Mailing List Archive: shambhala bug (v0.5.2)

shambhala bug (v0.5.2)

Jul 6, 1995, 7:49 PM

Post #1 of 4 (455 views)

It seems 5.2 (and earlier I think) did not log "file not found" stuff in
the error log. Here's a patch:

*** http_core.c-orig Thu Jul 6 19:46:33 1995
--- http_core.c Thu Jul 6 19:48:19 1995
***************
*** 638,644 ****
/* Generic case... */

if (r->method_number != M_GET) return DECLINED;
! if (r->finfo.st_mode == 0) return NOT_FOUND;

if ((errstatus = set_content_length (r, r->finfo.st_size))
|| (errstatus = set_last_modified (r, r->finfo.st_mtime)))
--- 638,647 ----
/* Generic case... */

if (r->method_number != M_GET) return DECLINED;
! if (r->finfo.st_mode == 0) {
! log_reason("File does not exist", r->filename, r);
! return NOT_FOUND;
! }

if ((errstatus = set_content_length (r, r->finfo.st_size))
|| (errstatus = set_last_modified (r, r->finfo.st_mtime)))

Re: shambhala bug (v0.5.2) [ In reply to ]

chuck at telebase

Jul 7, 1995, 8:57 AM

Post #2 of 4 (437 views)

Permalink

Re: shambhala bug (v0.5.2) [ In reply to ]

rst at ai

Jul 7, 1995, 9:06 AM

Post #3 of 4 (440 views)

Permalink

Date: Thu, 6 Jul 1995 19:49:39 -0700
From: Cliff Skolnick <cliffs@organic.com>

It seems 5.2 (and earlier I think) did not log "file not found" stuff in
the error log. Here's a patch:

OK --- FITNR. How's it performing otherwise?

rst

Re: shambhala bug (v0.5.2) [ In reply to ]

cliffs at organic

Jul 7, 1995, 11:53 AM

Post #4 of 4 (443 views)

Permalink

So far things seem to be quite excellent. I think Brian and I have
replace all Apache servers we had with Shambhala now, and all is well.
Performance is very good, no problems so far. One annoying thing is
I run the server at home, I get about 200 hits/day there, and I
want the forking behavior back for that :) If I use a small number of
processes (2 or 3), my slow link kills netscape since it tries to make multiple
connections. So I end up having all these servers hanging out idle most of
the time.

The "fork a process if no child was available" and running zero children
would be a nice option :) [.In other words, give me the old behavior
back by option]

On another note, are we going to revive the patch database?

On Fri, 07 Jul 1995 09:06:10 EDT, rst@ai.mit.edu (Robert S. Thau) wrote:

} Date: Thu, 6 Jul 1995 19:49:39 -0700
} From: Cliff Skolnick <cliffs@organic.com>
}
} It seems 5.2 (and earlier I think) did not log "file not found" stuff in
} the error log. Here's a patch:
}
} OK --- FITNR. How's it performing otherwise?
}
} rst
--