775132 - When templates change, even a force refresh is not causing pages using them to update.

Status: VERIFIED FIXED

(developer.mozilla.org Graveyard :: Editing, defect)

Description

[Trevor]

It seems that the template is being cached. If you change the parameters passed it will produce the correct output for the old version of the template.

After several minutes a force refresh will pick up the changes.

Comment 1

[John Karahalis [:openjck]]

Thanks as always, Trevor.

Les: Is this a dupe of bug 771236?

Comment 2

[Les Orchard [:lorchard]]

(In reply to John Karahalis [:openjck] from comment #1)
> Thanks as always, Trevor.
> 
> Les: Is this a dupe of bug 771236?

No I don't think so, that bug is about the editor, this is about rendering pages. separate systems

Comment 3

[Les Orchard [:lorchard]]

Investigating this. Here's what I've tried so far this morning, no ideas yet:

* No luck reproducing on laptop VM, or on mdn.lmorchard.com

* Can reproduce on developer-new.mozilla.org

* Editing a template and fetching its ?raw source works as expected - ie. the content is updated after an edit

* Creating a new template and adding a macro to call it in a document results in changes to the page. It's just editing existing templates that come sup with stale content. That feels like caching.

* Doesn't seem like memcache: I edited a template; shift-reload on test page did nothing; flushed memcache on developer{1,2}; shift-reload still did nothing

* Watching the kumascript logs on both developer{1,2} shows that shift-reload does indeed trigger a kumascript request.

Next step is probably to add some more log output to kumascript, especially around cache hits/misses and see what that says.

Comment 4

[Les Orchard [:lorchard]]

(In reply to Les Orchard [:lorchard] from comment #3)

> * Editing a template and fetching its ?raw source works as expected - ie.
> the content is updated after an edit

Actually, I take this back. I just saw a curl fetch of ?raw template source come back stale on developer1 from the command line. Maybe a lead, thinking Zeus + caching headers might be involved.

Comment 5

[[github robot]]

Commits pushed to master at https://github.com/mozilla/kuma

https://github.com/mozilla/kuma/commit/84b3cdd74cb4d0a576a56ccfcc356c7f8b39812c
fix bug 775132: Kumascript update for sending headers in Kuma requests

https://github.com/mozilla/kuma/commit/f255106466471c03f5155640bfc41bf83de91522
Merge pull request #446 from lmorchard/send-the-headers-please-bug-775132

fix bug 775132: Kumascript update for sending headers in Kuma requests

Comment 6

[raymond [:retornam] (inactive)]

verified fixed https://developer-new.mozilla.org/en-US/docs/Template:CXCLicenseBlock