModule @ loader.js:24
> i @ index.ts:121
> \_getQPMeta @ index.ts:101
It works normally after disabling plugins, but I can’t pinpoint which plugin is causing the issue. Can you help me take a look?
---
### Author:  [@xwj](https://meta.discourse.org/u/xwj)
#### Post date: [March 17, 2026, 1:15pm UTC](https://meta.discourse.org/t/v3-5-3/398634/2 "2026-03-17T13:15:50Z")
The issue persists when “Disable unofficial client plugin customizations” is enabled in [Safe Mode](https://meta.discourse.org/t/53504?silent=true), but it works correctly when “Disable all client plugin customizations” is enabled.
---
### Author:  [@NateDhaliwal](https://meta.discourse.org/u/NateDhaliwal)
#### Post date: [March 17, 2026, 1:56pm UTC](https://meta.discourse.org/t/v3-5-3/398634/3 "2026-03-17T13:56:45Z")
3.5.3 is no longer supported. Please see [https://releases.discourse.org](https://releases.discourse.org).
---
### Author:  [@Heliosurge](https://meta.discourse.org/u/Heliosurge)
#### Post date: [March 17, 2026, 2:28pm UTC](https://meta.discourse.org/t/v3-5-3/398634/4 "2026-03-17T14:28:49Z")
As Nate said that is a very old version now. Maybe try a command line update.
---
### Author:  [@xwj](https://meta.discourse.org/u/xwj)
#### Post date: [March 18, 2026, 1:54am UTC](https://meta.discourse.org/t/v3-5-3/398634/5 "2026-03-18T01:54:06Z")
Which version do you recommend upgrading to? Will skipping multiple versions cause more compatibility issues?
---
### Author:  [@xwj](https://meta.discourse.org/u/xwj)
#### Post date: [March 18, 2026, 7:38am UTC](https://meta.discourse.org/t/v3-5-3/398634/6 "2026-03-18T07:38:15Z")
It failed immediately after upgrading to 2026.1.1
---
### Author:  [@merefield](https://meta.discourse.org/u/merefield)
#### Post date: [March 18, 2026, 7:45am UTC](https://meta.discourse.org/t/v3-5-3/398634/7 "2026-03-18T07:45:34Z")
Check your `/logs` for suspect messages.
---
### Author:  [@Ed\_S](https://meta.discourse.org/u/Ed_S)
#### Post date: [March 18, 2026, 8:13am UTC](https://meta.discourse.org/t/v3-5-3/398634/8 "2026-03-18T08:13:33Z")
If the web upgrade fails, you need to use ssh to login as root and do a command line upgrade.
---
### Author:  [@xwj](https://meta.discourse.org/u/xwj)
#### Post date: [March 18, 2026, 8:48am UTC](https://meta.discourse.org/t/v3-5-3/398634/9 "2026-03-18T08:48:19Z")
---
### Author:  [@darkpixlz](https://meta.discourse.org/u/darkpixlz)
#### Post date: [March 18, 2026, 8:50am UTC](https://meta.discourse.org/t/v3-5-3/398634/10 "2026-03-18T08:50:56Z")
Can you try a command line rebuild as the reply above suggests? That will likely fix it, I’ve never seen an issue from version jumps..
---
### Author:  [@xwj](https://meta.discourse.org/u/xwj)
#### Post date: [March 19, 2026, 1:40am UTC](https://meta.discourse.org/t/v3-5-3/398634/11 "2026-03-19T01:40:52Z")
What is the specific command? My service is packaged into a Docker image and deployed on a remote server’s K8s cluster. The deployment uses the `web_only` method.
Additionally, when an error occurs on the administrator page, the following error message appears:
---
### Author:  [@darkpixlz](https://meta.discourse.org/u/darkpixlz)
#### Post date: [March 19, 2026, 1:42am UTC](https://meta.discourse.org/t/v3-5-3/398634/12 "2026-03-19T01:42:42Z")
> [@xwj](#):
>
> Could you please tell me the specific command?
Go to where you installed Discourse and type `./launcher rebuild app` - wherever `discourse_docker` is. I believe `/var/discourse` is the “recommended” place to put the script but it might be different for you.
---
### Author:  [@pfaffman](https://meta.discourse.org/u/pfaffman)
#### Post date: [March 19, 2026, 2:07am UTC](https://meta.discourse.org/t/v3-5-3/398634/13 "2026-03-19T02:07:37Z")
> [@xwj](#):
>
> What is the specific command? My service is packaged into a Docker image and deployed on a remote server’s K8s cluster
That’s not supported. You’ll need to do something to load the latest image. You’ll need to migrate the database. There’s usually no problem upgrading multiple versions.
---
### Author:  [@xwj](https://meta.discourse.org/u/xwj)
#### Post date: [March 19, 2026, 9:45am UTC](https://meta.discourse.org/t/v3-5-3/398634/14 "2026-03-19T09:45:13Z")
It seems the database migration failed. The following error was reported.
> TOP =\> db:migrate =\> assets:precompile:asset\_processor │
> │ full trace by running task with --trace) │
> │ executing /etc/runit/1.d/00-ensure-links │
> │ executing /etc/runit/1.d/01-cleanup-web-pids │
> │ executing /etc/runit/1.d/anacron │
> │ executing /etc/runit/1.d/cleanup-pids │
> │ stale PID files │
> │ executing /etc/runit/1.d/copy-env │
> │ runsvdir, PID is 1126 │
> │ aborted! │
> │ pnpm -C=frontend/asset-processor node build.js (Discourse::Utils::CommandError) │
> │ Failed to switch pnpm to v10.28.0. Looks like pnpm CLI is missing at “/home/discourse/.local/share/pnpm/.tools/pnpm/10.28.0/bin” or is incorrect │
> │ /home/discourse/.local/share/pnpm/.tools/pnpm/10.28.0/bin/pnpm EACCES
When I checked the specific pnpm path, I found that the path existed, but the `pnpm.cjs` file lacked execute permissions, which caused the error.
> drwxrwxrwx root root /
> drwxr-xr-x root root home
> drwxr-x— discourse discourse discourse
> drwxr-x— discourse discourse .local
> drwxr-x— discourse discourse share
> drwxr-x— discourse discourse pnpm
> drwxr-x— discourse discourse .tools
> drwxr-x— discourse discourse pnpm
> lrwxrwxrwx discourse discourse 10.28.0 → 10.28.0\_tmp\_9\_0
> drwxr-x— discourse discourse 10.28.0\_tmp\_9\_0
> drwxr-x— discourse discourse bin
> lrwxrwxrwx discourse discourse pnpm → ../node\_modules/pnpm/bin/pnpm.cjs
> drwxr-x— discourse discourse ..
> drwxr-x— discourse discourse node\_modules
> drwxr-x— discourse discourse pnpm
> drwxr-x— discourse discourse bin
> -rw-r----- discourse discourse pnpm.cjs
Later, after pre-installing the corresponding version of pnpm using `corepack` in the image, the database migration succeeded. The service is now normal.