3 *) Add a -min-server-life option to the FastCgiConfig and
4 FastCgiServer directives to provide better control of the
5 restart backoff feature. Benjamin Osheroff [ben@gimbo.net]
9 *) [WIN] Fix handle leaks in the process manager.
10 Sascha Schumann [sascha@schumann.cx]
12 *) [WIN] Use a permanent pool for allocating the SystemRoot
13 environment variable. Sakamoto [ringring@zb4.so-net.ne.jp]
15 *) [WIN] Fix starting of scripts under AP2.
17 *) Do the connect() to the application *after* collecting a chunk
18 of client data. This reinstates the pre-2.4.0 behaviour.
19 James Jurach [muaddib@erf.net].
21 *) Remove an assert that was triggering on WIN when spawn() failed.
23 *) Provide the NO_SUEXEC_FOR_AP_USER_N_GROUP macro for building
24 mod_fastcgi with the AP13 suexec behaviour (don't use suexec
25 if httpd's user and group match that needed for the application).
27 *) Prevent the use of all but the "auth" directives from being used
28 anywhere but in global scope. Prevent more than one instance
29 of the FastCgiWrapper directive.
31 *) Return NOT_FOUND (404) or FORBIDDEN (403) instead of
32 INTERNAL_SERVER_ERROR (500) when there are configuration issues
33 or the script does't exist. Suggested by Jeff Lawson [bovine@ud.com].
37 *) [*nix Security] - When FastCgiWrapper (FastCgiSuexec) was in use
38 and a vhost configured to use the same uid/gid as the main
39 server, mod_fastcgi would not bother using the wrapper (suexec)
40 because its effective uid/gid was already appropriate. This is
41 consistent with Apache's v1.3 mod_cgi behaviour. There are two
42 problems with this approach: 1) when FastCgWrapper is in use
43 mod_fastcgi's process manager keeps its root privileges (as its
44 real uid/gid) so it can terminate the applications its starts -
45 this privilege was being passed to applications when the use of
46 the wrapper was bypassed 2) wrappers are often employed to
47 perform functionality beyond setting the uid/gid - by not calling
48 the wrapper under certain circumstances, application invocation
49 environments were inconsistent. With this change, the wrapper is
50 always used (when enabled) under both Apache 1.3 and 2.
51 Reported by Michael Richards [michael@fastmail.ca].
53 *) [*nix/AP2] Use the vhost uid/gid instead of the server uid/gid
54 for dynamic application invocation when the FastCgiWrapper is in use.
55 Reported by Michael Richards [michael@fastmail.ca]
57 *) [*nix] Fix handling of FastCgiWrapper when passed a real path,
58 i.e. other than "on" or "off". Michael Richards [michael@fastmail.ca]
60 *) Eliminate the logging of "incomplete headers (0 bytes) received from
61 server" when a client aborts.
63 *) [WIN32] Fix a delay in handling large POSTs to named pipe based
64 servers. Philip Gladstone [philip@okena.com]
66 *) [*nix/AP2] Prevent the module from being initalized twice at startup
67 (resulting in confusing error messages to the log).
69 *) Eliminate the need for SetHandler or AddHandler with static or
70 external applications.
72 *) Limit PM requests to start a dynamic application
73 to 5sec to prevent endless spinning (this is a drop-dead
74 limit that should only occur if the socket/named_pipe directory
75 is removed out from under a running server).
77 *) [*nix] Change the default socket directory from /tmp/fcgi to:
79 Apache2 - RUNTIMEDIR/fastcgi
81 *) Add -user & -group args to FastCgiServer and FastCgiExternalServer
82 for use with wrappers (in lieu of finding the user/group associated
83 with a virtual host - under Apache2 this isn't accomodated).
85 *) [WIN32] Under Apache2, require v2.0.41 or later in order to pickup my
86 apr_proc_create() changes.
88 *) Log when invoking and restoring the restart backoff policy.
90 *) [WIN32] Prevent intermittent ReadFile() failures (properly initialize the
91 OVERLAPPED structure).
93 *) Eliminate need for dummy files for external servers under Apache2
95 *) Fix auth compatibility mode handling for access checker and authorizer
97 *) Fix HEAD request handling. Based on a patch by
98 Chris Lightfoot [chris@ex-parrot.com]
100 *) [*nix] When autoupdate is enabled touch the socket when restarting
101 the processes to prevent further requests.
102 Eckebrecht von Pappenheim [evp@heise.de]
104 *) Apache 2.0 support.
106 *) [WIN32] Don't read from a potentially closed named pipe.
107 Philip Gladstone [philip@okena.com]
109 *) Require the Apache version 1.3.6 or later to eliminate some signal
112 *) [WIN32] Use asyncronous io with named pipes instead of polled
113 nonblocking io. This should eliminate the last of the npipe issues.
115 *) Handle an application returning a complete and valid response without
116 having consumed all of the data sent to it.
118 *) Consume remaining client data (RESPONDERs only) if any.
120 *) Add support for backing off attempts to start applications that continuously
121 fail to start. Three new macros defined in mod_fastcgi.h control this
122 behaviour: MAX_FAILED_STARTS, RUNTIME_SUCCESS_INTERVAL, FAILED_STARTS_DELAY
124 *) [WIN32] Add (back) support for use of TerminateProcess() to accomodate
125 applications that do not (properly) support the shutdown event (this
126 feature was introduced in fcgi2 2.2.2 and improved in 2.2.4). The
127 new macro WIN32_SHUTDOWN_GRACEFUL_WAIT in mod_fastcgi.h conrols the
128 interval between signaling a proper shutdown and wacking the process(s)
129 with a TerminateProcess().
131 *) [WIN32] Don't set the OVERLAPPED_IO flag on NamedPipe listen HANDLEs -
132 setting it was just plain broken.
134 *) [WIN32] Fix the accept mutex - all applications were sharing one!?
136 *) Fix 'FastCgiConfig -autoUpdate'.
138 *) Fix 'FastCgiConfig -flush'.
140 *) Prevent silly maxProcesses and processSlack combinations.
141 Dmitry Dorofeev [dima@yasp.com]
143 *) Properly handle the killing of idle processes when one takes a long time
144 to exit once signaled down (or the config is funky).
145 Dmitry Dorofeev [dima@yasp.com]
147 *) Always kill the youngest instance of an application. Suggested by
148 Dmitry Dorofeev [dima@yasp.com]
152 *) Delay the logging of write errors to the pm to account for shutdown/restart.
154 *) (Win32) An assortment of fixes.
156 *) Fix some broken casts that were likely the cause of an assert.
158 *) Win32. Eliminate forward slashes from the named pipe path name.
159 Gerald Richter [richter@ecos.de]
161 *) SIGUSR2 is no longer blocked in the process manager and the fastcgi
162 applications it spawns. [ryans@amazon.com]
164 *) Added support for the -flush argument to FastCgiConfig.
165 Eric Sit [esit@alum.mit.edu]
167 *) Change the "which call to module_init() is this" check to a more
168 reliable approach. Doru Petrescu [pdoru@kappa.ro]
170 *) Close the old pipe file descriptor in apache main on USR1/HUP
171 (elimnates a small leak). James E. Jurach Jr. jjurach@fundsxpress.com
173 *) Fix a bug in fcgi_config_set_authoritative_slot(). Tetsuya Furukawa
174 [tetsuya@secom-sis.co.jp]
176 *) Eliminate the use of locks to assist in the clean shutdown of
177 applications. Instead, it is assumed that applications handle
178 termination signals properly (this is now embedded in the C
181 *) Fix Win32 process termination. Proper operation requires the use of
182 an updated application lib (termination is now signalled with an
183 Event and handled by specialized thread).
185 *) Docs cleanup. Andrew Benham [adsb@bigfoot.com]
187 *) Added code so if the last instance of a dynamic application died without
188 provocation, then don't restart it if singleThreshold > 0 (i.e. if the
189 configuration allows the last instance to be killed, then allow it to die).
190 Andrew Benham [adsb@bigfoot.com]
192 *) Fix the loadFactor calculation used to determine when dyanmic
193 applications could be killed off due to low demand [adsb@bigfoot.com].
195 *) Fix a deadlock condition that could occur with Win32 named
198 *) Fix a potential deadlock condition when FastCGI application
199 sent responses while still reading the client request (POST data).
203 *) Allow absolute pathnames in the -socket argument. Suggested by
204 Christian Jaeger [christian.jaeger@sl.ethz.ch].
206 *) Don't invoke suexec when the user/group for the fastcgi application
207 is the same as the apache main server. This is consistent with
208 apache's suexec handling. Suggested by Nikolaus Rath [Nikolaus@rath.org].
210 *) Reset the apache drop dead timer upon successful read or writes
211 to/from the client. This eliminates timeouts that were occuring
212 during the large file transfers to/from slow clients.
214 *) Support generic wrappers such as cgiwrap by eliminating dependencies
215 on Apache's SUEXEC, renaming the FastCgiSuexec directive
216 FastCgiWrapper and eliminating any checks regarding the target
217 application (this is the repsonibility of the wrapper).
219 *) Fix a nasty bug that occurred when a client aborted a POST request
220 before the connection to a dynamic FastCGI application was opened.
221 The application's lock file descriptor wasn't setup but was being
222 closed which resulted in FD0 being closed. Normally this is open to
223 /dev/null and should pose no problems except that because the FD
224 was available it was being returned by Apache's accept(). This
225 caused it to be registered for pool cleanup. mod_cgi though moves
226 the CGI stdin pipe to FD0 and thus it was getting waxed during pool
227 cleanup. Problem identified by checksum@163.net.
229 Changes with mod_fastcgi 2.2.8
231 *) Eliminate the concept of disabled applications. If a failure
232 occurs trying to setup an application (e.g. bind() error) its
233 tried repeatedly every init-start-delay seconds.
235 *) Tweak to Makefile.tmpl to support DSOs. Dave Hill [ddhill@zk3.dec.com]
237 Changes with mod_fastcgi 2.2.6
239 *) Shutdown the PM when Apache appears to have disappeared.
241 *) seteuid() tweak for HP-UX 11. Milton L. Hankins [mlh@swl.msd.ray.com]
243 *) (Win32) More dynamic fixes.
245 *) (Win32) Eliminate the per application named pipe mutex. Its now
246 per process to keep the lib happy (the use _FCGI_MUTEX_ should
247 removed from the lib). This allows pipe-based applications to
248 now handle multiple simoultanous requests (one per process).
250 *) Increase the number of open FDs we look to close when spawning apps.
252 *) Prevent an assert from popping unnecessarily.
254 *) (Win32) Add support for interpretter scripts.
256 *) (Win32) Fix named pipe handling (problems with large responses).
258 *) (Win32) Remove the "can exec" check.
260 Changes with mod_fastcgi 2.2.4
262 *) Beta WinNT support. David Allen [djallen@raleigh.ibm.com] and
263 Rob Saccoccio [robs@chelsea.net]
265 *) Remove request type restriction (GET and POST only) in order to support
266 Web DAV requests. Scott Robertson [sroberts@codeit.com]
268 *) Allow requests for URLs such as /server/some/other/qualifier to match a
269 FastCGI server defined as /server. This was done primarily for Java
270 Servlets, but is generically useful.
272 *) Change the comm between the PM and the request handlers from a regular
273 file to a pipe and an assortment of other dynamic fixes.
275 *) Log dynamic process termination scheduling and the resulting
276 process exit notification.
278 *) Change the default singleThreshhold from 10 to 0 to prevent the last dynamic
279 application process from being killed off due to low demand.
281 *) Clean up FastCGI application pathnames (e.g. remove duplicate slashes).
283 *) Fix bugs that prevented dynamic processes from being shutdown when
284 the load subsided. Lars Heete [heete@do.isst.fhg.de]
286 *) Prevent dynamic processes from being scheduled to be started before
287 the init-start-delay or restart-delay period expires. This prevents
288 the queuing of process start requests while an application which has
289 a long initialization period starts up.
291 *) When suexec is enabled, allow the process manager to become root again in
292 order to signal applications it spawned.
294 *) Change the "server started" log message level from INFO to WARN.
296 *) Fix dynamic server "Connection Refused" handling.
298 *) Fix demand-based dynamic process spawing (uses a more portable approach).
300 *) Add -idle-timeout arg to FastCgiServer, FastCgiExternalServer, and
301 FastCgiConfig. mod_fastcgi will now abort a connection if inactive for
302 longer than this period. It applies to the initiation of connections as
303 well (and thus is similar to appConnTimeout). Default is 30 seconds.
305 *) appConnTimeout is 0 by default now resulting in blocking connect()s. This
306 is more platform portable/predicable.
308 *) Leave STDOUT and STDERR open to the main server error_log. This should
309 eliminate a pile of protocol errors and having to track down 3rd party
310 libs writing to stderr inadvertantly. This doesn't mean these shouldn't
311 be fixed in the application to use the FastCGI I/O, it just allows these
312 applications to run when previously they'd crash and burn.
314 *) Add the pass-header arg to server directives.
316 *) Allow supplementary groups to be passed to spawned FastCGI applications (as
317 is done for CGI) - at least until PR2580 is resolved.
319 *) Initialize the default (empty) environment in a more socially acceptable
322 *) Miscellaneous doc improvements based on notes I've collected up over the
325 *) Fix a bug in the stderr handling introduced in 2.2.2. There were conditions
326 that resulted in not or improperly terminated strings.
328 *) Fix the call to setsockopt() to disable Nagle. [based on a bug report by
329 Johannes Plassmann <johannes.plassmann@pdb.siemens.de>]
331 Changes with mod_fastcgi 2.2.2
333 *) Added support for blocking connect()s by setting appConnTimeout to 0.
334 Non-blocking connect()s (the default) can be troublesome on some platforms.
335 Its expected that blocking connect()s will become the default (and
336 non-blocking connect()s will become optional) in the next release.
338 *) Dump a compile time error if the version of Apache is too old.
340 *) Wrap the SIGPIPE handler manipulation code such that it is only applicable
341 to Apache releases prior to 1.3.6.
343 *) Minor tweaks for RUSSIAN_APACHE. [Sergey Gershtein <sg@mplik.ru>]
347 Always restart a failed dynamic application if it is the last instance.
348 This means there once started there will always be at least one process
349 instance of a dynamic application.
351 Send PLEASE_START to the PM when a connect() results in an ECONNREFUSED.
352 ECONNREFUSED means the listen queue is full (or there isn't one). Asking
353 the PM to start (another) application instance may help empty it faster.
355 Change two sleep() calls to select() based snoozes because alarm() is in
356 effect and sleep() and alarm() don't always play nice together.
358 Fixed a couple of error messages.
360 *) Fix -listen-queue-depth arg on FastCgiConfig (dynamic). Previously it
361 was ignored and the default was always used.
363 *) Allow the -initial-env argument to be used to pass variables from the
364 Apache process environment to the FastCGI server (by specifying a
365 variable name without the "=" or a value). [suggested by
366 Martin Lichtin <lichtin@oanda.com>]
368 *) Cleanup some debug macros.
370 *) Improved script stderr handling. [based on suggestions by
371 David Birnbaum <davidb@chelsea.net>]
373 *) Added IRIX and FreeBSD to the list of supported platforms and other minor
374 updates to the INSTALL doc.
376 *) Changed the default listen-queue-depth (FCGI_DEFAULT_LISTEN_Q) from
377 5 to 100. Its still configurable with the -listen-queue-depth option.
378 This should help eliminate the FAQ - Why do I see "Connection Refused"
381 *) Fix a bug in FastCgiExternalServer that broke support for external servers
382 on other hosts. [Dave Neuer <dneuer@futuristics.net>]
384 Changes with mod_fastcgi 2.2.1
386 *) Updates to the INSTALL doc to describe building as a DSO.
388 *) If the FastCgiIpcDir directive was in use and "httpd -t" was issued while
389 Apache was running ("apachectl restart" does this), the contents of the
390 dynamic directory (dynamic sockets and the mbox) would be blown away
393 *) Add an extern declaration for ap_sys_siglist (its exposed but not in any
394 Apache header file) to prevent a compile error on systems without
397 *) During auth requests, the subprocess_env table was left holding variables
398 sent to the auth FastCGI server (including REMOTE_PASSWD!) which means they
399 were passed to other processing phases (such as CGI/FastCGI). The
400 subprocess_env is now restored to its pre-auth condition.
402 *) Added a FastCGI Authorizer Role compatibility mode which implements the
403 specification to the tee. Use -compat arg with any of the Auth server
404 directives. This is intended for new and existing FastCGI authorizer
405 applications that require compatibility with other server implementations.
407 *) Fix logging from the process manager. Previously, all calls to the
408 logging routines used NULL for the server_rec. This works fine in a dev
409 environment, but results in most of these messages being tossed by
410 log_error_core() because DEFAULT_LOGLEVEL is too low. Odd logic.
412 *) Always setup fr->header so that in the event the FastCGI server
413 doesn't get a valid FastCGI protocol header over the wire, we
414 can still print our error message to the log from do_work().
416 *) Fix a compile error in open_connection_to_fs() on systems with TCP_NODELAY
417 defined (Don Locrasto [locrasto@iceminer.com]).
419 Changes with mod_fastcgi 2.2.0
421 *) Serious rewrite of mod_fastcgi.html. An example conf will have
422 to wait until another time - I'm burned out on docs updates.
424 *) Add to the mod_fastcgi to the server version string.
426 *) SCO doesn't like the const in the arg to inet_addr() and
429 *) Use Apache's NET_SIZE_T to get the correct arg type for getsockopt().
431 *) Deleted redundant header files from fcgi.h.
433 *) Full path names are no longer required for directives which take
434 file names as arguments. Like other Apache directives you can
435 specify paths from server_root (don't start with a "/").
437 *) The fcgiKillMgr is gone. This intermediate process sat between
438 the Apache parent and the FastCGI Process Manager under Apache
439 1.2. The process manager tries to name itself fcgi_pm (it used
442 *) Authentication, Authorization, and Access phases are now
443 supported. I've bent the FastCgi spec a bit: as many of the standard
444 environment variables (that were easy) are sent (the spec says don't
445 send some by name), all headers (except Status) sent by the auth
446 FastCgi server are passed to subprocesses (CGI/FastCGI invocations)
447 as environment variables rather than just those prefixed by
448 "Variable-". Custom responses for auth failures aren't supported
449 (yet.. we'll see what the demand for this is like). In addition to
450 the FastCGI protocol defined environment variable "FCGI_ROLE" being
451 set to "AUTHORIZER" an environment variable "FCGI_APACHE_ROLE" is
452 set indicating which of the three phases is being processed. See
453 the docs for info on the new per-directory directives:
455 FastCgiAuthenticator, FastCgiAuthenticatorAuthoritative,
456 FastCgiAuthorizer, FastCgiAuthorizerAuthoritative,
457 FastCgiAccessChecker, FastCgiAccessCheckerAuthoritative
459 *) The code has been broken into logical chunks making figuring it
460 out much easier. All of the #defines a user may want to change are
461 now in mod_fastcgi.h.
463 *) All the logging has been converted to the "supported" Apache
464 logging routines, ap_log_error() and ap_log_rerror(). Log entries
465 for request specific errors should now be directed to the correct
466 server errorlog (if your running more than one). Some attempt has
467 been made at setting appropriate log levels (comments of course are
468 welcome) and printing errno info when its of value. All of log
469 messages now have "FastCGI" in them (nice for grep).
471 *) All of the calls to Apache routines have been updated to the ap_
472 convention, eliminating the need for the compat header.
474 *) By default we no longer flush every piece of data we get from the
475 FastCGI server to the client. This allows the FastCGI server to
476 release without having to wait for the client flush to complete.
477 The old behaviour can revived with the -flush argument to AppClass.
479 *) I renamed some of the directives for consistency (the old names
482 AppClass -> FastCgiServer,
483 FCGIConfig -> FastCgiConfig,
484 ExternalAppClass -> FastCgiExternalServer
486 *) Directive arguments are no longer case sensitive.
488 *) The module now uses hard_timeout() rather than soft_timeout. This
489 means that the module will longjump out of the request if the
490 drop-dead timeout expires (set with Apache's Timeout directive, the
491 default is 5min) or if the client closes the connection (SIGPIPE).
492 This is more typical for a Apache modules and I'm not convinced we
493 properly handle all of the error cases well enough to use
494 soft_timeout (I can think of one place we don't anyway). This means
495 your FastCGI application can see SIGPIPE.
497 *) SIGPIPE is now ignored by default in FastCGI servers spawned by
498 Apache. Without this the default behaviour is exit().
500 *) It should (I think) run now under DSO now.
502 *) The bug which prevented sending of binary data in 2.1b1 is fixed.
504 *) The broken -initial-env bug ins 2.1b1 is fixed.
506 *) Maybe some other stuff.
510 *** Well there's a big gap here. Maybe I'll go back and extract the CVS
511 commit notes and stick 'em in here.. when I get some free time. ;)
516 *** Originally from docs/README..
518 Apache[X1.03.01]/mod_fastcgi[02.00.05] 19970909 unsupported
519 From: Stanley Gambarin <stanleyg@cs.bu.edu>
521 *) Yet more changes to the source distribution. Separated out
522 the Tcl dynamic string and buffer libraries into separate files.
523 Did the same for the OS library, however, it is not really an
524 abstraction... more like a bunch of wrappers. Added to DEVNOTES
525 a note about the problem of Apache not allowing including header
528 Apache[X1.03.01]/mod_fastcgi[02.00.04] 19970908 unsupported
529 From: Stanley Gambarin <stanleyg@cs.bu.edu>
531 *) More changes to the installation scripts. Theoretically we
532 should be able to to use the same Makefile/installation script
533 for both Apache 1.2.x and 1.3.x. Need some sort of the abstraction
534 layer to use the same source code for both 1.2.x and 1.3.x sources.
536 Apache[X1.03.01]/mod_fastcgi[02.00.03] 19970905 unsupported
537 From: Stanley Gambarin <stanleyg@cs.bu.edu>
539 *) Created Makefile to compile the module into the archive library,
540 just like the proxy module. Also initial draft of the installation
541 script, which should work for both 1.2 and 1.3 sources of Apache.
543 Apache[01.02.04]/mod_fastcgi[02.00.02] 19970903 unsupported
544 From: Stanley Gambarin <stanleyg@cs.bu.edu>
546 *) More source reorg. Separated out some header files. Need much
547 more work on this. Updated the TODO file. Added fcgivers.h and
548 CHANGES file to track the history.
550 Apache[01.02.04]/mod_fastcgi[02.00.01] 19970902 unsupported
551 From: Stanley Gambarin <stanleyg@cs.bu.edu>
553 *) Source reorganization. This is done to provide a basis for later
554 functionality implementation. See TODO file for more information
555 of what possible future enhancements are possible/desired.
557 Apache[01.02.04]/mod_fastcgi[02.00.00] 19970902 unsupported
558 From: David MacKenzie <djm@va.pubnix.com>
560 *) Create the "dynamic" dir and "mbox" as the user specified in the
561 User and Group directives instead of as root. They are created with
562 restrictive permissions, and then the module checks to see whether the
563 user specified in the User and Group directives has read, write, and
564 execute permissions on them. Those permissions had been explicitly
565 denied by creating them as "root".
567 *) Some documentation was garbled or incomplete.
569 *) When a FastCGI application can't be execl'd, the code used a
570 value of errno that may have been stomped on by intervening system
571 calls. That happened on BSD/OS 2.1, where the error_log was
572 reporting errno=25 (NOTTY) instead of the correct errno=13 (EACCES).
573 An intervening stdio call was setting errno as a side-effect, which
574 it has the right to do.
576 *) Erroneous fprintf arguments, a missing return value, an unused
577 function, missing declarations, and other problems detected by gcc
580 *) Duplicated code to create the lock file name merged into a single
581 function, closing several memory leaks.
583 *) FCGIConfig -minProcesses didn't allow a value of 0, so the last
584 FCGI app in a given dynamic class would live forever, even if it
585 hadn't been requested for weeks.
587 *) The test for whether to keep looking for dynamic app victims to
588 kill was backwards, so no dynamic apps were ever selected as
591 *) The killInterval and updateInterval were being ignored; the
592 sigsuspend() forced recalculations only when a child died,
593 instead of at the intervals specified.
595 *) When a dynamic app couldn't be started, the program SEGV'd when
596 trying to free the ipcAddrPtr twice.
598 *) Remove the dead lock file and socket when cleaning up after a
599 server whose last child has been killed, so the FastCGIHandler
600 isn't fooled into thinking there is still a process serving
603 *) Make file locking robust in the presence of signals.
605 *) Rename some badly named variables.
607 *) Fix typos in many comments.
609 *) Fix some memory and file descriptor leaks.
611 *) Make the blocking kill of a server closer to working.
613 *) More fixes to calculations in dynamic application management.
616 *** Originally from docs/README.OMI..
618 What's New: Version 2.0b1, 16 Apr 1997
620 *) Implemented a mechanism by which FastCGI applications are started by
621 the web server on the first request and continue running. Also, a
622 heuristics have been implemented to allow for dynamic killing of the
623 running FastCGI apps. The configuration options are supplied via
624 the new FCGIConfig directive, that is described in the
627 *) When performing internal redirect, since the original request's body
628 has already been read, do not allow the redirected request to think
629 that is has one [body].
631 *) More conditional compilation for OS/2.
633 *) Fixed occassional "Assertion failed: len > 0, file mod_fastcgi.c"
635 *) Fixed "failed assertion `count >= 0 && count <= bufPtr->length'"
637 *) Added bflush() call after bwrite() in DrainReqOutBuf function to
638 force some output to be written in the period of inactivity. This
639 circumvents the Apache's buffering during the server-push.
641 *) Added a header-parser function placeholder for compatibility with an
644 *) Additional checks have been placed with regard to permissions of the
645 FastCGI processes. These include the existence of the file,
646 ExecCGI and IncludesNOEXEC checks, disallowing of nph- scripts, etc.
648 *) Implemented a restart cleanup, so that no parentlesses FastCGI
649 applications are left after Apache is restarted and/or terminated.
650 This was accomplished via two independent mechanisms, where the
651 first one prevents a start of the process manager on the first
652 reading of the configuration files. The second mechanism deals away
653 from Apache's implementation of fork() (via spawn_child()) and
654 implements its own forking, thus removing any dependencies on the
655 Apache to cleanup processes during termination phase.
657 *) Removed the definition of the Sigfunc, as Apache 2.0b10 defines it
658 itself in the file conf.h
660 What's New: Version 1.4.3, 15 Jan 1997
662 *) Fixed compilation warnings for various platforms, as well as
663 conditional compilation for OS/2.
665 What's New: Version 1.4.2, 12 Dec 1996
667 *) mod_fastcgi.c is ported to Apache 1.2b. Any further development
668 will proceed under this version of Apache web server.
670 *) As the result of porting, the "include virtual" construct of SSI
671 will now work correctly using either <Location> or <Directory>
674 What's New: Version 1.4.1, 4 Dec 1996
676 *) Checks have been removed from the ScanCGIHeaders that provided
677 for the presence of both Status and Location headers as being
678 an error. Contradictory to CGI/1.1 Internet Draft, both of
679 these headers are used by the current CGI applications.
681 What's New: Version 1.4, 22 Nov 1996
683 *) Added the -port option to AppClass, allowing TCP/IP communication.
684 Added the -socket option to AppClass, allowing Unix domain
685 communication via a configurable pathname.
687 *) Added the ExternalAppClass directive, allowing TCP/IP
688 communication with remote FastCGI applications.
690 *) The handler had its own code for generating HTTP response headers;
691 now it uses Apache's. This reduces the size of the module.
692 More importantly, it fixes the bug in which "include virtual"
693 sees the HTTP response headers.
695 *) The response header parser performed very little checking.
696 Now the parser enforces the guidelines in the CGI/1.1
697 Internet-Draft: Status and Location are mutually exclusive,
698 Location can only be a response to GET or HEAD,
699 CGI response headers can't be repeated, etc. (The CGI response
700 headers are the ones the handler interprets: Status, Location,
703 *) The response header parser used to miss CGI headers
704 with no whitespace after the colon, e.g. "Status:200 OK".
706 *) The response header parser sometimes interpreted the first line
707 of content as an RFC822 continuation line.
709 *) The handler implemented a nonstandard version of Location
710 which never used internal redirects. The handler
711 now attempts to implement Location as specified in the
712 CGI/1.1 Internet-Draft. The module documentation explains
715 *) Error log entries from the response header parser were
716 pretty uninformative; they are better now. If a header is
717 malformed, the log entry includes it. If the headers are
718 unterminated rather than malformed, the log entry says that,
719 and says how many bytes were received from the app.
721 *) When the application manager forked a new process, and that
722 process ran into trouble before executing the first instruction
723 of application code, the process used to exit with status =
724 errno. This made certain configuration problems (e.g. incorrect
725 file permissions when server parent is root) quite difficult
726 to pin down. Now the failing child process opens up the error log
727 and writes an informative entry before exiting.
729 *) The module now correctly handles a slash at the end of the
730 DocumentRoot directive. This was a one-line fix.
732 What's New: Version 1.3.3, 17 Oct 1996
734 *) The module now registers its request handler under the name
735 fastcgi-script in addition to the name application/x-httpd-fcgi.
736 This was a one-line addition to the module, but had quite
737 a large impact on the documentation and sample configuration.
739 What's New: Version 1.3.2, 27 Sept 1996
741 *) On some systems (SunOS, Linux?), fopen for append has a bug
742 that strikes when two processes append to the same file.
743 This bug causes the process manager to corrupt the error log.
744 Work around the bug by calling open, then fdopen. A patch
745 to Apache 1.1.1 is also required, as described in mod_fastcgi.html.
746 (Reported by Bob Ramstad.)
748 What's New: Version 1.3.1, 17 Sept 1996
750 *) On Linux and SunOS, the Apache default user_id (-1) is not a
751 legal value of uid_t, so cast it. (Reported by Bob Ramstad,
752 Scott Langley, others.)
754 *) On some systems (Linux, some Solaris?, Irix?), connect requires
755 write access to a Unix Domain socket, so provide it. (Reported
756 by Scott Langley, freeform@wired.com, others.)
758 *) If you hit a .fcg file, but there's no AppClass defined, the
759 error should be NOT_FOUND instead of SERVER_ERROR. (Reported
762 What's New: Version 1.3, 4 Sept 1996
764 *) Module sometimes busy-waited in FastCgiDoWork. Fixed.
766 *) Module violated the Apache buff abstraction.
767 Fixed. Might now work with SSL (not tested).
768 As part of the fix, eliminated support for
769 Apache 1.0x versions.
771 *) Module failed to chmod the Unix domain listening sockets
772 it created, so protections were set according to the current
775 *) Module forked too many process manager processes: One
776 per AppClass. When the Apache parent ran as root, these
777 process manager processes ran as root. New process
778 manager is a single process (only started if AppClass is
779 used) and runs with same privileges as other children.
780 The new process manager doesn't do any polling, so
781 there's less system overhead than before.
783 *) AppClass insisted on getting at least two arguments. Fixed.
785 *) Module used setjmp/longjmp, causing compiler warnings
786 on some platforms (e.g. Linux.) The module no longer
789 *) Module created listening sockets in /tmp, where they were
790 sometimes wiped out by cleanup scripts. Added FastCgiIpcDir
791 directive to give control over the location of listening
794 *) Module wrote error log entries without a timestamp. Fixed.
796 *) AppClass directive wrote error messages to stderr in addition
797 to returning a char * message to the Apache core. Fixed.
799 What's New: Version 1.2, 3 June 1996
801 *) Ported from Apache-1.0-based code to Apache-1.1b2 internals by
804 Add version string: APACHEVERSION macro to mod_fastcgi.c
806 *) chown FastCGI socket to user_id and setuid to user_id for app
809 *) Modify GetFromStream() by having it call an OS dependent
810 function GetStreamSize(FILE *) which uses FILE internal data member.
811 Linux users might need to modify GetStreamSize().
813 What's New: Version 1.1, 10 May 1996
815 *) If you specify a non-existent executable in the AppClass directive,
816 or if the file exists and it does not have execute permission you
817 get a constant stream of error messages telling you that "program
818 terminated due to a signal".
820 *) The mod_fastcgi module should use the standard Apache error logging
821 facility instead of writing to stderr.
823 Version 1.0, 30 April 1996