Apache::Site [WAS] Re: Segmentation fault

[BeerBong <al...@samara.net>]

> > We already did this! The question is - with 2 apache config,
> > proxy-front-end apache logs every request, even if power apache
daemon
> > crushed with segfault, simple-proxy daemon logs request with URI
and
> > 502 status. We know time of segfault and can find 502 request in
> > access logs with such time, and we can find out a request, wich
causes
> > seg fault. All right. It's working for 2 apache config. If we use
> > simple 1 apache daemon, I think, we will have in logs segfault
message
> > only. Is there way to find out segfault request for simple
> > configuration ?
>
> Provide a custom loghandler, which runs before the content
generation phase
> or whichever phase the segfault happens at. Though you won't know
the
> status,
> but there is nothing you can do after segfault. Or I can think of
two steps
> handler, which let's say logs something before the potentially
segfaulting
> handler and the status after the handler is run. So in case of
segfaults
> the status will be missing.

(First of all, sorry for gruesome english)

Thank you for response. As I understand, with simple configuration we
should assign
global URI translation (for instance) handler where we will log
request's URI.
After segfault we can find (via awk or other text processor) logged
URI without status.
Right?
Huh... I think installation of mod_accel or mod_proxy front-end is
simpler solution.

Thank you again and I will be happy to hear some feedback from
mod_perl guru
about module, which was developed by my team. It's another framework
and Matt Sergeant
compared it with Mason although it is much simpler and we purposed
other aims.

Our target was to ease implementing web application as Apache handler,
because,
we consider web site not a collection of html generating page-scripts,
(Apache::Registry, ASP and many other technix way), but whole
application with one entry point - Apache handler. We think that URI
processing and translating is a task of application, at any case for
projects such large e-shop or central online information portal of
region.
Another aim was to integrate TemplateToolkit usage as standart output
technique - it is excellent app, although descendant from our module
can return another status (redirect and others) or generate output via
another way freely.
Our module helps to maintain high performance web application as pure
Apache handler is fast and easy and programmer focuses on business
tasks of site only.

I put part of our programmer's letter below and ask - is there future
for such module?

| We develope the Module that will be capable of:
|   * use simple interface for dynamic site creating (and dynamic
parts for
|     any almost static site);
|   * use simple urls with no query strings at all;
|   * use smart templates (standard header and footer, for example).
|
| That very early beta works on:
|   * http://www.neftechimia.ru/en/  -- business site
|   * http://review.samara.ru/       -- newspaper site
|
| Note that the module is filter-aware and on "Neftechimia" there is
AxKit
| postprocessing
| (that slowers down overall request performance six(!!!) times (I
think
this
| may be changed
| with new AxKit's Provider) but simplifies site supporting for
| non-professional webmaster
| -- all templates may be changed through the web-interface)
|
| III. THE FUTURE
|
| Though we need some advices:
|   * is there any need for such thing?
|   * if so, which name we should choose (it's Apache::Site for now)?
|   * what features must it also have?
|   * what do you think about it?
|
| IV. THE CODE
|
| As I said above it is very early beta. No POD, Comments only.
|   * http://www.webzavod.ru/software/Apache-Site.tar.gz
|
| V. THE AUTHOR
|
| Gregory "Grishace" [gree-shah-s] Belenky
| WebZavod (http://www.webzavod.ru/) programmer
|
--------------------------------------------
Sergey Polyakov               aka "BeerBong"
Chief of WebZavod     http://www.webzavod.ru
Tel. +7 (8462) 43-93-85 | +7 (8462) 43-93-86
mailto:alexei@samara.net