[thelist] Re: traceroutes

[Ron Thigpen]

Better yet, have these outside HTTP requests hit a page that tests your servers
end-to-end.  If this page calls to the database and any other backend systems, a
well-formed response should indicate that all the boxes and their respective
services are healthy.  

Do be sure that the requestor can differentiate between a positive and negative
response.  A healthy webserver backed by a munged database will generally return
a 200, but the content of the page (error text vs. well-formed output) will tell
the story.   

These HTTP requestors are pretty easy to write.  Perhaps we can come up with set
of them in different languages (CF,PHP,ASP,etc) and deploy a community of
symmetric services, an "I'll check yours if you'll check mine" network.  Just a
thought.  

--rt


"Daniel J. Cody" wrote:
> 
> I recommend a third party system(either yours or someone elses) that
> sits on a *different* network than the one you're checking and makes X
> amount of HTTP requests a day(for webservers) This checks for network
> connectivity and for the webserver software itself being up and
> accepting requests.