1 |
\input texinfo @c -*-texinfo-*- |
2 |
@c %**start of header |
3 |
@setfilename hydra.info |
4 |
@settitle The Hydra HTTP Daemon |
5 |
@set UPDATED Last Updated: 2 Jan 2001 |
6 |
@set COPYPHRASE Copyright @copyright{} 1996-2001 Jon Nelson and Larry Doolittle |
7 |
@set VERSION $Revision: 1.2 $ |
8 |
|
9 |
@paragraphindent asis |
10 |
@iftex |
11 |
@parindent 0pt |
12 |
@end iftex |
13 |
@c @setchapternewpage odd |
14 |
@c %**end of header |
15 |
|
16 |
@iftex |
17 |
@titlepage |
18 |
@title The Hydra HTTP Daemon |
19 |
@c @sp 2 |
20 |
@end iftex |
21 |
|
22 |
@ifinfo |
23 |
This file documents Hydra, an HTTP daemon for UN*X like machines. |
24 |
@end ifinfo |
25 |
|
26 |
@html |
27 |
<h1 align="center">The Hydra HTTP Daemon</h1> |
28 |
@end html |
29 |
|
30 |
@ifinfo |
31 |
@dircategory Networking |
32 |
@direntry |
33 |
* Hydra: (hydra). The Hydra Webserver |
34 |
@end direntry |
35 |
@end ifinfo |
36 |
|
37 |
@comment node-name, next, previous, up |
38 |
@node Top, Introduction, , (dir) |
39 |
|
40 |
Welcome to the documentation for Hydra, a high performance |
41 |
HTTP Server for UN*X-alike computers, covered by the |
42 |
@uref{Gnu_License,GNU General Public License}. This is a draft |
43 |
document based on Boa documentation. |
44 |
The on-line, updated copy of this documentation lives at |
45 |
@uref{http://hydra.hellug.gr/,http://hydra.hellug.gr/} |
46 |
@sp 1 |
47 |
@center @value{COPYPHRASE} |
48 |
@center @value{UPDATED}, @value{VERSION} |
49 |
|
50 |
@iftex |
51 |
@end titlepage |
52 |
@contents |
53 |
@end iftex |
54 |
|
55 |
@menu |
56 |
* Introduction:: |
57 |
* Installation and Usage:: |
58 |
* Limits and Design Philosophy:: |
59 |
* Appendix:: |
60 |
|
61 |
-- Detailed Node Listing -- |
62 |
|
63 |
Installation |
64 |
|
65 |
* Files Used by Hydra:: |
66 |
* Compile-Time and Command-Line Options:: |
67 |
* hydra.conf Directives:: |
68 |
* Security:: |
69 |
|
70 |
Limits and Design Philosophy |
71 |
|
72 |
* Limits:: |
73 |
* Differences between Hydra and other web servers:: |
74 |
* Unexpected Behavior:: |
75 |
|
76 |
Appendix |
77 |
|
78 |
* License:: |
79 |
* Acknowledgments:: |
80 |
* Reference Documents:: |
81 |
* Other HTTP Servers:: |
82 |
* Benchmarks:: |
83 |
* Tools:: |
84 |
* Authors:: |
85 |
|
86 |
@end menu |
87 |
|
88 |
@comment node-name, next, previous, up |
89 |
@node Introduction, Installation and Usage,top,top |
90 |
@chapter Introduction |
91 |
|
92 |
Hydra is a single-tasking HTTP server. That means that unlike |
93 |
traditional web servers, it does not fork for each incoming |
94 |
connection, nor does it fork many copies of itself to handle multiple |
95 |
connections. It internally multiplexes all of the ongoing HTTP |
96 |
connections, and forks only for CGI programs (which must be separate |
97 |
processes), automatic directory generation, and automatic file |
98 |
gunzipping. Preliminary tests show Hydra is capable of |
99 |
handling several thousand hits per second on a 300 MHz Pentium and |
100 |
dozens of hits per second on a lowly 20 MHz 386/SX. |
101 |
|
102 |
The primary design goals of Hydra are speed and security. Security, |
103 |
in the sense of @emph{can't be subverted by a malicious user,} not |
104 |
@emph{fine grained access control and encrypted communications}. |
105 |
Hydra is not intended as a feature-packed server; if you want one of those, |
106 |
check out |
107 |
WN (@uref{http://hopf.math.nwu.edu/}) from John Franks. |
108 |
Modifications to Hydra that improve its speed, security, robustness, and |
109 |
portability, are eagerly sought. Other features may be added if they |
110 |
can be achieved without hurting the primary goals. |
111 |
|
112 |
Hydra is based on Boa web server. |
113 |
Boa was created in 1991 by Paul Phillips (@email{psp@@well.com}). |
114 |
It is now being maintained and enhanced by Larry Doolittle |
115 |
(@email{ldoolitt@@boa.org}) and Jon Nelson |
116 |
(@email{jnelson@@boa.org}). |
117 |
Please see the acknowledgement section for further |
118 |
details. Hydra is maintained by |
119 |
Nikos Mavroyanopoulos (@email{nmav@@gnutls.org}). |
120 |
|
121 |
|
122 |
GNU/Linux is the development platform at the moment, other OS's are known to work. |
123 |
If you'd like to contribute to this effort, contact Larry or Jon via e-mail. |
124 |
|
125 |
@comment node-name, next, previous, up |
126 |
@node Installation and Usage, Limits and Design Philosophy, Introduction,top |
127 |
@chapter Installation and Usage |
128 |
|
129 |
Hydra is currently being developed and tested on GNU/Linux/i386. |
130 |
The code is straightforward (more so than most other servers), |
131 |
so it should run easily on most modern Unix-alike platforms. Recent |
132 |
versions of Hydra worked fine on FreeBSD, SunOS 4.1.4, GNU/Linux-SPARC, |
133 |
and HP-UX 9.0. Pre-1.2.0 GNU/Linux kernels may not work because of |
134 |
deficient mmap() implementations. |
135 |
|
136 |
@menu |
137 |
* Installation:: |
138 |
* Files Used by Hydra:: |
139 |
* Compile-Time and Command-Line Options:: |
140 |
* Security:: |
141 |
@end menu |
142 |
|
143 |
@comment node-name, next, previous, up |
144 |
@node Installation,Files Used by Hydra,,Installation and Usage |
145 |
@section Installation |
146 |
|
147 |
@enumerate |
148 |
@item Unpack |
149 |
@enumerate |
150 |
@item Choose, and cd into, a convenient directory for the package. |
151 |
@item @kbd{tar -xvzf hydra-x.y.z.tar.gz}, or for those of you with an archaic |
152 |
(non-GNU) tar; @kbd{gzip -cd < hydra-x.y.z.tar.gz | tar -xvf -} |
153 |
@item Read the documentation. Really. |
154 |
@end enumerate |
155 |
@item Build |
156 |
@enumerate |
157 |
@item cd into the @t{src} directory. |
158 |
@item (optional) Change the default SERVER_ROOT by setting the #define |
159 |
at the top of src/defines.h |
160 |
@item Type @kbd{./configure; make} |
161 |
@item Report any errors to the maintainers for resolution, or strike |
162 |
out on your own. |
163 |
@end enumerate |
164 |
@item Configure |
165 |
@enumerate |
166 |
@item Choose a user and server port under which Hydra can run. The |
167 |
traditional port is 80, and user @t{nobody} (create if |
168 |
you need to) is often a good selection for security purposes. |
169 |
If you don't have (or choose not to use) root privileges, you |
170 |
can not use port numbers less than 1024, nor can you switch user id. |
171 |
@item Choose a server root. The @t{conf} directory within the |
172 |
server root must hold your copy of the configuration file |
173 |
@emph{hydra.conf} |
174 |
@item Choose locations for log files, CGI programs (if any), and |
175 |
the base of your URL tree. |
176 |
@item Set the location of the @t{mime.types} file. |
177 |
@item Edit @emph{conf/hydra.conf} according to your |
178 |
choices above (this file documents itself). Read through this file |
179 |
to see what other features you can configure. |
180 |
@end enumerate |
181 |
@item Start |
182 |
@itemize |
183 |
@item Start Hydra. If you didn't build the right SERVER_ROOT into the |
184 |
binary, you can specify it on the command line with the -c option |
185 |
(command line takes precedence). |
186 |
@example |
187 |
Example: ./hydra -c /usr/local/hydra |
188 |
@end example |
189 |
@end itemize |
190 |
|
191 |
@item Test |
192 |
@itemize |
193 |
@item At this point the server should run and serve documents. |
194 |
If not, check the error_log file for clues. |
195 |
@end itemize |
196 |
|
197 |
@item Install |
198 |
@itemize |
199 |
@item Copy the binary to a safe place, and put the invocation into |
200 |
your system startup scripts. Use the same -c option you used |
201 |
in your initial tests. |
202 |
@end itemize |
203 |
@end enumerate |
204 |
|
205 |
@comment node-name, next, previous, up |
206 |
@node Files Used by Hydra, Compile-Time and Command-Line Options, Installation,Installation and Usage |
207 |
@section Files Used by Hydra |
208 |
|
209 |
@ftable @file |
210 |
@item hydra.conf |
211 |
This file is the sole configuration file for Hydra. The directives in this |
212 |
file are defined in the DIRECTIVES section. |
213 |
@item mime.types |
214 |
The MimeTypes <filename> defines what Content-Type Hydra will |
215 |
send in an HTTP/1.0 or better transaction. |
216 |
Set to /dev/null if you do not want to load a mime types file. |
217 |
Do *not* comment out (better use AddType!) |
218 |
@end ftable |
219 |
|
220 |
@comment node-name, next, previous, up |
221 |
@node Compile-Time and Command-Line Options, hydra.conf Directives, Files Used by Hydra,Installation and Usage |
222 |
@section Compile-Time and Command-Line Options |
223 |
|
224 |
@table @var |
225 |
@item SERVER_ROOT |
226 |
@itemx -c |
227 |
The default server root as #defined by @var{SERVER_ROOT} in |
228 |
@file{defines.h} can be overridden on the commandline using the |
229 |
@option{-c} option. The server root must hold your local copy of the |
230 |
configuration file @file{hydra.conf}. |
231 |
@example |
232 |
Example: /usr/sbin/hydra -c /etc/hydra |
233 |
@end example |
234 |
|
235 |
@end table |
236 |
|
237 |
@comment node-name, next, previous, up |
238 |
@node hydra.conf Directives, Security, Compile-Time and Command-Line Options, (top) |
239 |
@section hydra.conf Directives |
240 |
|
241 |
The Hydra configuration file is parsed with a lex/yacc or flex/bison |
242 |
generated parser. If it reports an error, the line number will be |
243 |
provided; it should be easy to spot. The syntax of each of these rules |
244 |
is very simple, and they can occur in any order. Where possible, these |
245 |
directives mimic those of NCSA httpd 1.3; I (Paul Phillips) saw no reason |
246 |
to introduce gratuitous differences. |
247 |
|
248 |
Note: the "ServerRoot" is not in this configuration file. It can be |
249 |
compiled into the server (see @file{defines.h}) or specified on the command |
250 |
line with the @command{-c} option. |
251 |
|
252 |
The following directives are contained in the @file{hydra.conf} file, and most, |
253 |
but not all, are required. |
254 |
|
255 |
@table @option |
256 |
@item Port <Integer> |
257 |
This is the port that Hydra runs on. The default port for http servers is 80. |
258 |
If it is less than 1024, the server must be started as root. |
259 |
|
260 |
@item Listen <IP> |
261 |
The Internet address to bind(2) to, in quadded-octet form (numbers). |
262 |
If you leave it out, it binds to all addresses (INADDR_ANY). |
263 |
|
264 |
The name you provide gets run through inet_aton(3), so you have to |
265 |
use dotted quad notation. This configuration is too important to trust some DNS. |
266 |
|
267 |
You only get one "Listen" directive, if you want service on multiple |
268 |
IP addresses, you have three choices: |
269 |
|
270 |
@enumerate |
271 |
@item Run hydra without a "Listen" directive: |
272 |
@itemize @bullet |
273 |
@item All addresses are treated the same; makes sense if the addresses |
274 |
are localhost, ppp, and eth0. |
275 |
@item Use the VirtualHost directive below to point requests to different files. |
276 |
Should be good for a very large number of addresses (web hosting clients). |
277 |
@end itemize |
278 |
@item Run one copy of hydra per IP address: |
279 |
@itemize @bullet |
280 |
@item Each instance has its own configuration with its own |
281 |
"Listen" directive. No big deal up to a few tens of addresses. Nice separation |
282 |
between clients. |
283 |
@end itemize |
284 |
@end enumerate |
285 |
|
286 |
@item User <username or UID> |
287 |
The name or UID the server should run as. For Hydra to attempt this, the |
288 |
server must be started as root. |
289 |
|
290 |
@item Group <groupname or GID> |
291 |
The group name or GID the server should run as. For Hydra to attempt this, |
292 |
the server must be started as root. |
293 |
|
294 |
@item ServerAdmin <email address> |
295 |
The email address where server problems should be sent. Note: this is not |
296 |
currently used. |
297 |
|
298 |
@item ErrorLog <filename> |
299 |
The location of the error log file. If this does not start with /, it is |
300 |
considered relative to the server root. Set to /dev/null if you don't want |
301 |
errors logged. |
302 |
|
303 |
@item AccessLog <filename> |
304 |
The location of the access log file. If this does not start with /, it is |
305 |
considered relative to the server root. Comment out or set to /dev/null |
306 |
(less effective) to disable access logging. |
307 |
|
308 |
@item VerboseCGILogs |
309 |
This is a logical switch and does not take any parameters. Comment out to |
310 |
disable. All it does is switch on or off logging of when CGIs are launched and when |
311 |
the children return. |
312 |
|
313 |
@item CgiLog <filename> |
314 |
The location of the CGI error log file. If |
315 |
specified, this is the file that the stderr of CGIs is tied to. Otherwise, writes |
316 |
to stderr meet the bit bucket. |
317 |
|
318 |
@item ServerName <server_name> |
319 |
The name of this server that should be sent back to clients if different |
320 |
than that returned by gethostname. |
321 |
|
322 |
@item VirtualHost |
323 |
This is a logical switch and does not take any parameters. |
324 |
Comment out to disable. Given DocumentRoot /var/www, requests on interface `A' or |
325 |
IP `IP-A' become /var/www/IP-A. Example: http://localhost/ becomes |
326 |
/var/www/127.0.0.1 |
327 |
|
328 |
@item DocumentRoot <directory> |
329 |
The root directory of the HTML documents. If this does not start with /, |
330 |
it is considered relative to the server root. |
331 |
|
332 |
@item UserDir <directory> |
333 |
The name of the directory which is appended onto a user's home directory |
334 |
if a ~user request is received. |
335 |
|
336 |
@item DirectoryIndex <filename> |
337 |
Name of the file to use as a pre-written HTML directory index. Please |
338 |
make and use these files. On the fly creation of directory indexes |
339 |
can be slow. |
340 |
|
341 |
@item DirectoryMaker <full pathname to program> |
342 |
Name of the program used |
343 |
to generate on-the-fly directory listings. The program must take one or two |
344 |
command-line arguments, the first being the directory to index (absolute), and the |
345 |
second, which is optional, should be the "title" of the document be. Comment out if |
346 |
you don't want on the fly directory listings. If this does not start with /, it is |
347 |
considered relative to the server root. |
348 |
|
349 |
@item DirectoryCache <directory> |
350 |
DirectoryCache: If DirectoryIndex doesn't exist, and DirectoryMaker has been |
351 |
commented out, the the on-the-fly indexing of Hydra can be used to generate indexes |
352 |
of directories. Be warned that the output is extremely minimal and can cause |
353 |
delays when slow disks are used. Note: The DirectoryCache must be writable by the |
354 |
same user/group that Hydra runs as. |
355 |
|
356 |
@item KeepAliveMax <integer> |
357 |
Number of KeepAlive requests to allow per connection. Comment out, or set |
358 |
to 0 to disable keepalive processing. |
359 |
|
360 |
@item KeepAliveTimeout <integer> |
361 |
Number of seconds to wait before keepalive connections time out. |
362 |
|
363 |
@item MimeTypes <file> |
364 |
The location of the mime.types file. If this does not start with /, it is |
365 |
considered relative to the server root. |
366 |
Comment out to avoid loading mime.types (better use AddType!) |
367 |
|
368 |
@item DefaultType <mime type> |
369 |
MIME type used if the file extension is unknown, or there is no file |
370 |
extension. |
371 |
|
372 |
@item AddType <mime type> <extension> extension... |
373 |
Associates a MIME type |
374 |
with an extension or extensions. |
375 |
|
376 |
@item Redirect, Alias, and ScriptAlias |
377 |
Redirect, Alias, and ScriptAlias all have the same semantics -- |
378 |
they match the beginning of a request and take appropriate action. |
379 |
Use Redirect for other servers, Alias for the same server, and |
380 |
ScriptAlias to enable directories for script execution. |
381 |
|
382 |
@item Redirect <path1> <path2> |
383 |
allows you to tell clients about documents which used to exist |
384 |
in your server's namespace, but do not anymore. This allows you |
385 |
tell the clients where to look for the relocated document. |
386 |
|
387 |
@item Alias <path1> <path2> |
388 |
aliases one path to another. Of course, symbolic links in the |
389 |
file system work fine too. |
390 |
|
391 |
@item ScriptAlias <path1> <path2> |
392 |
maps a virtual path to a directory for serving scripts. |
393 |
|
394 |
@item Allow, Deny |
395 |
Only supported if Hydra is compiled with --enable-access-control. |
396 |
Allow and Deny allows pattern based access control using shell |
397 |
wildcards. The string the matching is performed on is the absolute |
398 |
filesystem filename. The Allow, Deny directives are processed in |
399 |
order until the first match is found, so to allow files containing |
400 |
the substring 123 but not the ones containing 1234 you would do: |
401 |
@itemize @bullet |
402 |
@item Deny *1234* |
403 |
@item Allow *123* |
404 |
@end itemize |
405 |
|
406 |
@item Allow <pattern> |
407 |
Allows files matching <pattern> |
408 |
|
409 |
@item Deny <pattern> |
410 |
Disallowes files matching <pattern> |
411 |
|
412 |
@end table |
413 |
|
414 |
@comment node-name, next, previous, up |
415 |
@node Security, , hydra.conf Directives, Installation and Usage |
416 |
@section Security |
417 |
|
418 |
Hydra has been designed to use the existing file system security. In |
419 |
@file{hydra.conf}, the directives @emph{user} and |
420 |
@emph{group} determine who Hydra will run as, if launched by root. |
421 |
By default, the user/group is nobody/nogroup. This allows quite a bit |
422 |
of flexibility. For example, if you want to disallow access to otherwise |
423 |
accessible directories or files, simply make them inaccessible to |
424 |
nobody/nogroup. If the user that Hydra runs as is "hydra" and the groups that |
425 |
"hydra" belongs to include "web-stuff" then files/directories accessible |
426 |
by users with group "web-stuff" will also be accessible to Hydra. |
427 |
|
428 |
The February 2000 hoo-rah from |
429 |
@uref{http://www.cert.org/advisories/CA-2000-02.html,CERT advisory CA-2000-02} |
430 |
has little to do with Hydra. As of version 0.94.4, Hydra's escaping rules have |
431 |
been cleaned up a little, but they weren't that bad before. The example CGI |
432 |
programs have been updated to show what effort is needed there. If you |
433 |
write, maintain, or use CGI programs under Hydra (or any other server) it's |
434 |
worth your while to read and understand this advisory. The real problem, |
435 |
however, boils down to browser and web page designers emphasizing frills |
436 |
over content and security. The market leading browsers assume (incorrectly) |
437 |
that all web pages are trustworthy. |
438 |
|
439 |
@comment node-name, next, previous, up |
440 |
@node Limits and Design Philosophy,Appendix, Installation and Usage,top |
441 |
@chapter Limits and Design Philosophy |
442 |
|
443 |
There are many issues that become more difficult to resolve in a single |
444 |
tasking web server than in the normal forking model. Here is a partial |
445 |
list -- there are probably others that haven't been encountered yet. |
446 |
|
447 |
@menu |
448 |
* Limits:: |
449 |
* Differences between Hydra and other web servers:: |
450 |
* Unexpected Behavior:: |
451 |
@end menu |
452 |
|
453 |
@comment node-name, next, previous, up |
454 |
@node Limits,Differences between Hydra and other web servers,,Limits and Design Philosophy |
455 |
@section Limits |
456 |
|
457 |
@itemize @bullet |
458 |
@item Slow file systems |
459 |
|
460 |
The file systems being served should be much faster than the |
461 |
network connection to the HTTP requests, or performance will suffer. |
462 |
For instance, if a document is served from a CD-ROM, the whole server |
463 |
(including all other currently incomplete data transfers) will stall |
464 |
while the CD-ROM spins up. This is a consequence of the fact that Hydra |
465 |
mmap()'s each file being served, and lets the kernel read and cache |
466 |
pages as best it knows how. When the files come from a local disk |
467 |
(the faster the better), this is no problem, and in fact delivers |
468 |
nearly ideal performance under heavy load. Avoid serving documents |
469 |
from NFS and CD-ROM unless you have even slower inbound net |
470 |
connections (e.g., POTS SLIP). |
471 |
|
472 |
@item DNS lookups |
473 |
|
474 |
Writing a nonblocking gethostbyaddr is a difficult and not very |
475 |
enjoyable task. Paul Phillips experimented with several methods, |
476 |
including a separate logging process, before removing hostname |
477 |
lookups entirely. There is a companion program with Hydra |
478 |
@file{util/resolver.pl} that will postprocess the logfiles and |
479 |
replace IP addresses with hostnames, which is much faster no matter |
480 |
what sort of server you run. |
481 |
|
482 |
@item Identd lookups |
483 |
|
484 |
Same difficulties as hostname lookups; not included. |
485 |
Hydra provides a REMOTE_PORT environment variable, in addition |
486 |
to REMOTE_ADDR, so that a CGI program can do its own ident. |
487 |
See the end of @t{examples/cgi-test.cgi}. |
488 |
|
489 |
@item Password file lookups via NIS |
490 |
|
491 |
If users are allowed to serve HTML from their home directories, |
492 |
password file lookups can potentially block the process. To lessen |
493 |
the impact, each user's home directory is cached by Hydra so it need |
494 |
only be looked up once. |
495 |
|
496 |
@item Running out of file descriptors |
497 |
|
498 |
Since a file descriptor is needed for every ongoing connection |
499 |
(two for non-nph CGIs, directories, and automatic gunzipping of files), |
500 |
it is possible though highly improbable to run out of file |
501 |
descriptors. The symptoms of this conditions may vary with |
502 |
your particular unix variant, but you will probably see log |
503 |
entries giving an error message for @t{accept}. |
504 |
Try to build your kernel to give an adequate number for |
505 |
your usage - GNU/Linux provides 256 out of the box, more than |
506 |
enough for most people. |
507 |
@end itemize |
508 |
|
509 |
@comment node-name, next, previous, up |
510 |
@node Differences between Hydra and other web servers,Unexpected Behavior,Limits,Limits and Design Philosophy |
511 |
@section Differences between Hydra and other web servers |
512 |
|
513 |
In the pursuit of speed and simplicity, some aspects of Hydra differ |
514 |
from the popular web servers. In no particular order: |
515 |
|
516 |
@itemize @bullet |
517 |
@item @var{REMOTE_HOST} environment variable not set for CGI programs |
518 |
|
519 |
The @var{REMOTE_HOST} environment variable is not set for CGI programs, |
520 |
for reasons already described. This is easily worked around because the |
521 |
IP address is provided in the @var{REMOTE_HOST} variable, so (if the CGI |
522 |
program actually cares) gethostbyaddr or a variant can be used. |
523 |
|
524 |
@item There are no server side includes (@acronym{SSI}) in Hydra |
525 |
|
526 |
We don't like them, and they are too slow to parse. We will consider |
527 |
more efficient alternatives. |
528 |
|
529 |
@item There are no access control features |
530 |
|
531 |
Hydra will follow symbolic links, and serve any file that it can |
532 |
read. The expectation is that you will configure Hydra to run as user |
533 |
"nobody", and only files configured world readable will come |
534 |
out. |
535 |
|
536 |
@item No chroot option |
537 |
|
538 |
There is no option to run chrooted. If anybody wants this, and is |
539 |
willing to try out experimental code, contact the maintainers. |
540 |
@end itemize |
541 |
|
542 |
@comment node-name, next, previous, up |
543 |
@node Unexpected Behavior,,Differences between Hydra and other web servers,Limits and Design Philosophy |
544 |
@section Unexpected Behavior |
545 |
|
546 |
@itemize @bullet |
547 |
@item SIGHUP handling |
548 |
|
549 |
Like any good server, Hydra traps SIGHUP and rereads @file{hydra.conf}. |
550 |
However, under normal circumstances, it has already given away |
551 |
permissions, so many items listed in @file{hydra.conf} can not take effect. |
552 |
No attempt is made to change uid, gid, log files, or server port. |
553 |
All other configuration changes should take place smoothly. |
554 |
|
555 |
@item SIGUSR1 handling |
556 |
|
557 |
Hydra traps SIGUSR1 and prints useful statistics in the error log file. |
558 |
|
559 |
@item Relative URL handling |
560 |
|
561 |
Not all browsers handle relative URLs correctly. Hydra will not |
562 |
cover up for this browser bug, and will typically report 404 Not Found |
563 |
for URL's containing odd combinations of "../" 's. |
564 |
|
565 |
Note: As of version 0.95.0 (unreleased) the URL parser has been |
566 |
rewritten and *does* correctly handle relative URLs. |
567 |
@end itemize |
568 |
|
569 |
@comment node-name, next, previous, up |
570 |
@node Appendix,,Limits and Design Philosophy,top |
571 |
@appendix Appendix |
572 |
|
573 |
@menu |
574 |
* License:: |
575 |
* Acknowledgments:: |
576 |
* Reference Documents:: |
577 |
* Other HTTP Servers:: |
578 |
* Benchmarks:: |
579 |
* Tools:: |
580 |
* Authors:: |
581 |
@end menu |
582 |
|
583 |
@comment node-name, next, previous, up |
584 |
@node License,Acknowledgments,,Appendix |
585 |
@section License |
586 |
|
587 |
This program is distributed under the |
588 |
@uref{http://www.gnu.org/copyleft/gpl.html,GNU General Public License}. |
589 |
as noted in each source file: |
590 |
@* |
591 |
|
592 |
@smallexample |
593 |
/* |
594 |
* Hydra, an http server |
595 |
* Copyright (C) 1995 Paul Phillips <psp@@well.com> |
596 |
* |
597 |
* This program is free software; you can redistribute it and/or modify |
598 |
* it under the terms of the GNU General Public License as published by |
599 |
* the Free Software Foundation; either version 1, or (at your option) |
600 |
* any later version. |
601 |
* |
602 |
* This program is distributed in the hope that it will be useful, |
603 |
* but WITHOUT ANY WARRANTY; without even the implied warranty of |
604 |
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the |
605 |
* GNU General Public License for more details. |
606 |
* |
607 |
* You should have received a copy of the GNU General Public License |
608 |
* along with this program; if not, write to the Free Software |
609 |
* Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA. |
610 |
* |
611 |
*/ |
612 |
|
613 |
@end smallexample |
614 |
|
615 |
@comment node-name, next, previous, up |
616 |
@node Acknowledgments,Reference Documents,License,Appendix |
617 |
@section Acknowledgments |
618 |
|
619 |
Paul Phillips wrote the first versions of Hydra, up to and including |
620 |
version 0.91. Version 0.92 of Hydra was officially released December 1996 |
621 |
by Larry Doolittle. Version 0.93 was the development version of 0.94, |
622 |
which was released in February 2000. |
623 |
|
624 |
The Boa Webserver is currently (Feb 2000) maintained and enhanced by |
625 |
Larry Doolittle (@email{ldoolitt@@boa.org}) |
626 |
and Jon Nelson (@email{jnelson@@boa.org}). |
627 |
|
628 |
We would like to thank Russ Nelson (@email{nelson@@crynwr.com}) |
629 |
for hosting the @uref{http://www.boa.org,web site}. |
630 |
|
631 |
We would also like to thank Paul Philips for writing code that is |
632 |
worth maintaining and supporting. |
633 |
|
634 |
Many people have contributed to Hydra, including (but not |
635 |
limited to) Charles F. Randall (@email{randall@@goldsys.com}) |
636 |
Christoph Lameter (@email{<chris@@waterf.org>}), |
637 |
Russ Nelson (@email{<nelson@@crynwr.com>}), Alain Magloire |
638 |
(@email{<alain.magloire@@rcsm.ee.mcgill.ca>}), |
639 |
and more recently, M. Drew Streib (@email{<dtype@@linux.com>}). |
640 |
|
641 |
Paul Phillips records his acknowledgments as follows: |
642 |
@quotation |
643 |
Thanks to everyone in the WWW community, in general a great bunch of people. |
644 |
Special thanks to Clem Taylor (@email{<ctaylor@@eecis.udel.edu>}), who |
645 |
provided invaluable feedback on many of my ideas, and offered good |
646 |
ones of his own. Also thanks to John Franks, author of wn, for |
647 |
writing what I believe is the best webserver out there. |
648 |
@end quotation |
649 |
|
650 |
@comment node-name, next, previous, up |
651 |
@node Reference Documents,Other HTTP Servers,Acknowledgments,Appendix |
652 |
@section Reference Documents |
653 |
|
654 |
Links to documents relevant to |
655 |
@uref{http://hydra.hellug.gr/,Hydra} |
656 |
development and usage. Incomplete, we're still working on this. |
657 |
NCSA has a decent |
658 |
@uref{http://hoohoo.ncsa.uiuc.edu/docs/Library.html,page} along |
659 |
these lines, too. |
660 |
|
661 |
Also see Yahoo's List |
662 |
@* @uref{http://www.yahoo.com/Computers_and_Internet/Software/Internet/World_Wide_Web/Servers/} |
663 |
|
664 |
@itemize |
665 |
@item W3O HTTP page |
666 |
@* @uref{http://www.w3.org/pub/WWW/Protocols/} |
667 |
|
668 |
@item RFC 1945 HTTP-1.0 (informational) |
669 |
@* @uref{http://ds.internic.net/rfc/rfc1945.txt} |
670 |
|
671 |
@item IETF Working Group Draft 07 of HTTP-1.1 |
672 |
@* @uref{http://www.w3.org/pub/WWW/Protocols/HTTP/1.1/draft-ietf-http-v11-spec-07.txt} |
673 |
|
674 |
@item HTTP: A protocol for networked information |
675 |
@* @uref{http://www.w3.org/pub/WWW/Protocols/HTTP/HTTP2.html} |
676 |
|
677 |
@item The Common Gateway Interface (CGI) |
678 |
@* @uref{http://hoohoo.ncsa.uiuc.edu/cgi/overview.html} |
679 |
|
680 |
@item RFC 1738 URL syntax and semantics |
681 |
@* @uref{http://ds.internic.net/rfc/rfc1738.txt} |
682 |
|
683 |
@item RFC 1808 Relative URL syntax and semantics |
684 |
@* @uref{http://ds.internic.net/rfc/rfc1808.txt} |
685 |
@end itemize |
686 |
|
687 |
@comment node-name, next, previous, up |
688 |
@node Other HTTP Servers,Benchmarks,Reference Documents,Appendix |
689 |
@section Other HTTP Servers |
690 |
|
691 |
For unix-alike platforms, with published source code. |
692 |
|
693 |
@itemize |
694 |
@item tiny/turbo/throttling httpd very similar to Hydra, with a throttling |
695 |
feature |
696 |
@* @uref{http://www.acme.com/software/thttpd/} |
697 |
|
698 |
@item Roxen: based on ulpc interpreter, non-forking (interpreter implements |
699 |
threading), GPL'd |
700 |
@* @uref{http://www.roxen.com/} |
701 |
|
702 |
@item WN: featureful, GPL'd |
703 |
@* @uref{http://hopf.math.nwu.edu/} |
704 |
|
705 |
@item Apache: fast, PD |
706 |
@* @uref{http://www.apache.org/} |
707 |
|
708 |
@item NCSA: standard, legal status? |
709 |
@* @uref{http://hoohoo.ncsa.uiuc.edu/} |
710 |
|
711 |
@item CERN: standard, PD, supports proxy |
712 |
@* @uref{http://www.w3.org/pub/WWW/Daemon/Status.html} |
713 |
|
714 |
@item xs-httpd 2.0: small, fast, pseudo-GPL'd |
715 |
@* @uref{http://www.stack.nl/~sven/xs-httpd/} |
716 |
|
717 |
@item bozohttpd.tar.gz sources, in perl |
718 |
@* @uref{ftp://ftp.eterna.com.au/bozo/bsf/attware/bozohttpd.tar.gz} |
719 |
|
720 |
@item Squid is actually an "Internet Object Cache" |
721 |
@* @uref{http://squid.nlanr.net/Squid/} |
722 |
@end itemize |
723 |
|
724 |
Also worth mentioning is Zeus. |
725 |
It is commercial, with a free demo, so it doesn't belong on the list above. |
726 |
Zeus seems to be based on technology similar to Hydra and thttpd, |
727 |
but with more bells and whistles. |
728 |
@* @uref{http://www.zeus.co.uk/products/server/} |
729 |
|
730 |
@comment node-name, next, previous, up |
731 |
@node Benchmarks,Tools,Other HTTP Servers,Appendix |
732 |
@section Benchmarks |
733 |
|
734 |
@itemize |
735 |
@item ZeusBench (broken link) |
736 |
@* @uref{http://www.zeus.co.uk/products/server/intro/bench2/zeusbench.shtml} |
737 |
|
738 |
@item WebBench (binary-ware) |
739 |
@* @uref{http://web1.zdnet.com/zdbop/webbench/webbench.html} |
740 |
|
741 |
@item WebStone |
742 |
@* @uref{http://www.mindcraft.com/benchmarks/webstone/} |
743 |
|
744 |
@item SpecWeb96 |
745 |
@* @uref{http://www.specbench.org/osg/web96/} |
746 |
@end itemize |
747 |
|
748 |
@comment node-name, next, previous, up |
749 |
@node Tools,Authors,Benchmarks,Appendix |
750 |
@section Tools |
751 |
|
752 |
@itemize |
753 |
@item Analog logfile analyzer |
754 |
@* @uref{http://www.statslab.cam.ac.uk/@~sret1/analog/} |
755 |
|
756 |
@item wwwstat logfile analyzer |
757 |
@* @uref{http://www.ics.uci.edu/pub/websoft/wwwstat/} |
758 |
|
759 |
@item gwstat wwwstat postprocessor |
760 |
@* @uref{http://dis.cs.umass.edu/stats/gwstat.html} |
761 |
|
762 |
@item The Webalizer logfile analyzer |
763 |
@* @uref{http://www.usagl.net/webalizer/} |
764 |
|
765 |
@item cgiwrap |
766 |
@* @uref{http://www.umr.edu/@~cgiwrap/} |
767 |
|
768 |
@item suEXEC (Hydra would need to be ..umm.. "adjusted" to support this) |
769 |
@* @uref{http://www.apache.org/docs/suexec.html} |
770 |
@end itemize |
771 |
|
772 |
Note: References last checked: 06 October 1997 |
773 |
|
774 |
@comment node-name, next, previous, up |
775 |
@node Authors,,Tools,Appendix |
776 |
@section Authors |
777 |
|
778 |
@itemize |
779 |
@item Conversion from linuxdoc SGML to texinfo by Jon Nelson |
780 |
@item Conversion to linuxdoc SGML by Jon Nelson |
781 |
@item Original HTML documentation by Larry Doolittle |
782 |
@item @value{COPYPHRASE} |
783 |
@end itemize |
784 |
|
785 |
@c variable |
786 |
@c @printindex vr |
787 |
@c concept |
788 |
@c @printindex cp |
789 |
@c function |
790 |
@c @printindex fn |
791 |
@c key |
792 |
@c @printindex ky |
793 |
@c program |
794 |
@c @printindex pg |
795 |
@c data type |
796 |
@c @printindex tp |
797 |
|
798 |
@bye |