Issues with FastCGI setup using Apache/mod_fcgid.
I am not getting the results I would expect out of this setup based on experience with using the embedded server. I have been testing with our internal programs and with the example programs linked against wtfcgi. I am having three separate issues/symptoms. The first happens with both our internal programs and the example programs. I have Wt set to log to a file. I am getting a hunk of binary junk in the log file between the lines:
[2013-Nov-11 10:58:48.370584] 30975 - [info] "config: reading Wt config file: /etc/wt/wt_config.xml (location = '/srv/wt1.openrda.net/docroot/hello.wt')"
[2013-Nov-11 10:58:48.371687] 30975 - [info] "wtfcgi: reading FastCGI stream from stdin"
and the lines:
[2013-Nov-11 10:58:48.376290] 30962 [/hello.wt lqcCngA2qGw0JTEW] [info] "WebRequest: took 1113.4ms"
[2013-Nov-11 10:58:48.382212] 30962 [/hello.wt lqcCngA2qGw0JTEW] [info] "WebRequest: took 7.253ms"
I don't know where this is coming from but this is happening with all Wt examples I have tried. I am attaching the relevant wt_config.xml, apache vhost and mod_fcgi config files. My testing is being done under Debian Wheezy 64 bit and Ubuntu 12.04 64 bit. I am using Wt version 3.3.0.
My second issue/symptom is that debugging using http://myhost/myurl?debug has stopped working entirely. The program seems to run but acts like the request for debugging was never sent. I am not using valgrind. I am using ltrace in it's place at the moment. This was working putting ltrace in the valgrind path section of the wt_config.xml file. This isn't working at all for both internal programs and Wt demo programs now.
My last issue/symptom is only happening with our internal programs. I am getting these messages in addition to the binary junk from issue1:
[2013-Nov-11 11:17:26.717760] 13559 - [info] "config: reading Wt config file: /etc/wt/wt_config.xml (location = '/srv/wt1.openrda.net/docroot/fldtest.lnx')"
[2013-Nov-11 11:17:26.717973] 13559 - [info] "wtfcgi: reading FastCGI stream from stdin"
[2013-Nov-11 11:17:26.959048] 13559 - [error] "wtfcgi: error reading from application"
[2013-Nov-11 11:17:29.702622] 13559 - [info] "wtfcgi: shutdown (caught SIGTERM)"
error reading from application\"
[2013-Nov-11 11:17:29.702621] 13521 - [info] "wtfcgi: shutdown (caught SIGTERM)"
[binary junk here]
[2013-Nov-11 11:17:26.718650] 13532 [/fldtest.lnx Na9MYc4bGDrl2ByB] [info] "WebRequest: took 1064.93ms"
[2013-Nov-11 11:17:26.719024] 13532 [/fldtest.lnx Na9MYc4bGDrl2ByB] [info] "WebRequest: took 0.532ms"
One thing I notice is that the error line seems to be repeated improperly or somehow being stepped on by something else trying to write the log entries. Our program seems to run but never display a web page. I am not sure what is getting the SIGTERM or what is sending the signal. The hello.wt seems to be working ok and displays a web page. Our internal programs seem to lose communication with mod_fcgid somehow. Any insight you can provide on these issues would be greatly appreciated. If I can get the ltrace/valgrind debugging working again, issue 3 would probably make a lot more sense.