Re: apache segfaults on mod_perl dlclose
On Tue, 14 Dec 1999, Saar Picker wrote: Hrm. Does perl use 0x as a marker? yes, see malloc.c I thought once an sv was freed it gets a null value or something. sure, but with Perls before 5.005_60-ish, Perl's malloc overrides system malloc/free *everywhere*, not just for Perl data, but also for other C code using malloc/free in the same runtime In any case, this shouldn't be happening due to any strictly perl modules I'm loading. true, but pure Perl can't be ruled out as triggering such a bug. it might help if you can run your server under gdb and post a stacktrace after this core dump.
Re: troubleshooting seg fault/child exit
What I find odd is that all of the sudden people seem to be coming out of the woodwork about this stuff. Whe I searched through mailing list archives, some of them going back years, I found something odd... Many people were talking about this same problem between December of last year and January of this year. No one found a satisfactory solution (at least, I didn't see one). Some blamed it on pre-loading CGI.pm in startup.perl, while others blamed it on having a PerlFreshRestart On in the Apache confguration scripts. This seems to be a pretty big deal, if so many people are complaining under so many configurations, and apparently it has gone on for over a year. core dumps can happen for many different reasons. why there are so many these past couple of days, I don't know, but doubt they are all triggered by the same thing. if you could follow suit and post a stacktrace from gdb, that would help.
Re: Problems with dynamically-loaded perl modules under Solaris2.5.1
are you still stuck on this? did you link mod_perl static or dso? if dso, try static. you can also try configuring Perl with -Dusemymalloc=n, but that comes with a large performance hit. I've never had a problem with solaris and any combo of mod_perl/Perl On Mon, 1 Nov 1999, Dan Rench wrote: I did some experimenting with my Solaris2.5.1/mod_perl/Data::Dumper/Storable problem that I wrote about last week. A grid: | mod_perl 1.21 | 1.21-dev (19991101174047) -+---+-- perl5.004_05 | works fine| didn't try perl5.005_03 | doesn't work | doesn't work perl5.005_62 | won't compile | works fine! All versions of perl compiled with all solaris hints. So is this a 5.005_03-specific bug? Any word on 5.005_04?
RE: no luck uploading
On Mon, 1 Nov 1999, Tubbs, Derric L wrote:
Well, I finally got it to work with Apache::Request. I had to do "$args =
Apache::Request-new($r);" and "$file = $args-upload;" in the handler sub
and then pass $args and $file to any other subs that needed it. Maybe this
is normal but if so I was unaware of it. I was doing all of the above
within a different sub that gets called from handler.
BTW, Can I do a conditional on the creation of $file that would return false
if there was no file uploaded???
just test for -filename:
if (my $upload = $apr-upload) {
my $filename = $upload-filename;
unless ($filename) {
print "no file specified";
return;
}
...
Re: PerlFixupHandler vs. PerlHandler
On Mon, 1 Nov 1999, James G Smith wrote:
I have some code working, but I'm curious as to why it has to work in the
way it does. The code:
package My::TAMU;
sub handler ($$) {
my $class = shift;
my $r = shift;
if($r-current_callback eq "PerlHandler" || # this one makes sense
$r-current_callback eq "PerlFixupHandler") { # but why this one?
$r-current_callback is store in a global, set to PerlHandler when
entered, but then the setup of %ENV does a subrequest, and sets the global
to PerlFixupHandler in doing so. this bug will be fixed in 1.22
Re: setting cookies?
On Mon, 1 Nov 1999, Robin Berjon wrote: At 11:44 01/11/1999 -0800, Doug MacEachern wrote: On Mon, 1 Nov 1999, Wyman Eric Miles wrote: The next question is, when the cookie expires 2 hours later, the initial SecurID user/password has long since expired. How do I cause the module to force the basic auth dialogs again? $r - note_basic_auth_failure; return AUTH_REQUIRED; Doesn't seem to work. that's cause netscape, ie, etc, cache basic credentials. so even though a 401 code is sent to the client, they just reuse the existing username/password in memory. I've never tried this but doesn't sending two 401s in a row for the same document have the auth popup appear again ? yeah, except that clicking 'cancel' causes the browser to send the cached credentials. at least, that was my experience last I tried.
Re: suggestion about PerlRun
thanks, I'll add this for 1.22.
On Tue, 2 Nov 1999, Yasushi Nakajima wrote:
Doug the handler() in Apache::PerlRun was not originally indended to be
Doug subclassed. however, if somebody submits a patch to make is
Doug subclass-able that doesn't break the way Apache::PerlRun currently works,
Doug that would be fine.
Following modified code works itself and as derived module on my
Apache/1.3.9 (Unix) mod_perl/1.21.
259,261c259,267
sub handler {
my $r = shift;
my $pr = Apache::PerlRun-new($r);
---
sub handler($$) {
my($class, $r);
if( @_ = 2 ) {
($class, $r) = (shift, shift);
} else {
($class, $r) = (__PACKAGE__, shift);
}
my $pr = $class-new($r);
I copied this code from RegistryNG.pm.
Sey Nakajima [EMAIL PROTECTED]
Kyoto, Japan
Re: referenced symbol not found error
you need the cvs snapshot of mod_perl to use 5.005_62+, or wait for 1.22 On Tue, 2 Nov 1999, Arkadiy Goykhberg wrote: *This message was transferred with a trial version of CommuniGate(tm) Pro* Hello, I'm trying to compile mod_perl-1.21 as DSO module for apache version 1.3.9 on Solaris 2.6 using gcc version 2.8.1 and perl version 5.00562. Right now I'am using out the box vanilla configuration of apache. Everything compiles ok, but then I try to start apache, I get the following error: Syntax error on line 224 of /usr/local/apache/conf/httpd.conf: Cannot load /usr/local/apache/libexec/libperl.so into server: ld.so.1: /usr/local/apache/bin/httpd: fatal: relocation error: file /usr/local/apache/libexec/libperl.so: symbol perl_eval_pv: referenced symbol not found /usr/local/apache/bin/apachectl start: httpd could not be started. Here is the output of make test: SNIP /usr/bin/perl t/TEST 0 Syntax error on line 1 of /local/packages/apache/mod_perl-1.21/t/conf/httpd.conf: Cannot load /local/packages/apache/mod_perl-1.21/t/../../apache_1.3.9/src/modules /perl/libperl.so into server: ld.so.1: ../apache_1.3.9/src/httpd: fatal: relocation error: file /local/packages/apache/mod_perl-1.21/t/../../apache_1.3.9/src/modules /perl/libperl.so: symbol perl_eval_pv: referenced symbol not found still waiting for server to warm upnot ok server failed to start! at t/TEST line 95. make: *** [run_tests] Error 146 /SNIP # perl -V Summary of my perl5 (revision 5.0 version 5 subversion 62) configuration: Platform: osname=solaris, osvers=2.6, archname=sun4-solaris uname='sunos arkadiy 5.6 generic_105181-05 sun4u sparc sunw,ultra-5_10 ' config_args='' hint=previous, useposix=true, d_sigaction=define usethreads=undef useperlio=undef d_sfio=undef use64bits=define usemultiplicity=undef Compiler: cc='gcc', optimize='-O', gccversion=2.8.1 cppflags='-D_LARGEFILE_SOURCE -D_FILE_OFFSET_BITS=64 -DUSE_LONG_LONG -DSOCKS -I/usr/local/include -DUSE_LONG_LONG' ccflags ='-D_LARGEFILE_SOURCE -D_FILE_OFFSET_BITS=64 -DUSE_LONG_LONG -DSOCKS -I/usr/local/include -DUSE_LONG_LONG' stdchar='unsigned char', d_stdstdio=define, usevfork=false intsize=4, longsize=4, ptrsize=4, doublesize=8 d_longlong=define, longlongsize=8, d_longdbl=define, longdblsize=16 alignbytes=8, usemymalloc=y, prototype=define Linker and Libraries: ld='gcc', ldflags =' -L/usr/local/lib' libpth=/usr/local/lib /lib /usr/lib /usr/ccs/lib libs=-lsocket -lnsl -ldl -lm -lc -lcrypt \lsec libc=, so=so, useshrplib=false, libperl=libperl.a Dynamic Linking: dlsrc=dl_dlopen.xs, dlext=so, d_dlsymun=undef, ccdlflags=' ' cccdlflags='-fPIC', lddlflags='-G -L/usr/local/lib' Characteristics of this binary (from libperl): Built under solaris Compiled at Oct 29 1999 13:01:49 @INC: /usr/local/lib/perl5/5.00562/sun4-solaris /usr/local/lib/perl5/5.00562 /usr/local/lib/site_perl/5.00562/sun4-solaris /usr/local/lib/site_perl . Have anybody seen this before, or better yet, found a solution for this problem? Thanks in advance. PS I apologize if this message was sent more then once.
Re: hostname fails under Apache::Registry, but not as CGI??
try Apache-request-get_server_name instead of hostname() it's cheaper too, since Sys::Hostname::hostname forks, at least once per-process. not so bad if you call it in the parent though (in a PerlRequire script), then all children have the cached results On Wed, 3 Nov 1999, Nick Urbanik wrote: Dear folks, When I run a CGI program as such, hostname works. When I run it under Aapche::Registry, it fails. Can anyone understand why? Sorry, I am new to all this. [error] Cannot get host name of local machine at /home/httpd/dbi/frame_control.cgi line 13 use Sys::Hostname; my $hostname = hostname(); my $URL = "http://$hostname/dbi/"; -- Nick Urbanik, Dept. of Electrical Communications Engineering Hong Kong Institute of Vocational Education (Tsing Yi) email: [EMAIL PROTECTED], [EMAIL PROTECTED] Tel: (852) 2436 8660, (825) 2436 8674 Fax: (852) 2436 8643 pgp ID: 7529555D fingerprint: 53 B6 6D 73 52 EE 1F EE EC F8 21 98 45 1C 23 7B
Re: Problems building
On 3 Nov 1999, Greg Stark wrote: I'm trying to update to 1.3.9+1.21 but I'm having various problems: 1) The makefile seems to try to run ../apaci even though perl is two levels deep, so it can't find ../apaci, it would have to be ../../apaci === src/modules/perl gcc -I. -I../../include -I../../include/regex -I../../include/.././os/unix -DLINUX=2 -DUSE_HSREGEX -DUSE_EXPAT -I../lib/expat-lite `../apaci` `/data/app/perl/bin/perl -MExtUtils::Embed -e ccopts` -DNO_PERL_RESTART -DMOD_PERL_VERSION=\"1.21\" -I./os/unix -I../../include -c mod_perl_opmask.c -o mod_perl_opmask.o /bin/sh: ../apaci: No such file or directory 2) when it tries to link it gets pages and pages of these errors: gcc -DLINUX=2 -DUSE_HSREGEX -DUSE_EXPAT -I./lib/expat-lite `./apaci`\ -o httpd buildmark.o modules.o modules/perl/libperl.a modules/standard/libstandard.a main/libmain.a ./os/unix/libos.a ap/libap.a regex/libregex.a lib/expat-lite/libexpat.a -lm -lcrypt modules/perl/libperl.a(mod_perl.o): In function `perl_shutdown': did you ever get this sorted out? this link line isn't even close to correct, how did you go about building, Makefile.PL options, etc??
Re: mod_perl on Apache 2.0
On Wed, 3 Nov 1999, Jeffrey Baker wrote: There was a short discussion a while ago about getting mod_perl working with Apache 2.0. Since Apache 2.0 can actually be built and run on a few platforms now, I think it is worth taking a lot at this for real. I started to fiddle with 5.005_63 and perl_clone(), running concurrent interpreters in seperate threads. it didn't core dump, like old MULTIPLICTY Perls would, but didn't work quite right either. sarathy says it's not quite ready yet, but it looks very promising from what I've seen so far. I hope to start looking into 2.0 things once I catch up here, and get 1.22 released.
Re: make fails
did you ever get this sorted out? looks like an rpm version of Perl maybe? have you tried configuring/build Perl from source? On Thu, 4 Nov 1999, Eva Neuberger wrote: Hallo, I'm trying to install mod_perl_1.21 with apache 1.36 perl 5.005_03 and Linux Kernel 2.2.10 I follow the instructions in INSTALL.apaci perl Makefile.PL APACHE_SRC=../apache_1.3.6/src DO_HTTPD=1 PREP_HTTPD=1 USE_APACI=1 EVERYTHING=1 Everything is ok, then typing make . cc -c -Dbool=char -DHAS_BOOL -I/usr/local/include -O2 -pipe -DVERSION=\"1.00\" -DXS_VERSION=\"1.00\" -fpic -I/usr/lib/perl5/5.00503/i586-linux/CORE Leak.c In file included from Leak.xs:12: /usr/lib/perl5/5.00503/i586-linux/CORE/perl.h:2546: redefinition of `union semun' make[1]: *** [Leak.o] Error 1 make[1]: Leaving directory`/usr/src/mod_perl-1.21/Leak' make: *** [subdirs] Error 2 and with perl Makefile.PL APACHE_SRC=../apache_1.3.6/src APACHE_PREFIX=/usr/local/apache DO_HTTPD=1 USE_APACI=1 EVERYTHING=1 I've got cc -c -I../.. -I/usr/lib/perl5/5.00503/i586-linux/CORE -I../../os/unix -I../../include -DLINUX=2 -DMOD_PERL -DUSE_PERL_SSI -Dbool=char -DHAS_BOOL -I/usr/local/include -DUSE_HSREGEX `../../apaci` mod_include.c In file included from ../../modules/perl/mod_perl.h:41, from mod_include.c:85: /usr/lib/perl5/5.00503/i586-linux/CORE/perl.h:2546: redefinition of `union semun' make[5]: *** [mod_include.o] Error 1 make[4]: *** [all] Error 1 make[3]: *** [subdirs] Error 1 make[3]:Leaving directory`/usr/src/apache_1.3.6/src' make[2]: ***[build-std] Error 2 make[2]: Leaving directory`/usr/src/apache_1.3.6' make[1]: ***[build] Error 2 make[1]: Leaving directory`/usr/src/apache_1.3.6' make: *** [apaci_httpd] Error 2 the same "redefinition of union semun" I've no idea, what's wrong. Eva Neuberger
Re: Unresolved References
On Mon, 8 Nov 1999, Brad Bonkoski wrote: This message was sent from Geocrawler.com by "Brad Bonkoski" [EMAIL PROTECTED] Be sure to reply to that address. Hello, I am trying to build mod_perl and PHP together with apache and this is a sample of what I am getting: === src/modules/perl === src/modules gcc -c -I./os/unix -I./include -DLINUX=2 -DUSE_HSREGEX -DUSE_EXPAT -I./lib/ex pat-lite `./apaci` modules.c gcc -c -I./os/unix -I./include -DLINUX=2 -DUSE_HSREGEX -DUSE_EXPAT -I./lib/ex pat-lite `./apaci` buildmark.c gcc -DLINUX=2 -DUSE_HSREGEX -DUSE_EXPAT -I./lib/expat-lite `./apaci` -rdynami c \ -o httpd buildmark.o modules.o modules/perl/libperl.a modules/example/libe xample.a modules/php3/libphp3.a modules/experimental/libexperimental.a modules/s tandard/libstandard.a modules/proxy/libproxy.a main/libmain.a ./os/unix/libos.a ap/libap.a regex/libregex.a lib/expat-lite/libexpat.a modules/perl/libperl.a(mod_perl.o): In function `perl_shutdown': you're missing quite a bit here in the link line. I know folks have posted how to build Apache+mod_php+mod_perl, maybe Stas added it to the guide?
Re: My doesn't it start ?
try adding: PerlModule Apache to your httpd.conf On Tue, 9 Nov 1999, Shay Mandel wrote: Hi all, It is my first time I am building the apache with the mod_perl enabled. I have installed it as static module. Then I compiled the apache (1.3.9). Everything passed ok (I skipped the make test because I don't have the LWP package). Now, when I start the apache I get this in the error_log: Apache.pm failed to load!. [Tue Nov 9 17:10:46 1999] [warn] pid file /www/apache/v1.3.9/logs/httpd.pid overwritten -- Unclean shutdown of previous Apache run? Apache.pm failed to load!. What should I do ? should I add the addModule directive to the httpd.conf ? Should I alter the @INC variable ? when I run the Apache.pm by myself I see that it has a problem that it cannot find modules in the @INC - how do I change this environment variable ? which of the (three) Apache.pm I have is the one that the apache is running ? I read all the docs,readme's, Install's, but still I have all these questions - please help.
Re: mod_perl crash
more details please, Makefile.PL options, etc. On Thu, 11 Nov 1999, Dominic A. V. Amann wrote: I am trying to build apache 1.3.9 (or 1.3.anything) with mod_perl. It builds fine without, on libc-2.0.7 platform, using gcc 2.7.2.3 I have tried many options in the APACI make process, all have same result (segmentation fault). This is the gdb backtrace from running the mod-perl version: fido:/usr/src/mod_perl-1.21$ gdb ../apache_1.3.6/src/httpd GNU gdb 4.18 Copyright 1998 Free Software Foundation, Inc. GDB is free software, covered by the GNU General Public License, and you are welcome to change it and/or distribute copies of it under certain conditions. Type "show copying" to see the conditions. There is absolutely no warranty for GDB. Type "show warranty" for details. This GDB was configured as "i686-pc-linux-gnulibc1"... (gdb) run -X -f `pwd`/t/conf/httpd.conf -d `pwd`/t Starting program: /usr/src/mod_perl-1.21/../apache_1.3.6/src/httpd -X -f `pwd`/t/conf/httpd.conf -d `pwd`/t Program received signal SIGSEGV, Segmentation fault. 0x80c1e88 in perl_parse () (gdb) bt #0 0x80c1e88 in perl_parse () (gdb) Please advise. -- Dominic Amann, http://www.interlog.com/~damann/ Linux Based Solutions Ltd. Toronto, ON, M3J 1G8, Canada Tel: (416) 638-8649, Fax: (416) 630-1584
Re: Apache::Registry and 304 status code in log file
see Apache::RedirectLogFix On Thu, 11 Nov 1999, Bill Moseley wrote: In an Apache::Registry script I'm print()ing Status: 304 Not Modified when the client sends an If-Modified-Since: header. The 304 error is returned to the client properly, but the log file shows a 200 status. Can I make Apache log the access as a 304 with my Apache::Registry script? The same script runs someplace else as mod_cgi. Thanks, BTW -- In perldoc Apache::Registry I'm sure this is known, but... Location /perl SetHandler perl-script PerlHandler Apache::Registry Options ExecCGI ... /Directory -- shouldn't this be /Location Bill Moseley mailto:[EMAIL PROTECTED]
Re: Seg fault on fork w/ PerlModule
can you get a stacktrace with gdb? see the SUPPORT doc for hints.
On Tue, 16 Nov 1999, Tony Demark wrote:
I have compiled the following:
Apache 1.3.9
mod_perl 1.21 (PERL_AUTHEN PERL_LOG PERL_STACKED_HANDLERS)
on:
Ultra 10
Solaris 7
with:
gcc 2.95.2 19991024
perl 5.005_03
When starting an httpd process with a 'PerlModule' defined in the conf file,
the process will seg fault when it attempts to fork:
open("/opt/apache/logs/access_log", O_WRONLY|O_APPEND|O_CREAT, 0644) = 3
fcntl(3, F_DUPFD, 0x000F) = 16
close(3)= 0
chdir("/") = 0
fork() = 16024
Incurred fault #6, FLTBOUNDS %pc = 0xD002
siginfo: SIGSEGV SEGV_MAPERR addr=0xD002
Received signal #11, SIGSEGV [default]
siginfo: SIGSEGV SEGV_MAPERR addr=0xD002
*** process killed ***
However, comment out the 'PerlModule' statement, and the process forks fine
and exits:
open("/opt/apache/logs/access_log", O_WRONLY|O_APPEND|O_CREAT, 0644) = 3
fcntl(3, F_DUPFD, 0x000F) = 16
close(3)= 0
chdir("/") = 0
fork() = 16030
llseek(0, 0, SEEK_CUR) = 101628
_exit(0)
I didn't see anything about this on the maillist... get anyone give me an
idea of where I should at least start to track down this problem?
Thanks,
- Tony
Re: module Safe ?
On Thu, 18 Nov 1999, Ekaterina Ivannikova wrote: Hi All! I should have mentioned that there is a Safe.pm 2.06 in perl5 directory tree. But after browsing modperl archive I was under the impression that a special Apache::Safe was needed due to some subtle problems. Is it not so ? It seems there used to be a PerlSafe directive at some time but no mention of it in the current docs. yes, there was indeed an Apache::Safe at one time, but it was never finished, too some thorny problems. there was an experimental PerlOpMask directive added in 1.15_01 over a year ago (see Changes), derived from Safe/Opcode.pm logic, but I haven't looked at that for quite a while.
Re: multipart POST problems
On Thu, 18 Nov 1999, John S. Evans wrote: I've been attempting to write a perl module that handles POSTs of type multipart/form-data, and have been having a rough time. I'm using Apache::Request to process the request. I have dumped the content-type of the incoming request, and verified that it's "multipart/form-data". I can use param() to get the parameters, but I can't seem to use upload() to access the blocks directly. I've included the code from my test handler below. Basically it attempts to access a large parameter MIME block two ways - using param() and using upload(). The param() version works fine, but the upload() version can't find the block. Any clues? The only thing that I can think of is that for this test case, the MIME type of the "message" block is text/plain (it's in a TEXTAREA field, for testing purposes). Is it possible that Apache::Request will not allow me to process "normal" form fields with upload()? does the libapreq-x.xx/eg/perl/file_upload.pl script work for you?
Re: Apache::Cookie confusion
On Fri, 19 Nov 1999, John Siracusa wrote: Apache::Cookie seems to have two different interfaces...or maybe there are two different distributions of Apache::Cookie? Whatever it is, the interface seems different on two machines here at work. One has 5.004 and one has 5.005, but that shouldn't change the Apache::Cookie interface should it? there is an Apache::Cookie in the libapreq bundle, which is supported here. I'm not sure how another module named Apache::Cookie made it onto CPAN.
Re: Access handlers for CGI's...
why don't you just use a PerlAuthenHandler? see chapter 6 online at
modperl.com
On Tue, 23 Nov 1999, Trevor Phillips wrote:
I've written an Authentication Handler using the Access phase, where someone
must validate themselves before accessing a resource. The way I've done it, if
the client is not validated, it changes the PerlHandler to reference a "Login"
subroutine, which produces the login page. This works fine, although currently
the protected documents are also handled by the module (to append a footer
relating to their session), and so the normal handler is already associated
with "perl-script".
I'm now encountering a problem where I'm trying to use the same module to
protect simple CGI's, where I'm trying to avoid requiring any custom code
additions to the CGI's (and would also work with binary CGI's). The problem is
the normal content handler. Is there a way to leave the handler as the normal
CGI handler, but in an instance where a login is required, override the handler
and enable the perl-script handler??
ie;
If authenticated
{
handle CGI as per usual
}
else
{
replace handler with custom login screen page
}
This is the code I've currently got, which alters the handlers:
$r-handler("perl-script");
$r-set_handlers(PerlHandler=[\LoginScreen]);
return OK;
As far as I can see, this should both change the handler to use the Perl
Script, AND define the perl handler as referencing a subroutine LoginScreen.
However, in practice the CGI is being executed as per usual (ie; the CGI
handler isn't being overridden by the mod_perl one).
Any ideas?
--
. Trevor Phillips - http://jurai.murdoch.edu.au/ .
: CWIS Technical Officer - [EMAIL PROTECTED] :
| IT Services - Murdoch University |
--- Member of the #SAS# #CFC#
| On nights such as this, evil deeds are done. And good deeds, of /
| course. But mostly evil, on the whole. /
\ -- (Terry Pratchett, Wyrd Sisters) /
Re: Limiting CPU (was Re: embperl pages and braindead sucking robots)
My CPU-based limiter is working quite nicely. It lets oodles of static pages be served, but if someone starts doing CPU intensive stuff, they get booted for hogging my server machine. The nice thing is that I return a standard "503" error including a "retry-after", so if it is a legitimate mirroring program, it'll know how to deal with the error. choice! Doug - one thing I noticed is that mod_cgi isn't charging the child-process time to the server anywhere between post-read-request and log phases. Does that mean there's no "wait" or "waitpid" until cleanup? it should be, mod_cgi waits for the child, parsing it's header output, etc. Also, Doug, can there be only one $r-cleanup_handler? I was getting intermittent results until I changed my -cleanup_handler into a push'ed loghandler. I also use -cleanup_handler in other modules, so I'm wondering if there's a conflict. you should be able to use any number of cleanup_handlers. do you have a small test case to reproduce the problem?
Re: VelociGen for Perl, Competetive Analysis?
Here's a snip from an email that I just receved from one of your co-workers, Alex Shah [EMAIL PROTECTED] in response to this same email: ---begin quote Why the comparison with mod_perl? This was part of the agreement we made with Sun in order to bundle our product with their web server. It was a strategic decision which step on mod_perl toes. Sun needed us to come up with a white paper to show how the new iPlanet web server was superior to the free Apache solution. ---end quote hmm, guess we should take this as a complement! I understand that you guys even use core mod_perl code in your product, and yet you have to resort to FUD aginst mod_perl to sell it. VelociGen was written completely independently of mod_perl - in fact mod_perl wasn't really around when we start writing the code for the Netscape and IIS versions of VelociGen. Our Apache port came out quite a bit later, and we didn't use any mod_perl code for that. Our architecture is quite a bit different, so not a lot of the code is applicable. We do use Apache::DBI, and possibly one or two other Perl modules, which could be classified as parts of mod_perl, but to say VelociGen uses core mod_perl code is entirely inaccurate. this isn't true. mod_perl was around for quite a while before velocigen. in fact, I downloaded the first version to see what is was all about and saw a copy-n-pasted file from perlembed.pod, the "Maintaining a persistent interpreter" section. that code and documentation was written (by me), based on mod_perl/Apache::Registry. I do agree that velocigen is quite different from mod_perl, but let's be honest here, mod_perl certainly helped to kick start your product.
Re: Altering handler not working...
the TypeHandler phase will override your changes, I think. try setting
$r-handler with a PerlTypeHandler, apache stops after the first one
returns OK, so mod_mime won't stomp your $r-handler settings in that
case.
On Thu, 25 Nov 1999, Trevor Phillips wrote:
I'm trying to write an access routine which requires altering the handler if
certain conditions are (not) met. There are a few interesting examples of this
in the "Apache Modules in Perl C" book, chapter 7, which do something similar
within the TransHandler and some other phases, but I'm trying to do it in the
AccessHandler phase.
Here's a simple test module:
package Access;
use Apache::Constants qw(:common);
sub handler
{
my $r = shift;
if ( -- some condition test -- )
{
$r-handler("perl-script");
$r-set_handlers(PerlHandler=[\SomeRoutine]);
return OK;
}
return DECLINED;
}
sub SomeRoutine
{
my $r = shift;
$r-content_type('text/html');
$r-send_http_header;
print "Some content...";
return OK;
}
So, basically, if a condition is met, then I return Access as OK, but I also
override whatever handler is there with a custom one.
The problem is, I cannot get this to work!! If a URI's handler is already
perl-script, then SomeRoutine is called, but it is NOT overriding other
handlers.
If I use this module as a TransHandler, then it DOES work correctly, but I
really need this to come in at the Access phase (as that's what it relates to).
Any ideas?
--
. Trevor Phillips - http://jurai.murdoch.edu.au/ .
: CWIS Technical Officer - [EMAIL PROTECTED] :
| IT Services - Murdoch University |
--- Member of the #SAS# #CFC#
| On nights such as this, evil deeds are done. And good deeds, of /
| course. But mostly evil, on the whole. /
\ -- (Terry Pratchett, Wyrd Sisters) /
Re: Release for 1.22?
On Fri, 26 Nov 1999, Ken Williams wrote: Hi, I've got an ISP running Apache/1.3.4 mod_perl/1.18, and I'm trying to get them to fix some of the broken functionality therein by upgrading to the latest versions. I really need the latest CVS snapshot so that Perl sections work correctly, but I doubt I'll be able to convince them to upgrade without a formal release. Any ideas about when that might happen? soonish, before the new year, I hope.
Re: Eagle Book - mod_hello.c, hello.pl :)
On Mon, 29 Nov 1999, Michael Dearman wrote: ... When looking at http_config.h at the handler_rec structure, the elements don't seem to match the way they're used in the example in the book. Going to the book site, the src's for the examples are NOT available. *shrug* all of the book source and then some is available from the site, there's a link on the homepage. it includes an apxs/Makefile for building the C modules.
Re: PerlLogHandler - bytes always zero for proxy requests
On Thu, 2 Dec 1999, Brian S. Craigie wrote: Hi Doug, Thanks and sorry if I looked impatient :-) nah, sorry if I looked like I thought you looked impatient ;-) I'm glad that this was spotted. I thought I was doing something wrong. I'm assuming you are referring to the cvs snapshot of mod_perl? It wasn't clear in my own mind where the problem lay, but I was pretty certain it wasn't in either of the DBI logger modules. yes. I'll check out the CVS. Many thanks. did it cure the problem for you?
Re: PerlLogHandler - bytes always zero for proxy requests
On Wed, 1 Dec 1999, Brian S. Craigie wrote: I presume from the lack of response that nobody knows the answer to this one? Or have I sent this to the wrong mailing list? Or is it a stupid question? or because you just posted on monday and some people have been extra super buzy with other things? have a look at the cvs snapshot and the Changes file, the problem was fixed a while back: if $r-bytes_sent has not been calculated, then do so before returning the value. thanks to Dirk Melchers for the spot
Re: mod_perl programming logic error
thanks Michael! I've committed your change to the cvs tree. actually, Eric Cholet did, mine failed, cvs rocks. need more coffee. cvs server: Up-to-date check failed for `Changes' cvs server: Up-to-date check failed for `src/modules/perl/Apache.xs' cvs [server aborted]: correct above errors first! cvs commit: saving log message in /tmp/cvs00820baa
Re: Intentional Premature Finish in an Apache Module
On Mon, 1 Nov 1999, John Siracusa wrote: On 11/1/99 2:32 AM, Doug MacEachern wrote: or use this undocumented feature: $Apache::ERRSV_CAN_BE_HTTP = 1; #set this anywhere, startup script is best die FORBIDDEN; mod_perl peeks at $@, if it's a 3 digit http status code, it propagates that value to Apache. Sounds great, but I'm hesitant to use "undocumented features." Will anything like this ever make it into the official spec? For now I'm using eval, I guess, but I'm all for official support for this functionality! :) Of course, the method could be a bit more friendly. I mean, no one likes seeing "SV" on the Perl side of things, even if it is just a variable name ;) well, it's only undocumented and awkward to enable because I wasn't sure how useful the feature would be. it's in the code base for when the need arises, like right now, so try it, if it's useful then it can be documented, etc.
Re: no luck uploading
On Mon, 1 Nov 1999, Tubbs, Derric L wrote: I've been trying to use the "file" input field to upload some files and I'm having no luck at all. I first tried using Apache::Request with no luck and the error message "Can't call method "filename" without a package or object reference at ..." but I had in fact done a $args = Apache::Request-new($r) and $file = Apache::Request-upload. I don't understand what you are trying to say, without seeing an example. try the libapreq-x.xx/eg/perl/file_upload.pl script
RE: mod-perl logo
On Mon, 1 Nov 1999, Kent, Mr. John wrote: Doug, I asked O'Reilly's permission to use the "camel and feather" without acknowledging them, and they said NO, see letter below from Edie Freedman. gad, I hate politics. Tim O'Reilly gave us permission to use the camel in the mod_perl logo two years ago. I'm can't remember if that means every site that uses the logo also needs to include the permission text like perl.apache.org does. I say mod-perl needs to adopt a non-proprietary logo. Wasn't there a page somewhere that had competing designs? Perhaps I can use one of them. perl.apache.org/logos/
Re: Using Apache::Util outside mod_perl
So you can't use Apache::Util unless you are running Apache. exactly. but it might be possible with the SHARED_CORE configure option, that builds libhttpd.so, I've never tried though.
Re: follow-up - no luck uploading
Our httpd's were spinning out of control on uploads until Doug sent us this patch. I don't know whether it made it into the distribution or not. yabut, that was only when netscape was trying to upload a directory name and lied about the content length. anyhow, the patch is in the distribution, just not released yet.
Re: PATH env garbled in mod_perl 1.21 patch
On 1 Nov 1999, Greg Stark wrote: Was this ever resolved? yes, fixed in cvs. I'm concerned because for Oracle the environment variable ORACLE_HOME is critical and I think the environment corruption explains some of our past database problems. PATH was the only corrupted variable, this was explained too, see the archive for details.
Re: setting cookies?
On Mon, 1 Nov 1999, Wyman Eric Miles wrote: I should have realized that. I managed to architect a kludge that sets the cookie but causes a fragment of the HTTP header to appear in the browser the first time. For now, I'm content with it. The next question is, when the cookie expires 2 hours later, the initial SecurID user/password has long since expired. How do I cause the module to force the basic auth dialogs again? $r - note_basic_auth_failure; return AUTH_REQUIRED; Doesn't seem to work. that's cause netscape, ie, etc, cache basic credentials. so even though a 401 code is sent to the client, they just reuse the existing username/password in memory.
Re: Make errors
On Fri, 15 Oct 1999, John Whitehead wrote: I can succesfully make Apache and perl with the following versions ; apache_1.2.6 perl5.004_05 when trying to make mod_perl-1.21 mod_perl no longer supports 1.2.x, try 1.3.9 instead.
Re: setting cookies?
On Mon, 18 Oct 1999, Wyman Eric Miles wrote: System: Solaris 2.6 Apache 1.3.9/mod_perl 1.19/gcc 2.8.1/perl-5.004.04 I'm using SecurID to authenticate for an Apache proxy server. I've written a little perl module that uses a username/tokencode returned by basic auth to validate a user and return a session cookie. The SecurID auth works fine and I'm able to generate a cookie for the user. The problem is, I can't get the module to return the cookie to the browser before the proxy request is completed. mod_proxy doesn't look at r-headers_out table, it just passing along the headers from the downstream server.
Re: mod_perl with APXS plus Raven equals segfault
try building Perl with Configure -Uusemymalloc, solaris+perl malloc don't seem to get along when mod_perl is a dso. you're better off to link static though, solaris system malloc is 25%-30% slower than using Perl's malloc On Mon, 18 Oct 1999, Steve Snodgrass wrote: I've been using mod_perl with Raven's SSL package for some time now, but I'm building a refresh of our environment with new versions of everything and I ran into trouble. I decided to use APXS this time instead of building mod_perl statically. Everything compiled and installed fine but Apache immediately segfaults on startup. The details: Sun Ultra Enterprise 3500 Solaris 7 (HW 5/99) Apache 1.3.9 (built from Raven pre-patched source) Raven SSL 1.4.1 mod_perl 1.21 perl 5.005_03 gcc 2.95.1 (regular Solaris ld, GNU ld is not even on the system) I guess I can go back to compiling mod_perl statically, but it would be nice to get this fixed. Any thoughts? Thanks. -- Steve "Pheran" Snodgrass * [EMAIL PROTECTED] * FORE Systems Unix Administrator Geek Code: GCS d? s: a- C++ US$ P+++ L+ w PS+ 5++ b++ DI+ D++ e++ r++ y+* "What to do I find it hard to know/The road I walk is not the one I chose" -Yes
Re: redirect after POST again?
if you read POST data, then redirect, you need to do this before the
redirect or apache will hang:
$r-method_number(M_GET);
$r-method('GET');
$r-headers_in-unset('Content-length');
On Wed, 20 Oct 1999, Ilya Obshadko wrote:
Hello,
Weird things: this simple plain code
$req-header_out('Location' = $ENV{SCRIPT_NAME});
$req-status(REDIRECT);
$req-send_http_header;
in POST script doesn't work with Netscape. It results in 'document
contains no data' message. If $req-send_http_header is omitted,
Netscape waits indefinitely until timeout. With IE, it's all ok.
Even more strange - some weeks ago I've installed a VERY similar
server with the same mod_perl apache, and all such scripts
worked fine. But don't advice to diff on all the config files -- it
would take too long ;)
Best regards,
Ilya mailto:[EMAIL PROTECTED]
Re: Problem at startup (Apache.pm)
try adding to httpd.conf: PerlModule Apache On Thu, 21 Oct 1999, Alessio Bragadini wrote: I am sorry to bother you with what seems a FAQ, but the solutions found on the list archive (do make install, check permissions, use PerlSetEnv PERL5LIB) are not working for me: after make make install I have the perl modules in /usr/local/apache+mod_perl/lib/site_perl/5.005/alpha-dec_osf put server startup fails with Apache.pm failed to load! Looks like I am missing sometimes, but I cannot find what, probabily because it's my firtst mod_perl build. Apache 1.3.9, mod_perl 1.21, perl 5.005_03, Digital Unix 4.0, Digital cc Thanks in advance -- Alessio F. Bragadini [EMAIL PROTECTED] APL Financial Serviceshttp://staff.dsnet.it/~alessio Nicosia, Cyprus phone: +357-2-750652 You are welcome, sir, to Cyprus. -- Shakespeare's "Othello"
Re: undefined reference problem
On Thu, 21 Oct 1999, Scott Anderson wrote: am having the following errors trying to compile apache with mod_perl my mod perl configuration script looks like this perl Makefile.PL \ APACHE_SRC=/usr/local/src/apache_1.3.9 \ NO_HTTPD=1 \ PREP_HTTPD=1 \ EVERYTHING=1 i answer yes when it asks about the src dir and no when it asks me if i want to build httpd. then i do make make install and everything seems to be ok then i goto the apache dir and use this configure script CC=cc \ ./configure \ --prefix=/usr/local/apache \ --enable-rule=SHARED_CORE \ --enable-module=so \ --enable-module=auth_dbm \ --enable-module=auth_db \ --enable-module=rewrite \ --enable-module=info \ --activate-module=src/modules/jserv/mod_jserv \ --activate-module=src/modules/php3/libphp3.a \ --activate-module=src/modules/perl/libperl.a you're better off to let Makefile.PL do all the work, you can pass those options to Apache via APACI_ARGS, see also: makepl_args.mod_perl
Re: Apache-read timed out?
sounds like somebody has already read the POST data before your code is
reached.
On Fri, 22 Oct 1999, Marler, Gordon wrote:
I'm running mod_perl 1.21, Apache 1.39 on Solaris 2.5.1.
I'm using the proxy example from pp. 374-381 of the "Writing Apache Modules"
book, and I've come up with a strange case in handling a POST request in the
following part of the code (explained on page 377). Here's my doctored
version of the code snippet:
# Copy POST data, if any
if ($r-method eq 'POST') {
$log-debug("Posting!");
my $len = $r-header_in('Content-length');
$log-debug("Got the content length for POSTing: $len");
my $buf;
$r-read($buf,$len);
$log-debug("Read the content for POSTing into a buffer");
$request-content($buf);
$log-debug("Copied POSTed data into the LWP request");
}
Turns out that this code hangs (just for this particular POST) at the
$r-read($buf,$len) for a few minutes, then I get the
"Apache-read timed out" message in the Apache error log.
Any ideas in what I could look for in the URI or the headers that might give
me a clue on why this is happening or how I could debug this? Since the
Content-length is greater than 0, I can't imagine why this would time out!
I'm sure this is a special case, since most POSTs work just fine with this
code.
Gordon Marler
ATT Wireless Services, WLTG
Unix System Architect
[EMAIL PROTECTED]
Phone: 425 702 2980
FAX: 425 702 2568
Re: Hanging process: detection and determination (was Re: Runawayprocesses)
I've checked the Apache::SIG and $r-connection-aborted, but is there a
way to "write" without actually writing, probably some control char will
do? Something like:
while(1){
$r-print("\0");
last if $r-connection-aborted;
$i++;
sleep (1);
}
have a look at what Eric recently added to the api:
=item $c-fileno( [$direction] )
Returns the client file descriptor. If $direction is 0, the input fd
is returned. If $direction is not null or ommitted, the output fd is
returned.
This can be used to detect client disconnect without doing any I/O,
e.g. using IO::Select.
See the output of strace, when I press Stop - it detects the SIGPIPE but
doesn't quit!
...
So Apache::SIG doesn't set correctly the mod_perl's default behaviour,
since when I add:
use Apache::SIG ();
Apache::SIG-set;
hmm, I thought we were done with SIGPIPE, the patch below will re-enable
auto-calling of Apache::SIG-set
Index: mod_perl.c
===
RCS file: /home/cvs/modperl/src/modules/perl/mod_perl.c,v
retrieving revision 1.91
diff -u -u -r1.91 mod_perl.c
--- mod_perl.c 1999/08/16 07:50:15 1.91
+++ mod_perl.c 1999/11/01 07:15:47
@@ -484,14 +484,12 @@
exit(1);
}
-#if !HAS_MMN_136
static void set_sigpipe(void)
{
char *dargs[] = { NULL };
perl_require_module("Apache::SIG", NULL);
perl_call_argv("Apache::SIG::set", G_DISCARD, dargs);
}
-#endif
void perl_module_init(server_rec *s, pool *p)
{
@@ -543,9 +541,7 @@
if(PERL_RUNNING() PERL_STARTUP_IS_DONE) {
saveINC;
mp_check_version();
-#if !HAS_MMN_136
set_sigpipe();
-#endif
}
if(perl_is_running == 0) {
Re: Apache::SSI and lookup_uri and POST req fails
On Tue, 26 Oct 1999, Pouneh Mortazavi wrote:
i'm so close to getting apache::ssi to work correctly but right now my
work has come to a halt...
I've tracked the problem down to POST-ed forms + SSI + virtual includes +
$r-lookup_uri, but i don't know how to fix it!
I use Apache::SSI to parse the contents of an HTML string and include any
ssi variables, templates etc before printing to the screen.
my modperl handler has a snippet of code like this:
my $ssi = Apache::SSI-new($output_html, $r);
$output_html = $ssi-output();
the modperl handler is used to
- display a login page
- authenticate the user and display any error information
I have a POST form for people to type in username,password and trigger
off this handler. i'm noticing that if i use a POST request, the returned
page just hangs forever, in the $r-lookup_uri stage of implementing a
virtual include via SSI. If I use a GET request, $r-lookup_uri completes
normally.
after the first time you read post data, you need the code below to
prevent somebody else from trying to read post data that's already been
read.
$r-method_number(M_GET);
$r-method('GET');
$r-headers_in-unset('Content-length');
Re: suggestion about PerlRun
On Thu, 28 Oct 1999, Yasushi Nakajima wrote:
Hello All
I have made a module derived from PerlRun. It overrides only readscript()
method in PerlRun. First I wrote as follows (essencial part only)
package Apache::PerlRunFake;
use Apache::PerlRun;
@ISA = qw(Apache::PerlRun);
sub readscript {
...
}
*handler = \Apache::PerlRun::handler;
But Apache::PerlRun::handler() inclueds following code.
my $pr = Apache::PerlRun-new($r);
the handler() in Apache::PerlRun was not originally indended to be
subclassed. however, if somebody submits a patch to make is
subclass-able that doesn't break the way Apache::PerlRun currently works,
that would be fine.
Re: forking from the main Apache process
On Sat, 23 Oct 1999, Stas Bekman wrote:
After investigating deeper into a hanging detection problem, I understood
that I can quite easily write a monitor that will detect these processes
and kill them off. I wrote a module Apache::SafeHang with help of
Apache::Scoreboard that should solve the problem.
The only problem I have with this module, is that it should be forked from
the main process after the server has completed the configuration. I've
tried to fork it during the startup (from startup.pl), but then I don't
get a valid Apache::Scoreboard image, so it's like starting it from shell.
Is it possible to execute some perl code in the parent process, a moment
before a first child starts or any time after that?
if you put your code in a Perl section and wait until the second time it
is called, apache will have initialized the scoreboard by then. if you
use a PerlRequire script, that's only call during the first init-modules
call, before the scoreboard is created.
The third question is again regarding the main process. If I fork the
process during a startup, when I perform 'apachectl stop' all the
processes die but the one I've forked. Is there any way to install END{}
block in the parent main process, so the children wouldn't inherit it?
Otherwise if they do and it supposed to kill the forked process, a first
child that quits will kill off the monitor. (I could probably test the pid
of the process that executes the END block, since I know the pid of the
parent, I could execute the block only if it's parent process...)
not really, you'd have to put logic in the END routines to do things (or
not do things) based on pid or some other flag.
Re: Intentional Premature Finish in an Apache Module
On Thu, 28 Oct 1999, Public Interactive wrote: I'd like to be able to prematurely end the thread of execution within a Perl apache module from someplace *other than* the PerlHandler entry point subroutine (usually "handler()"). That is, when I'm a few subroutines deep inside my module, I want to be able to spit out an error page and have the module finish as if handler() had returned OK. Right now I'm painstakingly propagating return values back up to my handler() subroutine, but I'm hoping there's a better way. here's a possible option (from my Mail/.sent-mail a few weeks back): ... or use this undocumented feature: $Apache::ERRSV_CAN_BE_HTTP = 1; #set this anywhere, startup script is best die FORBIDDEN; mod_perl peeks at $@, if it's a 3 digit http status code, it propagates that value to Apache.
Re: Memory Leaks?
sounds like you have PerlFreshRestart On, try turning it Off. scan the archives for more info. On Fri, 29 Oct 1999, Ben Bell wrote: Hi, I'm using the Debian package of mod_perl (1.21) and apache 1.3.9 and I've noticed quite nasty memory leaks on server restart. I've noticed unresolved bug reports on the Debian pages about this. Is it a known issue with this version? The leak is ca. 2MB each restart (or graceful) with my startup script enabled (which queries a database, and uses the following modules: Apache Apache::PerlSections Apache::DBI Data::Dumper Carp VI::Utils All my vars are declared as "my". When I disable all Perl stuff (my startup script, the perl sections etc) I still see a memory leak, albeit a smaller one. Can anyone shed any light on this? Cheers, Ben -- +-Ben Bell - "A song, a perl script and the occasional silly sig.-+ /// email: [EMAIL PROTECTED]www: http://www.deus.net/~bjb/ bjbDon't try to drive me crazy... \_/...I'm close enough to walk.
Re: Profiling
On 31 Oct 1999, Greg Stark wrote: Does anyone have any idea how much overhead Apache::DProf or Apache::SmallProf add? Will it be possible to use these on a production system without having a severe impact? I wouldn't using profiling on a production site, there is considerable overhead involved with profiling. and, as you've seen, possible oddities, such as random core dumps. I've seen cases such this code: my $len = $#array + 1; #dump core vs. my $len = scalar @array; #no problem
Re: modules to be left enabled in a mod_perl back-end server?
On 1 Oct 1999, Stephen Zander wrote: "Stephen" == Stephen Zander [EMAIL PROTECTED] writes: Stephen Is the SetHandler functionaity of mod_mime available from Stephen within mod_perl somehow? This was the one ovbious Stephen problem I could see with removing everything. Nevermind. The Eagle book, chapter 8 p407 is my friend. that's just an example though, you're probably better off just letting mod_mime take care of things if you don't need to change it's behavior. image Stephen goes off to thwap himself unconscious with Doug Lincoln's excellent book /image :) -Doug
RE: HTML::Embperl and 'undefined symbol: perl_eval_sv'
On Sun, 3 Oct 1999, Gerald Richter wrote: When I use HTML::Embperl for instance like: perl -MHTML::Embperl -e 'HTML::Embperl::Execute(some_embperl.epl)' I got: perl: error in loading shared libraries: /usr/local/lib/site_perl/5.00561/i686-linux/auto/HTML/Embperl/Embperl.so: undefined symbol: perl_eval_sv it seams that it can't be linked to perl. The versions is: Linux 2.3.10 5.005_61 built for i686-linux HTML-Embperl-1.2b9 there was no errors making the Embperl package. Does the "make test" runs without problem, or does it show the same error? The newest perl version I tried was 5.005_57, where it works without problem. Maybe something has changed inside perl since that version? yup, have a look at mod_perl.h in cvs: #ifdef eval_pv # ifndef perl_eval_pv # define perl_eval_pv eval_pv # endif #endif #ifdef eval_sv # ifndef perl_eval_sv # define perl_eval_sv eval_sv # endif #endif -Doug
