I'm unable to get a Rails application running using LSAPI on Solaris 10 with Ruby 1.8.6 from Blastwave. I'm getting a 503 error.
This looks similar to the problem reported in thread [thread]2065[/thread], however I had no issues compiling the ruby-lsapi from source.
The workaround mentioned in that thread is to compile Ruby from source. Unfortunately, maintaining anything compiled directly from source is not an option. Is there any sensible reason why Ruby LSAPI could not be usable with Ruby from Blastwave? I had no issues with all the other gems/modules so far.
truss output of the ruby /opt/lsws/fcgi-bin/RailsRunner.rb process is attached as well as the manifest used for starting LSWS.
The Rails application is running fine in LiteSpeed through FastCGI or in Mongrel.
OS: Solaris 10, 64-bit, i86pc
LSWS 3.3.17 Standard
Ruby 1.8.6 from Blastwave, Rails 2.1.0 from gem, Ruby LSAPI 3.2 from source
Code:
2008-09-06 14:11:53.934 [NOTICE] Loading LiteSpeed/3.3.17 Standard ...
2008-09-06 14:11:53.936 [NOTICE] [ADMIN] server socket: uds://opt/lsws/admin/tmp/admin.sock.7309
2008-09-06 14:11:53.936 [NOTICE] Loading configuration from /opt/lsws/conf/httpd_config.xml ...
2008-09-06 14:11:53.937 [INFO] old priority: 0, new priority: 0
2008-09-06 14:11:53.937 [INFO] [config:server:basic] For better obscurity, server version number is hidden in the response header.
2008-09-06 14:11:54.375 [NOTICE] The maximum number of file descriptor limit is set to 65536.
2008-09-06 14:11:54.379 [NOTICE] [AutoRestarter] new child process with pid=28554 is forked!
2008-09-06 14:11:54.379 [NOTICE] [child: 28554] Successfully change current user to litesped
2008-09-06 14:11:54.380 [INFO] [Rails:RailsLSAPI:/] add child process pid: 28555
2008-09-06 14:11:54.380 [INFO] [Rails:RailsLSAPI:/] pid list size: 1
2008-09-06 14:11:54.381 [NOTICE] [Child: 28554] Setup swapping space...
2008-09-06 14:11:54.381 [NOTICE] [Child: 28554] LiteSpeed/3.3.17 Standard starts successfully!
2008-09-06 14:11:55.114 [NOTICE] [AutoUpdate] Checking for new releases...
2008-09-06 14:12:35.753 [INFO] [127.0.0.1:40411-0#RailsLSAPI] connection to [uds://tmp/lshttpd/RailsLSAPI:_.sock] on request #0, error: Connection reset by peer!
2008-09-06 14:12:35.754 [INFO] [127.0.0.1:40411-0#RailsLSAPI] connection to [uds://tmp/lshttpd/RailsLSAPI:_.sock] on request #0, error: Connection reset by peer!
2008-09-06 14:12:35.755 [INFO] [127.0.0.1:40411-0#RailsLSAPI] connection to [uds://tmp/lshttpd/RailsLSAPI:_.sock] on request #0, error: Connection reset by peer!
2008-09-06 14:12:35.755 [NOTICE] [127.0.0.1:40411-0#RailsLSAPI] oops! 503 Service Unavailable
2008-09-06 14:12:35.755 [NOTICE] [127.0.0.1:40411-0#RailsLSAPI] Content len: 0, Request line:
GET / HTTP/1.1
2008-09-06 14:12:35.755 [NOTICE] [127.0.0.1:40411-0#RailsLSAPI] Redirect: #2, URL: /dispatch.lsapiThe workaround mentioned in that thread is to compile Ruby from source. Unfortunately, maintaining anything compiled directly from source is not an option. Is there any sensible reason why Ruby LSAPI could not be usable with Ruby from Blastwave? I had no issues with all the other gems/modules so far.
truss output of the ruby /opt/lsws/fcgi-bin/RailsRunner.rb process is attached as well as the manifest used for starting LSWS.
The Rails application is running fine in LiteSpeed through FastCGI or in Mongrel.
OS: Solaris 10, 64-bit, i86pc
LSWS 3.3.17 Standard
Ruby 1.8.6 from Blastwave, Rails 2.1.0 from gem, Ruby LSAPI 3.2 from source