3 *) Eliminate the use of locks to assist in the clean shutdown of
4 applications. Instead, it is assumed that applications handle
5 termination signals properly (this is now embedded in the C
8 *) Fix Win32 process termination. Proper operation requires the use of
9 an updated application lib (termination is now signalled with an
10 Event and handled by specialized thread).
12 *) Docs cleanup. Andrew Benham [adsb@bigfoot.com]
14 *) Added code so if the last instance of a dynamic application died without
15 provocation, then don't restart it if singleThreshold > 0 (i.e. if the
16 configuration allows the last instance to be killed, then allow it to die).
17 Andrew Benham [adsb@bigfoot.com]
19 *) Fix the loadFactor calculation used to determine when dyanmic
20 applications could be killed off due to low demand [adsb@bigfoot.com].
22 *) Fix a deadlock condition that could occur with Win32 named
25 *) Fix a potential deadlock condition when FastCGI application
26 sent responses while still reading the client request (POST data).
30 *) Allow absolute pathnames in the -socket argument. Suggested by
31 Christian Jaeger [christian.jaeger@sl.ethz.ch].
33 *) Don't invoke suexec when the user/group for the fastcgi application
34 is the same as the apache main server. This is consistent with
35 apache's suexec handling. Suggested by Nikolaus Rath [Nikolaus@rath.org].
37 *) Reset the apache drop dead timer upon successful read or writes
38 to/from the client. This eliminates timeouts that were occuring
39 during the large file transfers to/from slow clients.
41 *) Support generic wrappers such as cgiwrap by eliminating dependencies
42 on Apache's SUEXEC, renaming the FastCgiSuexec directive
43 FastCgiWrapper and eliminating any checks regarding the target
44 application (this is the repsonibility of the wrapper).
46 *) Fix a nasty bug that occurred when a client aborted a POST request
47 before the connection to a dynamic FastCGI application was opened.
48 The application's lock file descriptor wasn't setup but was being
49 closed which resulted in FD0 being closed. Normally this is open to
50 /dev/null and should pose no problems except that because the FD
51 was available it was being returned by Apache's accept(). This
52 caused it to be registered for pool cleanup. mod_cgi though moves
53 the CGI stdin pipe to FD0 and thus it was getting waxed during pool
54 cleanup. Problem identified by checksum@163.net.
56 Changes with mod_fastcgi 2.2.8
58 *) Eliminate the concept of disabled applications. If a failure
59 occurs trying to setup an application (e.g. bind() error) its
60 tried repeatedly every init-start-delay seconds.
62 *) Tweak to Makefile.tmpl to support DSOs. Dave Hill [ddhill@zk3.dec.com]
64 Changes with mod_fastcgi 2.2.6
66 *) Shutdown the PM when Apache appears to have disappeared.
68 *) seteuid() tweak for HP-UX 11. Milton L. Hankins [mlh@swl.msd.ray.com]
70 *) (Win32) More dynamic fixes.
72 *) (Win32) Eliminate the per application named pipe mutex. Its now
73 per process to keep the lib happy (the use _FCGI_MUTEX_ should
74 removed from the lib). This allows pipe-based applications to
75 now handle multiple simoultanous requests (one per process).
77 *) Increase the number of open FDs we look to close when spawning apps.
79 *) Prevent an assert from popping unnecessarily.
81 *) (Win32) Add support for interpretter scripts.
83 *) (Win32) Fix named pipe handling (problems with large responses).
85 *) (Win32) Remove the "can exec" check.
87 Changes with mod_fastcgi 2.2.4
89 *) Beta WinNT support. David Allen [djallen@raleigh.ibm.com] and
90 Rob Saccoccio [robs@chelsea.net]
92 *) Remove request type restriction (GET and POST only) in order to support
93 Web DAV requests. Scott Robertson [sroberts@codeit.com]
95 *) Allow requests for URLs such as /server/some/other/qualifier to match a
96 FastCGI server defined as /server. This was done primarily for Java
97 Servlets, but is generically useful.
99 *) Change the comm between the PM and the request handlers from a regular
100 file to a pipe and an assortment of other dynamic fixes.
102 *) Log dynamic process termination scheduling and the resulting
103 process exit notification.
105 *) Change the default singleThreshhold from 10 to 0 to prevent the last dynamic
106 application process from being killed off due to low demand.
108 *) Clean up FastCGI application pathnames (e.g. remove duplicate slashes).
110 *) Fix bugs that prevented dynamic processes from being shutdown when
111 the load subsided. Lars Heete [heete@do.isst.fhg.de]
113 *) Prevent dynamic processes from being scheduled to be started before
114 the init-start-delay or restart-delay period expires. This prevents
115 the queuing of process start requests while an application which has
116 a long initialization period starts up.
118 *) When suexec is enabled, allow the process manager to become root again in
119 order to signal applications it spawned.
121 *) Change the "server started" log message level from INFO to WARN.
123 *) Fix dynamic server "Connection Refused" handling.
125 *) Fix demand-based dynamic process spawing (uses a more portable approach).
127 *) Add -idle-timeout arg to FastCgiServer, FastCgiExternalServer, and
128 FastCgiConfig. mod_fastcgi will now abort a connection if inactive for
129 longer than this period. It applies to the initiation of connections as
130 well (and thus is similar to appConnTimeout). Default is 30 seconds.
132 *) appConnTimeout is 0 by default now resulting in blocking connect()s. This
133 is more platform portable/predicable.
135 *) Leave STDOUT and STDERR open to the main server error_log. This should
136 eliminate a pile of protocol errors and having to track down 3rd party
137 libs writing to stderr inadvertantly. This doesn't mean these shouldn't
138 be fixed in the application to use the FastCGI I/O, it just allows these
139 applications to run when previously they'd crash and burn.
141 *) Add the pass-header arg to server directives.
143 *) Allow supplementary groups to be passed to spawned FastCGI applications (as
144 is done for CGI) - at least until PR2580 is resolved.
146 *) Initialize the default (empty) environment in a more socially acceptable
149 *) Miscellaneous doc improvements based on notes I've collected up over the
152 *) Fix a bug in the stderr handling introduced in 2.2.2. There were conditions
153 that resulted in not or improperly terminated strings.
155 *) Fix the call to setsockopt() to disable Nagle. [based on a bug report by
156 Johannes Plassmann <johannes.plassmann@pdb.siemens.de>]
158 Changes with mod_fastcgi 2.2.2
160 *) Added support for blocking connect()s by setting appConnTimeout to 0.
161 Non-blocking connect()s (the default) can be troublesome on some platforms.
162 Its expected that blocking connect()s will become the default (and
163 non-blocking connect()s will become optional) in the next release.
165 *) Dump a compile time error if the version of Apache is too old.
167 *) Wrap the SIGPIPE handler manipulation code such that it is only applicable
168 to Apache releases prior to 1.3.6.
170 *) Minor tweaks for RUSSIAN_APACHE. [Sergey Gershtein <sg@mplik.ru>]
174 Always restart a failed dynamic application if it is the last instance.
175 This means there once started there will always be at least one process
176 instance of a dynamic application.
178 Send PLEASE_START to the PM when a connect() results in an ECONNREFUSED.
179 ECONNREFUSED means the listen queue is full (or there isn't one). Asking
180 the PM to start (another) application instance may help empty it faster.
182 Change two sleep() calls to select() based snoozes because alarm() is in
183 effect and sleep() and alarm() don't always play nice together.
185 Fixed a couple of error messages.
187 *) Fix -listen-queue-depth arg on FastCgiConfig (dynamic). Previously it
188 was ignored and the default was always used.
190 *) Allow the -initial-env argument to be used to pass variables from the
191 Apache process environment to the FastCGI server (by specifying a
192 variable name without the "=" or a value). [suggested by
193 Martin Lichtin <lichtin@oanda.com>]
195 *) Cleanup some debug macros.
197 *) Improved script stderr handling. [based on suggestions by
198 David Birnbaum <davidb@chelsea.net>]
200 *) Added IRIX and FreeBSD to the list of supported platforms and other minor
201 updates to the INSTALL doc.
203 *) Changed the default listen-queue-depth (FCGI_DEFAULT_LISTEN_Q) from
204 5 to 100. Its still configurable with the -listen-queue-depth option.
205 This should help eliminate the FAQ - Why do I see "Connection Refused"
208 *) Fix a bug in FastCgiExternalServer that broke support for external servers
209 on other hosts. [Dave Neuer <dneuer@futuristics.net>]
211 Changes with mod_fastcgi 2.2.1
213 *) Updates to the INSTALL doc to describe building as a DSO.
215 *) If the FastCgiIpcDir directive was in use and "httpd -t" was issued while
216 Apache was running ("apachectl restart" does this), the contents of the
217 dynamic directory (dynamic sockets and the mbox) would be blown away
220 *) Add an extern declaration for ap_sys_siglist (its exposed but not in any
221 Apache header file) to prevent a compile error on systems without
224 *) During auth requests, the subprocess_env table was left holding variables
225 sent to the auth FastCGI server (including REMOTE_PASSWD!) which means they
226 were passed to other processing phases (such as CGI/FastCGI). The
227 subprocess_env is now restored to its pre-auth condition.
229 *) Added a FastCGI Authorizer Role compatibility mode which implements the
230 specification to the tee. Use -compat arg with any of the Auth server
231 directives. This is intended for new and existing FastCGI authorizer
232 applications that require compatibility with other server implementations.
234 *) Fix logging from the process manager. Previously, all calls to the
235 logging routines used NULL for the server_rec. This works fine in a dev
236 environment, but results in most of these messages being tossed by
237 log_error_core() because DEFAULT_LOGLEVEL is too low. Odd logic.
239 *) Always setup fr->header so that in the event the FastCGI server
240 doesn't get a valid FastCGI protocol header over the wire, we
241 can still print our error message to the log from do_work().
243 *) Fix a compile error in open_connection_to_fs() on systems with TCP_NODELAY
244 defined (Don Locrasto [locrasto@iceminer.com]).
246 Changes with mod_fastcgi 2.2.0
248 *) Serious rewrite of mod_fastcgi.html. An example conf will have
249 to wait until another time - I'm burned out on docs updates.
251 *) Add to the mod_fastcgi to the server version string.
253 *) SCO doesn't like the const in the arg to inet_addr() and
256 *) Use Apache's NET_SIZE_T to get the correct arg type for getsockopt().
258 *) Deleted redundant header files from fcgi.h.
260 *) Full path names are no longer required for directives which take
261 file names as arguments. Like other Apache directives you can
262 specify paths from server_root (don't start with a "/").
264 *) The fcgiKillMgr is gone. This intermediate process sat between
265 the Apache parent and the FastCGI Process Manager under Apache
266 1.2. The process manager tries to name itself fcgi_pm (it used
269 *) Authentication, Authorization, and Access phases are now
270 supported. I've bent the FastCgi spec a bit: as many of the standard
271 environment variables (that were easy) are sent (the spec says don't
272 send some by name), all headers (except Status) sent by the auth
273 FastCgi server are passed to subprocesses (CGI/FastCGI invocations)
274 as environment variables rather than just those prefixed by
275 "Variable-". Custom responses for auth failures aren't supported
276 (yet.. we'll see what the demand for this is like). In addition to
277 the FastCGI protocol defined environment variable "FCGI_ROLE" being
278 set to "AUTHORIZER" an environment variable "FCGI_APACHE_ROLE" is
279 set indicating which of the three phases is being processed. See
280 the docs for info on the new per-directory directives:
282 FastCgiAuthenticator, FastCgiAuthenticatorAuthoritative,
283 FastCgiAuthorizer, FastCgiAuthorizerAuthoritative,
284 FastCgiAccessChecker, FastCgiAccessCheckerAuthoritative
286 *) The code has been broken into logical chunks making figuring it
287 out much easier. All of the #defines a user may want to change are
288 now in mod_fastcgi.h.
290 *) All the logging has been converted to the "supported" Apache
291 logging routines, ap_log_error() and ap_log_rerror(). Log entries
292 for request specific errors should now be directed to the correct
293 server errorlog (if your running more than one). Some attempt has
294 been made at setting appropriate log levels (comments of course are
295 welcome) and printing errno info when its of value. All of log
296 messages now have "FastCGI" in them (nice for grep).
298 *) All of the calls to Apache routines have been updated to the ap_
299 convention, eliminating the need for the compat header.
301 *) By default we no longer flush every piece of data we get from the
302 FastCGI server to the client. This allows the FastCGI server to
303 release without having to wait for the client flush to complete.
304 The old behaviour can revived with the -flush argument to AppClass.
306 *) I renamed some of the directives for consistency (the old names
309 AppClass -> FastCgiServer,
310 FCGIConfig -> FastCgiConfig,
311 ExternalAppClass -> FastCgiExternalServer
313 *) Directive arguments are no longer case sensitive.
315 *) The module now uses hard_timeout() rather than soft_timeout. This
316 means that the module will longjump out of the request if the
317 drop-dead timeout expires (set with Apache's Timeout directive, the
318 default is 5min) or if the client closes the connection (SIGPIPE).
319 This is more typical for a Apache modules and I'm not convinced we
320 properly handle all of the error cases well enough to use
321 soft_timeout (I can think of one place we don't anyway). This means
322 your FastCGI application can see SIGPIPE.
324 *) SIGPIPE is now ignored by default in FastCGI servers spawned by
325 Apache. Without this the default behaviour is exit().
327 *) It should (I think) run now under DSO now.
329 *) The bug which prevented sending of binary data in 2.1b1 is fixed.
331 *) The broken -initial-env bug ins 2.1b1 is fixed.
333 *) Maybe some other stuff.
337 *** Well there's a big gap here. Maybe I'll go back and extract the CVS
338 commit notes and stick 'em in here.. when I get some free time. ;)
343 *** Originally from docs/README..
345 Apache[X1.03.01]/mod_fastcgi[02.00.05] 19970909 unsupported
346 From: Stanley Gambarin <stanleyg@cs.bu.edu>
348 *) Yet more changes to the source distribution. Separated out
349 the Tcl dynamic string and buffer libraries into separate files.
350 Did the same for the OS library, however, it is not really an
351 abstraction... more like a bunch of wrappers. Added to DEVNOTES
352 a note about the problem of Apache not allowing including header
355 Apache[X1.03.01]/mod_fastcgi[02.00.04] 19970908 unsupported
356 From: Stanley Gambarin <stanleyg@cs.bu.edu>
358 *) More changes to the installation scripts. Theoretically we
359 should be able to to use the same Makefile/installation script
360 for both Apache 1.2.x and 1.3.x. Need some sort of the abstraction
361 layer to use the same source code for both 1.2.x and 1.3.x sources.
363 Apache[X1.03.01]/mod_fastcgi[02.00.03] 19970905 unsupported
364 From: Stanley Gambarin <stanleyg@cs.bu.edu>
366 *) Created Makefile to compile the module into the archive library,
367 just like the proxy module. Also initial draft of the installation
368 script, which should work for both 1.2 and 1.3 sources of Apache.
370 Apache[01.02.04]/mod_fastcgi[02.00.02] 19970903 unsupported
371 From: Stanley Gambarin <stanleyg@cs.bu.edu>
373 *) More source reorg. Separated out some header files. Need much
374 more work on this. Updated the TODO file. Added fcgivers.h and
375 CHANGES file to track the history.
377 Apache[01.02.04]/mod_fastcgi[02.00.01] 19970902 unsupported
378 From: Stanley Gambarin <stanleyg@cs.bu.edu>
380 *) Source reorganization. This is done to provide a basis for later
381 functionality implementation. See TODO file for more information
382 of what possible future enhancements are possible/desired.
384 Apache[01.02.04]/mod_fastcgi[02.00.00] 19970902 unsupported
385 From: David MacKenzie <djm@va.pubnix.com>
387 *) Create the "dynamic" dir and "mbox" as the user specified in the
388 User and Group directives instead of as root. They are created with
389 restrictive permissions, and then the module checks to see whether the
390 user specified in the User and Group directives has read, write, and
391 execute permissions on them. Those permissions had been explicitly
392 denied by creating them as "root".
394 *) Some documentation was garbled or incomplete.
396 *) When a FastCGI application can't be execl'd, the code used a
397 value of errno that may have been stomped on by intervening system
398 calls. That happened on BSD/OS 2.1, where the error_log was
399 reporting errno=25 (NOTTY) instead of the correct errno=13 (EACCES).
400 An intervening stdio call was setting errno as a side-effect, which
401 it has the right to do.
403 *) Erroneous fprintf arguments, a missing return value, an unused
404 function, missing declarations, and other problems detected by gcc
407 *) Duplicated code to create the lock file name merged into a single
408 function, closing several memory leaks.
410 *) FCGIConfig -minProcesses didn't allow a value of 0, so the last
411 FCGI app in a given dynamic class would live forever, even if it
412 hadn't been requested for weeks.
414 *) The test for whether to keep looking for dynamic app victims to
415 kill was backwards, so no dynamic apps were ever selected as
418 *) The killInterval and updateInterval were being ignored; the
419 sigsuspend() forced recalculations only when a child died,
420 instead of at the intervals specified.
422 *) When a dynamic app couldn't be started, the program SEGV'd when
423 trying to free the ipcAddrPtr twice.
425 *) Remove the dead lock file and socket when cleaning up after a
426 server whose last child has been killed, so the FastCGIHandler
427 isn't fooled into thinking there is still a process serving
430 *) Make file locking robust in the presence of signals.
432 *) Rename some badly named variables.
434 *) Fix typos in many comments.
436 *) Fix some memory and file descriptor leaks.
438 *) Make the blocking kill of a server closer to working.
440 *) More fixes to calculations in dynamic application management.
443 *** Originally from docs/README.OMI..
445 What's New: Version 2.0b1, 16 Apr 1997
447 *) Implemented a mechanism by which FastCGI applications are started by
448 the web server on the first request and continue running. Also, a
449 heuristics have been implemented to allow for dynamic killing of the
450 running FastCGI apps. The configuration options are supplied via
451 the new FCGIConfig directive, that is described in the
454 *) When performing internal redirect, since the original request's body
455 has already been read, do not allow the redirected request to think
456 that is has one [body].
458 *) More conditional compilation for OS/2.
460 *) Fixed occassional "Assertion failed: len > 0, file mod_fastcgi.c"
462 *) Fixed "failed assertion `count >= 0 && count <= bufPtr->length'"
464 *) Added bflush() call after bwrite() in DrainReqOutBuf function to
465 force some output to be written in the period of inactivity. This
466 circumvents the Apache's buffering during the server-push.
468 *) Added a header-parser function placeholder for compatibility with an
471 *) Additional checks have been placed with regard to permissions of the
472 FastCGI processes. These include the existence of the file,
473 ExecCGI and IncludesNOEXEC checks, disallowing of nph- scripts, etc.
475 *) Implemented a restart cleanup, so that no parentlesses FastCGI
476 applications are left after Apache is restarted and/or terminated.
477 This was accomplished via two independent mechanisms, where the
478 first one prevents a start of the process manager on the first
479 reading of the configuration files. The second mechanism deals away
480 from Apache's implementation of fork() (via spawn_child()) and
481 implements its own forking, thus removing any dependencies on the
482 Apache to cleanup processes during termination phase.
484 *) Removed the definition of the Sigfunc, as Apache 2.0b10 defines it
485 itself in the file conf.h
487 What's New: Version 1.4.3, 15 Jan 1997
489 *) Fixed compilation warnings for various platforms, as well as
490 conditional compilation for OS/2.
492 What's New: Version 1.4.2, 12 Dec 1996
494 *) mod_fastcgi.c is ported to Apache 1.2b. Any further development
495 will proceed under this version of Apache web server.
497 *) As the result of porting, the "include virtual" construct of SSI
498 will now work correctly using either <Location> or <Directory>
501 What's New: Version 1.4.1, 4 Dec 1996
503 *) Checks have been removed from the ScanCGIHeaders that provided
504 for the presence of both Status and Location headers as being
505 an error. Contradictory to CGI/1.1 Internet Draft, both of
506 these headers are used by the current CGI applications.
508 What's New: Version 1.4, 22 Nov 1996
510 *) Added the -port option to AppClass, allowing TCP/IP communication.
511 Added the -socket option to AppClass, allowing Unix domain
512 communication via a configurable pathname.
514 *) Added the ExternalAppClass directive, allowing TCP/IP
515 communication with remote FastCGI applications.
517 *) The handler had its own code for generating HTTP response headers;
518 now it uses Apache's. This reduces the size of the module.
519 More importantly, it fixes the bug in which "include virtual"
520 sees the HTTP response headers.
522 *) The response header parser performed very little checking.
523 Now the parser enforces the guidelines in the CGI/1.1
524 Internet-Draft: Status and Location are mutually exclusive,
525 Location can only be a response to GET or HEAD,
526 CGI response headers can't be repeated, etc. (The CGI response
527 headers are the ones the handler interprets: Status, Location,
530 *) The response header parser used to miss CGI headers
531 with no whitespace after the colon, e.g. "Status:200 OK".
533 *) The response header parser sometimes interpreted the first line
534 of content as an RFC822 continuation line.
536 *) The handler implemented a nonstandard version of Location
537 which never used internal redirects. The handler
538 now attempts to implement Location as specified in the
539 CGI/1.1 Internet-Draft. The module documentation explains
542 *) Error log entries from the response header parser were
543 pretty uninformative; they are better now. If a header is
544 malformed, the log entry includes it. If the headers are
545 unterminated rather than malformed, the log entry says that,
546 and says how many bytes were received from the app.
548 *) When the application manager forked a new process, and that
549 process ran into trouble before executing the first instruction
550 of application code, the process used to exit with status =
551 errno. This made certain configuration problems (e.g. incorrect
552 file permissions when server parent is root) quite difficult
553 to pin down. Now the failing child process opens up the error log
554 and writes an informative entry before exiting.
556 *) The module now correctly handles a slash at the end of the
557 DocumentRoot directive. This was a one-line fix.
559 What's New: Version 1.3.3, 17 Oct 1996
561 *) The module now registers its request handler under the name
562 fastcgi-script in addition to the name application/x-httpd-fcgi.
563 This was a one-line addition to the module, but had quite
564 a large impact on the documentation and sample configuration.
566 What's New: Version 1.3.2, 27 Sept 1996
568 *) On some systems (SunOS, Linux?), fopen for append has a bug
569 that strikes when two processes append to the same file.
570 This bug causes the process manager to corrupt the error log.
571 Work around the bug by calling open, then fdopen. A patch
572 to Apache 1.1.1 is also required, as described in mod_fastcgi.html.
573 (Reported by Bob Ramstad.)
575 What's New: Version 1.3.1, 17 Sept 1996
577 *) On Linux and SunOS, the Apache default user_id (-1) is not a
578 legal value of uid_t, so cast it. (Reported by Bob Ramstad,
579 Scott Langley, others.)
581 *) On some systems (Linux, some Solaris?, Irix?), connect requires
582 write access to a Unix Domain socket, so provide it. (Reported
583 by Scott Langley, freeform@wired.com, others.)
585 *) If you hit a .fcg file, but there's no AppClass defined, the
586 error should be NOT_FOUND instead of SERVER_ERROR. (Reported
589 What's New: Version 1.3, 4 Sept 1996
591 *) Module sometimes busy-waited in FastCgiDoWork. Fixed.
593 *) Module violated the Apache buff abstraction.
594 Fixed. Might now work with SSL (not tested).
595 As part of the fix, eliminated support for
596 Apache 1.0x versions.
598 *) Module failed to chmod the Unix domain listening sockets
599 it created, so protections were set according to the current
602 *) Module forked too many process manager processes: One
603 per AppClass. When the Apache parent ran as root, these
604 process manager processes ran as root. New process
605 manager is a single process (only started if AppClass is
606 used) and runs with same privileges as other children.
607 The new process manager doesn't do any polling, so
608 there's less system overhead than before.
610 *) AppClass insisted on getting at least two arguments. Fixed.
612 *) Module used setjmp/longjmp, causing compiler warnings
613 on some platforms (e.g. Linux.) The module no longer
616 *) Module created listening sockets in /tmp, where they were
617 sometimes wiped out by cleanup scripts. Added FastCgiIpcDir
618 directive to give control over the location of listening
621 *) Module wrote error log entries without a timestamp. Fixed.
623 *) AppClass directive wrote error messages to stderr in addition
624 to returning a char * message to the Apache core. Fixed.
626 What's New: Version 1.2, 3 June 1996
628 *) Ported from Apache-1.0-based code to Apache-1.1b2 internals by
631 Add version string: APACHEVERSION macro to mod_fastcgi.c
633 *) chown FastCGI socket to user_id and setuid to user_id for app
636 *) Modify GetFromStream() by having it call an OS dependent
637 function GetStreamSize(FILE *) which uses FILE internal data member.
638 Linux users might need to modify GetStreamSize().
640 What's New: Version 1.1, 10 May 1996
642 *) If you specify a non-existent executable in the AppClass directive,
643 or if the file exists and it does not have execute permission you
644 get a constant stream of error messages telling you that "program
645 terminated due to a signal".
647 *) The mod_fastcgi module should use the standard Apache error logging
648 facility instead of writing to stderr.
650 Version 1.0, 30 April 1996