1 Changes with mod_fastcgi 2.2.5
3 *) Fix Win32 named pipe handling (problems with large responses).
5 *) (Win32) Remove the "can exec" check.
7 Changes with mod_fastcgi 2.2.4
9 *) Beta WinNT support. David Allen [djallen@raleigh.ibm.com] and
10 Rob Saccoccio [robs@chelsea.net]
12 *) Remove request type restriction (GET and POST only) in order to support
13 Web DAV requests. Scott Robertson [sroberts@codeit.com]
15 *) Allow requests for URLs such as /server/some/other/qualifier to match a
16 FastCGI server defined as /server. This was done primarily for Java
17 Servlets, but is generically useful.
19 *) Change the comm between the PM and the request handlers from a regular
20 file to a pipe and an assortment of other dynamic fixes.
22 *) Log dynamic process termination scheduling and the resulting
23 process exit notification.
25 *) Change the default singleThreshhold from 10 to 0 to prevent the last dynamic
26 application process from being killed off due to low demand.
28 *) Clean up FastCGI application pathnames (e.g. remove duplicate slashes).
30 *) Fix bugs that prevented dynamic processes from being shutdown when
31 the load subsided. Lars Heete [heete@do.isst.fhg.de]
33 *) Prevent dynamic processes from being scheduled to be started before
34 the init-start-delay or restart-delay period expires. This prevents
35 the queuing of process start requests while an application which has
36 a long initialization period starts up.
38 *) When suexec is enabled, allow the process manager to become root again in
39 order to signal applications it spawned.
41 *) Change the "server started" log message level from INFO to WARN.
43 *) Fix dynamic server "Connection Refused" handling.
45 *) Fix demand-based dynamic process spawing (uses a more portable approach).
47 *) Add -idle-timeout arg to FastCgiServer, FastCgiExternalServer, and
48 FastCgiConfig. mod_fastcgi will now abort a connection if inactive for
49 longer than this period. It applies to the initiation of connections as
50 well (and thus is similar to appConnTimeout). Default is 30 seconds.
52 *) appConnTimeout is 0 by default now resulting in blocking connect()s. This
53 is more platform portable/predicable.
55 *) Leave STDOUT and STDERR open to the main server error_log. This should
56 eliminate a pile of protocol errors and having to track down 3rd party
57 libs writing to stderr inadvertantly. This doesn't mean these shouldn't
58 be fixed in the application to use the FastCGI I/O, it just allows these
59 applications to run when previously they'd crash and burn.
61 *) Add the pass-header arg to server directives.
63 *) Allow supplementary groups to be passed to spawned FastCGI applications (as
64 is done for CGI) - at least until PR2580 is resolved.
66 *) Initialize the default (empty) environment in a more socially acceptable
69 *) Miscellaneous doc improvements based on notes I've collected up over the
72 *) Fix a bug in the stderr handling introduced in 2.2.2. There were conditions
73 that resulted in not or improperly terminated strings.
75 *) Fix the call to setsockopt() to disable Nagle. [based on a bug report by
76 Johannes Plassmann <johannes.plassmann@pdb.siemens.de>]
78 Changes with mod_fastcgi 2.2.2
80 *) Added support for blocking connect()s by setting appConnTimeout to 0.
81 Non-blocking connect()s (the default) can be troublesome on some platforms.
82 Its expected that blocking connect()s will become the default (and
83 non-blocking connect()s will become optional) in the next release.
85 *) Dump a compile time error if the version of Apache is too old.
87 *) Wrap the SIGPIPE handler manipulation code such that it is only applicable
88 to Apache releases prior to 1.3.6.
90 *) Minor tweaks for RUSSIAN_APACHE. [Sergey Gershtein <sg@mplik.ru>]
94 Always restart a failed dynamic application if it is the last instance.
95 This means there once started there will always be at least one process
96 instance of a dynamic application.
98 Send PLEASE_START to the PM when a connect() results in an ECONNREFUSED.
99 ECONNREFUSED means the listen queue is full (or there isn't one). Asking
100 the PM to start (another) application instance may help empty it faster.
102 Change two sleep() calls to select() based snoozes because alarm() is in
103 effect and sleep() and alarm() don't always play nice together.
105 Fixed a couple of error messages.
107 *) Fix -listen-queue-depth arg on FastCgiConfig (dynamic). Previously it
108 was ignored and the default was always used.
110 *) Allow the -initial-env argument to be used to pass variables from the
111 Apache process environment to the FastCGI server (by specifying a
112 variable name without the "=" or a value). [suggested by
113 Martin Lichtin <lichtin@oanda.com>]
115 *) Cleanup some debug macros.
117 *) Improved script stderr handling. [based on suggestions by
118 David Birnbaum <davidb@chelsea.net>]
120 *) Added IRIX and FreeBSD to the list of supported platforms and other minor
121 updates to the INSTALL doc.
123 *) Changed the default listen-queue-depth (FCGI_DEFAULT_LISTEN_Q) from
124 5 to 100. Its still configurable with the -listen-queue-depth option.
125 This should help eliminate the FAQ - Why do I see "Connection Refused"
128 *) Fix a bug in FastCgiExternalServer that broke support for external servers
129 on other hosts. [Dave Neuer <dneuer@futuristics.net>]
131 Changes with mod_fastcgi 2.2.1
133 *) Updates to the INSTALL doc to describe building as a DSO.
135 *) If the FastCgiIpcDir directive was in use and "httpd -t" was issued while
136 Apache was running ("apachectl restart" does this), the contents of the
137 dynamic directory (dynamic sockets and the mbox) would be blown away
140 *) Add an extern declaration for ap_sys_siglist (its exposed but not in any
141 Apache header file) to prevent a compile error on systems without
144 *) During auth requests, the subprocess_env table was left holding variables
145 sent to the auth FastCGI server (including REMOTE_PASSWD!) which means they
146 were passed to other processing phases (such as CGI/FastCGI). The
147 subprocess_env is now restored to its pre-auth condition.
149 *) Added a FastCGI Authorizer Role compatibility mode which implements the
150 specification to the tee. Use -compat arg with any of the Auth server
151 directives. This is intended for new and existing FastCGI authorizer
152 applications that require compatibility with other server implementations.
154 *) Fix logging from the process manager. Previously, all calls to the
155 logging routines used NULL for the server_rec. This works fine in a dev
156 environment, but results in most of these messages being tossed by
157 log_error_core() because DEFAULT_LOGLEVEL is too low. Odd logic.
159 *) Always setup fr->header so that in the event the FastCGI server
160 doesn't get a valid FastCGI protocol header over the wire, we
161 can still print our error message to the log from do_work().
163 *) Fix a compile error in open_connection_to_fs() on systems with TCP_NODELAY
164 defined (Don Locrasto [locrasto@iceminer.com]).
166 Changes with mod_fastcgi 2.2.0
168 *) Serious rewrite of mod_fastcgi.html. An example conf will have
169 to wait until another time - I'm burned out on docs updates.
171 *) Add to the mod_fastcgi to the server version string.
173 *) SCO doesn't like the const in the arg to inet_addr() and
176 *) Use Apache's NET_SIZE_T to get the correct arg type for getsockopt().
178 *) Deleted redundant header files from fcgi.h.
180 *) Full path names are no longer required for directives which take
181 file names as arguments. Like other Apache directives you can
182 specify paths from server_root (don't start with a "/").
184 *) The fcgiKillMgr is gone. This intermediate process sat between
185 the Apache parent and the FastCGI Process Manager under Apache
186 1.2. The process manager tries to name itself fcgi_pm (it used
189 *) Authentication, Authorization, and Access phases are now
190 supported. I've bent the FastCgi spec a bit: as many of the standard
191 environment variables (that were easy) are sent (the spec says don't
192 send some by name), all headers (except Status) sent by the auth
193 FastCgi server are passed to subprocesses (CGI/FastCGI invocations)
194 as environment variables rather than just those prefixed by
195 "Variable-". Custom responses for auth failures aren't supported
196 (yet.. we'll see what the demand for this is like). In addition to
197 the FastCGI protocol defined environment variable "FCGI_ROLE" being
198 set to "AUTHORIZER" an environment variable "FCGI_APACHE_ROLE" is
199 set indicating which of the three phases is being processed. See
200 the docs for info on the new per-directory directives:
202 FastCgiAuthenticator, FastCgiAuthenticatorAuthoritative,
203 FastCgiAuthorizer, FastCgiAuthorizerAuthoritative,
204 FastCgiAccessChecker, FastCgiAccessCheckerAuthoritative
206 *) The code has been broken into logical chunks making figuring it
207 out much easier. All of the #defines a user may want to change are
208 now in mod_fastcgi.h.
210 *) All the logging has been converted to the "supported" Apache
211 logging routines, ap_log_error() and ap_log_rerror(). Log entries
212 for request specific errors should now be directed to the correct
213 server errorlog (if your running more than one). Some attempt has
214 been made at setting appropriate log levels (comments of course are
215 welcome) and printing errno info when its of value. All of log
216 messages now have "FastCGI" in them (nice for grep).
218 *) All of the calls to Apache routines have been updated to the ap_
219 convention, eliminating the need for the compat header.
221 *) By default we no longer flush every piece of data we get from the
222 FastCGI server to the client. This allows the FastCGI server to
223 release without having to wait for the client flush to complete.
224 The old behaviour can revived with the -flush argument to AppClass.
226 *) I renamed some of the directives for consistency (the old names
229 AppClass -> FastCgiServer,
230 FCGIConfig -> FastCgiConfig,
231 ExternalAppClass -> FastCgiExternalServer
233 *) Directive arguments are no longer case sensitive.
235 *) The module now uses hard_timeout() rather than soft_timeout. This
236 means that the module will longjump out of the request if the
237 drop-dead timeout expires (set with Apache's Timeout directive, the
238 default is 5min) or if the client closes the connection (SIGPIPE).
239 This is more typical for a Apache modules and I'm not convinced we
240 properly handle all of the error cases well enough to use
241 soft_timeout (I can think of one place we don't anyway). This means
242 your FastCGI application can see SIGPIPE.
244 *) SIGPIPE is now ignored by default in FastCGI servers spawned by
245 Apache. Without this the default behaviour is exit().
247 *) It should (I think) run now under DSO now.
249 *) The bug which prevented sending of binary data in 2.1b1 is fixed.
251 *) The broken -initial-env bug ins 2.1b1 is fixed.
253 *) Maybe some other stuff.
257 *** Well there's a big gap here. Maybe I'll go back and extract the CVS
258 commit notes and stick 'em in here.. when I get some free time. ;)
263 *** Originally from docs/README..
265 Apache[X1.03.01]/mod_fastcgi[02.00.05] 19970909 unsupported
266 From: Stanley Gambarin <stanleyg@cs.bu.edu>
268 *) Yet more changes to the source distribution. Separated out
269 the Tcl dynamic string and buffer libraries into separate files.
270 Did the same for the OS library, however, it is not really an
271 abstraction... more like a bunch of wrappers. Added to DEVNOTES
272 a note about the problem of Apache not allowing including header
275 Apache[X1.03.01]/mod_fastcgi[02.00.04] 19970908 unsupported
276 From: Stanley Gambarin <stanleyg@cs.bu.edu>
278 *) More changes to the installation scripts. Theoretically we
279 should be able to to use the same Makefile/installation script
280 for both Apache 1.2.x and 1.3.x. Need some sort of the abstraction
281 layer to use the same source code for both 1.2.x and 1.3.x sources.
283 Apache[X1.03.01]/mod_fastcgi[02.00.03] 19970905 unsupported
284 From: Stanley Gambarin <stanleyg@cs.bu.edu>
286 *) Created Makefile to compile the module into the archive library,
287 just like the proxy module. Also initial draft of the installation
288 script, which should work for both 1.2 and 1.3 sources of Apache.
290 Apache[01.02.04]/mod_fastcgi[02.00.02] 19970903 unsupported
291 From: Stanley Gambarin <stanleyg@cs.bu.edu>
293 *) More source reorg. Separated out some header files. Need much
294 more work on this. Updated the TODO file. Added fcgivers.h and
295 CHANGES file to track the history.
297 Apache[01.02.04]/mod_fastcgi[02.00.01] 19970902 unsupported
298 From: Stanley Gambarin <stanleyg@cs.bu.edu>
300 *) Source reorganization. This is done to provide a basis for later
301 functionality implementation. See TODO file for more information
302 of what possible future enhancements are possible/desired.
304 Apache[01.02.04]/mod_fastcgi[02.00.00] 19970902 unsupported
305 From: David MacKenzie <djm@va.pubnix.com>
307 *) Create the "dynamic" dir and "mbox" as the user specified in the
308 User and Group directives instead of as root. They are created with
309 restrictive permissions, and then the module checks to see whether the
310 user specified in the User and Group directives has read, write, and
311 execute permissions on them. Those permissions had been explicitly
312 denied by creating them as "root".
314 *) Some documentation was garbled or incomplete.
316 *) When a FastCGI application can't be execl'd, the code used a
317 value of errno that may have been stomped on by intervening system
318 calls. That happened on BSD/OS 2.1, where the error_log was
319 reporting errno=25 (NOTTY) instead of the correct errno=13 (EACCES).
320 An intervening stdio call was setting errno as a side-effect, which
321 it has the right to do.
323 *) Erroneous fprintf arguments, a missing return value, an unused
324 function, missing declarations, and other problems detected by gcc
327 *) Duplicated code to create the lock file name merged into a single
328 function, closing several memory leaks.
330 *) FCGIConfig -minProcesses didn't allow a value of 0, so the last
331 FCGI app in a given dynamic class would live forever, even if it
332 hadn't been requested for weeks.
334 *) The test for whether to keep looking for dynamic app victims to
335 kill was backwards, so no dynamic apps were ever selected as
338 *) The killInterval and updateInterval were being ignored; the
339 sigsuspend() forced recalculations only when a child died,
340 instead of at the intervals specified.
342 *) When a dynamic app couldn't be started, the program SEGV'd when
343 trying to free the ipcAddrPtr twice.
345 *) Remove the dead lock file and socket when cleaning up after a
346 server whose last child has been killed, so the FastCGIHandler
347 isn't fooled into thinking there is still a process serving
350 *) Make file locking robust in the presence of signals.
352 *) Rename some badly named variables.
354 *) Fix typos in many comments.
356 *) Fix some memory and file descriptor leaks.
358 *) Make the blocking kill of a server closer to working.
360 *) More fixes to calculations in dynamic application management.
363 *** Originally from docs/README.OMI..
365 What's New: Version 2.0b1, 16 Apr 1997
367 *) Implemented a mechanism by which FastCGI applications are started by
368 the web server on the first request and continue running. Also, a
369 heuristics have been implemented to allow for dynamic killing of the
370 running FastCGI apps. The configuration options are supplied via
371 the new FCGIConfig directive, that is described in the
374 *) When performing internal redirect, since the original request's body
375 has already been read, do not allow the redirected request to think
376 that is has one [body].
378 *) More conditional compilation for OS/2.
380 *) Fixed occassional "Assertion failed: len > 0, file mod_fastcgi.c"
382 *) Fixed "failed assertion `count >= 0 && count <= bufPtr->length'"
384 *) Added bflush() call after bwrite() in DrainReqOutBuf function to
385 force some output to be written in the period of inactivity. This
386 circumvents the Apache's buffering during the server-push.
388 *) Added a header-parser function placeholder for compatibility with an
391 *) Additional checks have been placed with regard to permissions of the
392 FastCGI processes. These include the existence of the file,
393 ExecCGI and IncludesNOEXEC checks, disallowing of nph- scripts, etc.
395 *) Implemented a restart cleanup, so that no parentlesses FastCGI
396 applications are left after Apache is restarted and/or terminated.
397 This was accomplished via two independent mechanisms, where the
398 first one prevents a start of the process manager on the first
399 reading of the configuration files. The second mechanism deals away
400 from Apache's implementation of fork() (via spawn_child()) and
401 implements its own forking, thus removing any dependencies on the
402 Apache to cleanup processes during termination phase.
404 *) Removed the definition of the Sigfunc, as Apache 2.0b10 defines it
405 itself in the file conf.h
407 What's New: Version 1.4.3, 15 Jan 1997
409 *) Fixed compilation warnings for various platforms, as well as
410 conditional compilation for OS/2.
412 What's New: Version 1.4.2, 12 Dec 1996
414 *) mod_fastcgi.c is ported to Apache 1.2b. Any further development
415 will proceed under this version of Apache web server.
417 *) As the result of porting, the "include virtual" construct of SSI
418 will now work correctly using either <Location> or <Directory>
421 What's New: Version 1.4.1, 4 Dec 1996
423 *) Checks have been removed from the ScanCGIHeaders that provided
424 for the presence of both Status and Location headers as being
425 an error. Contradictory to CGI/1.1 Internet Draft, both of
426 these headers are used by the current CGI applications.
428 What's New: Version 1.4, 22 Nov 1996
430 *) Added the -port option to AppClass, allowing TCP/IP communication.
431 Added the -socket option to AppClass, allowing Unix domain
432 communication via a configurable pathname.
434 *) Added the ExternalAppClass directive, allowing TCP/IP
435 communication with remote FastCGI applications.
437 *) The handler had its own code for generating HTTP response headers;
438 now it uses Apache's. This reduces the size of the module.
439 More importantly, it fixes the bug in which "include virtual"
440 sees the HTTP response headers.
442 *) The response header parser performed very little checking.
443 Now the parser enforces the guidelines in the CGI/1.1
444 Internet-Draft: Status and Location are mutually exclusive,
445 Location can only be a response to GET or HEAD,
446 CGI response headers can't be repeated, etc. (The CGI response
447 headers are the ones the handler interprets: Status, Location,
450 *) The response header parser used to miss CGI headers
451 with no whitespace after the colon, e.g. "Status:200 OK".
453 *) The response header parser sometimes interpreted the first line
454 of content as an RFC822 continuation line.
456 *) The handler implemented a nonstandard version of Location
457 which never used internal redirects. The handler
458 now attempts to implement Location as specified in the
459 CGI/1.1 Internet-Draft. The module documentation explains
462 *) Error log entries from the response header parser were
463 pretty uninformative; they are better now. If a header is
464 malformed, the log entry includes it. If the headers are
465 unterminated rather than malformed, the log entry says that,
466 and says how many bytes were received from the app.
468 *) When the application manager forked a new process, and that
469 process ran into trouble before executing the first instruction
470 of application code, the process used to exit with status =
471 errno. This made certain configuration problems (e.g. incorrect
472 file permissions when server parent is root) quite difficult
473 to pin down. Now the failing child process opens up the error log
474 and writes an informative entry before exiting.
476 *) The module now correctly handles a slash at the end of the
477 DocumentRoot directive. This was a one-line fix.
479 What's New: Version 1.3.3, 17 Oct 1996
481 *) The module now registers its request handler under the name
482 fastcgi-script in addition to the name application/x-httpd-fcgi.
483 This was a one-line addition to the module, but had quite
484 a large impact on the documentation and sample configuration.
486 What's New: Version 1.3.2, 27 Sept 1996
488 *) On some systems (SunOS, Linux?), fopen for append has a bug
489 that strikes when two processes append to the same file.
490 This bug causes the process manager to corrupt the error log.
491 Work around the bug by calling open, then fdopen. A patch
492 to Apache 1.1.1 is also required, as described in mod_fastcgi.html.
493 (Reported by Bob Ramstad.)
495 What's New: Version 1.3.1, 17 Sept 1996
497 *) On Linux and SunOS, the Apache default user_id (-1) is not a
498 legal value of uid_t, so cast it. (Reported by Bob Ramstad,
499 Scott Langley, others.)
501 *) On some systems (Linux, some Solaris?, Irix?), connect requires
502 write access to a Unix Domain socket, so provide it. (Reported
503 by Scott Langley, freeform@wired.com, others.)
505 *) If you hit a .fcg file, but there's no AppClass defined, the
506 error should be NOT_FOUND instead of SERVER_ERROR. (Reported
509 What's New: Version 1.3, 4 Sept 1996
511 *) Module sometimes busy-waited in FastCgiDoWork. Fixed.
513 *) Module violated the Apache buff abstraction.
514 Fixed. Might now work with SSL (not tested).
515 As part of the fix, eliminated support for
516 Apache 1.0x versions.
518 *) Module failed to chmod the Unix domain listening sockets
519 it created, so protections were set according to the current
522 *) Module forked too many process manager processes: One
523 per AppClass. When the Apache parent ran as root, these
524 process manager processes ran as root. New process
525 manager is a single process (only started if AppClass is
526 used) and runs with same privileges as other children.
527 The new process manager doesn't do any polling, so
528 there's less system overhead than before.
530 *) AppClass insisted on getting at least two arguments. Fixed.
532 *) Module used setjmp/longjmp, causing compiler warnings
533 on some platforms (e.g. Linux.) The module no longer
536 *) Module created listening sockets in /tmp, where they were
537 sometimes wiped out by cleanup scripts. Added FastCgiIpcDir
538 directive to give control over the location of listening
541 *) Module wrote error log entries without a timestamp. Fixed.
543 *) AppClass directive wrote error messages to stderr in addition
544 to returning a char * message to the Apache core. Fixed.
546 What's New: Version 1.2, 3 June 1996
548 *) Ported from Apache-1.0-based code to Apache-1.1b2 internals by
551 Add version string: APACHEVERSION macro to mod_fastcgi.c
553 *) chown FastCGI socket to user_id and setuid to user_id for app
556 *) Modify GetFromStream() by having it call an OS dependent
557 function GetStreamSize(FILE *) which uses FILE internal data member.
558 Linux users might need to modify GetStreamSize().
560 What's New: Version 1.1, 10 May 1996
562 *) If you specify a non-existent executable in the AppClass directive,
563 or if the file exists and it does not have execute permission you
564 get a constant stream of error messages telling you that "program
565 terminated due to a signal".
567 *) The mod_fastcgi module should use the standard Apache error logging
568 facility instead of writing to stderr.
570 Version 1.0, 30 April 1996