[7.3.2] Dashboard/nginx log spam from stale Docker template fallback icons

[Claude was used to diagnose this bug, fix the underlying issue and provide the bug report summary]

Expected Behavior

The Dashboard should be able to load Docker container icons without generating filesystem errors, and Docker Manager's own fallback icon file should exist on a standard install.

Actual Behavior

nginx's error log filled up with hundreds of entries per second, all the same:

nginx: [error] ... open() "/usr/local/emhttp/plugins/dynamix.docker.manager/images/question.png" failed (2: No such file or directory) while sending to client, ... referrer: "https://<myserver>/Dashboard"

/var/log is a 128MB tmpfs on my system and it filled to 100% in under an hour from this alone.

The file /usr/local/emhttp/plugins/dynamix.docker.manager/images/question.png (Docker Manager's generic fallback icon) was missing entirely. That alone would just be one broken image, but I found 9 of my Docker templates under /boot/config/plugins/dockerMan/templates-user/ had that exact path saved directly as their <Icon> value. It looks like Docker Manager couldn't fetch these containers' real icon URLs at some point (dead links / moved repos) and substituted question.png, but instead of only falling back at render time, it wrote that substitution back into the template file permanently. All 9 were for containers I'd removed a long time ago — the template files were never cleaned up when the containers were removed, so they were still being enumerated on every Dashboard load, each one requesting the same missing file.

Steps to Reproduce

1. Add a Docker container from a template whose icon URL is (or becomes) unreachable.

2. Docker Manager fails to fetch the icon and saves the question.png path into that template's <Icon> field instead of just falling back visually.

3. Remove the container without manually deleting its template file.

4. Repeat for enough containers over time, and/or have question.png itself go missing from the install.

5. Load the Dashboard — each affected template's icon is requested on every load/refresh.
   

Additional Context

No recent upgrades or hardware changes around when this started. Docker enabled with a large number of containers configured. The 9 affected templates had no matching container left (checked against docker ps -a).

Workarounds

• Copied webGui/images/default.png over the missing question.png path so requests stop failing.

• Deleted the stale template XML files for containers no longer in use.

• Truncated /var/log/syslog and /var/log/nginx/error.log to reclaim space.

Screenshots / Videos / Diagnostics

Diagnostics attached.