Strange caching? problem


#1

I got up this morning to discover a strange occurrence on my blog
server.

My blog http://talklikeaduck.denhaven2.com/ seems to be serving up a
blank page for JUST the landing page.

The admin pages are fine, if I enter the url of an article page it
comes back fine. But the base url comes back blank.

Tailing the production log indicates that requests to that url aren’t
even hitting the rails app. I’ve swept the page cache using the typo
admin ui, I’ve restarted the blog using cap.

The same thing happens with Safari 4, Firefox, and Curl, so it’s not a
browser caching issue.

I suspect that Apache and Passenger are not sending the request
through, I restarted apache on the server and still the same thing.

Ideas anyone?


Rick DeNatale

Blog: http://talklikeaduck.denhaven2.com/
Twitter: http://twitter.com/RickDeNatale
WWR: http://www.workingwithrails.com/person/9021-rick-denatale
LinkedIn: http://www.linkedin.com/in/rickdenatale


#2

On May 26, 12:26 pm, Rick DeNatale removed_email_address@domain.invalid wrote:

The same thing happens with Safari 4, Firefox, and Curl, so it’s not a
browser caching issue.

I suspect that Apache and Passenger are not sending the request
through, I restarted apache on the server and still the same thing.

Ideas anyone?

Not so much an idea as what I would look at next; have you tried using
tcpdump (or wireshark, ethereal, etc) to see what the http requests/
responses look like ?

Fred


#3

On Tue, May 26, 2009 at 7:34 AM, Frederick C.
removed_email_address@domain.invalid wrote:

Not so much an idea as what I would look at next; have you tried using
tcpdump (or wireshark, ethereal, etc) to see what the http requests/
responses look like ?

No. I guess I could set up to do that on the server, but I’m not sure
what to look for, I don’t know enough about Passenger to know where
tcp is used between Apache - Passenger and the rails app, I think that
the rails app actually runs in one or more Passenger processes doesn’t
it?


Rick DeNatale

Blog: http://talklikeaduck.denhaven2.com/
Twitter: http://twitter.com/RickDeNatale
WWR: http://www.workingwithrails.com/person/9021-rick-denatale
LinkedIn: http://www.linkedin.com/in/rickdenatale


#4

On Tue, May 26, 2009 at 8:13 AM, Hassan S.
removed_email_address@domain.invalid wrote:

On Tue, May 26, 2009 at 4:26 AM, Rick DeNatale removed_email_address@domain.invalid wrote:

…, I restarted apache on the server and still the same thing.

Did you do a restart, or did you stop httpd and use ps to make sure
all threads had exited, then start clean?

If not, I’d try that first.

Tried that, no joy.


Rick DeNatale

Blog: http://talklikeaduck.denhaven2.com/
Twitter: http://twitter.com/RickDeNatale
WWR: http://www.workingwithrails.com/person/9021-rick-denatale
LinkedIn: http://www.linkedin.com/in/rickdenatale


#5

On Tue, May 26, 2009 at 4:26 AM, Rick DeNatale removed_email_address@domain.invalid
wrote:

…, I restarted apache on the server and still the same thing.

Did you do a restart, or did you stop httpd and use ps to make sure
all threads had exited, then start clean?

If not, I’d try that first.

Hassan S. ------------------------ removed_email_address@domain.invalid


#6

On Tue, May 26, 2009 at 5:23 AM, Rick DeNatale removed_email_address@domain.invalid
wrote:

Did you do a restart, or did you stop httpd and use ps to make sure
all threads had exited, then start clean?

Tried that, no joy.

Ouch. And you’re absolutely sure there’s no caching appliance
anywhere in the path to your server? Shutting down httpd and
sending a request results in a no connection message?

Any recent config changes? System auto-updates?


Hassan S. ------------------------ removed_email_address@domain.invalid


#7

On Tue, May 26, 2009 at 7:26 AM, Rick DeNatale removed_email_address@domain.invalid
wrote:

admin ui, I’ve restarted the blog using cap.

The same thing happens with Safari 4, Firefox, and Curl, so it’s not a
browser caching issue.

I suspect that Apache and Passenger are not sending the request
through, I restarted apache on the server and still the same thing.

Okay, I discovered that there was an empty file in public/index.html,
I renamed it and now the root page is back.

Not sure how it got blank. Looks like a bug in typo.


Rick DeNatale

Blog: http://talklikeaduck.denhaven2.com/
Twitter: http://twitter.com/RickDeNatale
WWR: http://www.workingwithrails.com/person/9021-rick-denatale
LinkedIn: http://www.linkedin.com/in/rickdenatale


#8

On Tue, May 26, 2009 at 8:46 AM, Rick DeNatale removed_email_address@domain.invalid
wrote:

Ouch. And you’re absolutely sure there’s no caching appliance

So the blog is back but I still don’t understand why the cached page got empty.

Actually, I think I figured it out.

A day or so ago, I made a stupid sysop mistake and did a bulk copy
which filled up the Logical Volume where the blog is kept. I suspect
that Typo tried to refresh that cached file when the disk was full
hence the empty file.


Rick DeNatale

Blog: http://talklikeaduck.denhaven2.com/
Twitter: http://twitter.com/RickDeNatale
WWR: http://www.workingwithrails.com/person/9021-rick-denatale
LinkedIn: http://www.linkedin.com/in/rickdenatale


#9

On Tue, May 26, 2009 at 8:44 AM, Hassan S.
removed_email_address@domain.invalid wrote:

sending a request results in a no connection message?

Any recent config changes? System auto-updates?

Nope.

I just posted in reply to my original post that I discovered that
there was an empty index.html in the public directory. I got rid of
that and the page came back (Typo regenerated it).

So the blog is back but I still don’t understand why the cached page got
empty.


Rick DeNatale

Blog: http://talklikeaduck.denhaven2.com/
Twitter: http://twitter.com/RickDeNatale
WWR: http://www.workingwithrails.com/person/9021-rick-denatale
LinkedIn: http://www.linkedin.com/in/rickdenatale


#10

Le 26 mai 09 à 14:45, Rick DeNatale a écrit :

Okay, I discovered that there was an empty file in public/index.html,
I renamed it and now the root page is back.

Not sure how it got blank. Looks like a bug in typo.

It’s not a Typo issue. We generate the index.html. We delete it only
if it’s Typo to create it. If you create it manually, Typo can’t
delete it. We can’t assume, all page generate if you don’t generate it.


Cyril M.
http://blog.shingara.fr


#11

On Tue, May 26, 2009 at 5:46 AM, Rick DeNatale
removed_email_address@domain.invalidwrote:

Tried that, no joy.
there was an empty index.html in the public directory. I got rid of
WWR: http://www.workingwithrails.com/person/9021-rick-denatale
LinkedIn: http://www.linkedin.com/in/rickdenatale

If you’re caching the index action using caches_page which produces
an index.html in the public directory, then Typo appears to be
overwriting
that version.

-Conrad