wiki madness

[Toshio Kuratomi]

Matt Domsch wrote:
> It's stale on page refresh.  Yesterday, I changed the name of a Site
> using the web UI.  I hit "submit", and the change is committed and the
> page refreshed automatically.  The Site name was unchanged (old value)
> in the text box.  I waited a few seconds, and reloaded the page, and
> the Site name was changed to the new value.
> 
> In controller.py, Site.update() calls site.sync() at the end of the
> update call, and the next call will be to Site.get() which should have
> the new value you'd think.
> 
Matt and I looked at this and we have a theory of what's going wrong as 
well as a fix.

I tried to replicate this on publictest1 without success.  When I tried 
to replicate it on the app servers it returned stale data to me every 
time.  The theory is that since the app servers are alternating in 
sending the page, we're ending up with this sequence of events:

1) app3 serves the site form.
2) User fills out form and submits
3) app4 processes the form results (including the call to site.sync()) 
and raises a redirect back to the form.
4) app3 gets the request and pulls the stale data out of its cache 
because site.sync() wasn't called on this server.

I'm attaching a patch that invalidates the site cache before serving the 
page.  This should fix the stale cache but we'll have to check whether 
it brings back the performance issues that require turning on the cache 
in the first place.  (It should be better than not having caching on at 
all but it still might not be acceptable.  Needs testing.)

Also, the other data on the forms probably needs to have similar calls 
made to prevent stale data.

-Toshio
-------------- next part --------------
A non-text attachment was scrubbed...
Name: mirrorman-controllers.patch
Type: text/x-patch
Size: 526 bytes
Desc: not available
URL: <http://listman.redhat.com/archives/fedora-infrastructure-list/attachments/20071102/d77826b8/attachment.bin>