Re: [phusion-passenger] Segmentation fault with 3.0.17

[Hongli Lai]

On Wed, Oct 17, 2012 at 11:52 AM, Christoph <s...@tachium.org> wrote:

Hey,

I'm getting a segfault with 3.0.17 and have no idea why

https://gist.github.com/735164bdf41607e524ab

Does anyone know what causes this?

The crash reports indicates that the app crashes in the switch_user function, which is used by Phusion Passenger to change the process's user and group for security purposes. This function calls the corresponding system calls (setuid() and friends) directly, and should never crash, but we've received several reports about it crashing anyway. The reason thus far remains unknown, although we suspect that it has got something to do with broken user databases on the system. We just know that the same code, when called from C++, doesn't crash. There's something in Ruby that makes it crash.

You should give Phusion Passenger 4 a try. A large part of the codebase has been moved into C++ so this problem is avoided. Phusion Passenger 4 hasn't been released yet but you can get the preelease code from Github under the 'experimental' branch.

--
Phusion | Ruby & Rails deployment, scaling and tuning solutions

Web: http://www.phusion.nl/
E-mail: in...@phusion.nl
Chamber of commerce no: 08173483 (The Netherlands)

[Christoph]

Not sure if I can convince my company to use an experimental branch :)

Would a downgrade make sense?

[Hongli Lai]

On Wed, Oct 17, 2012 at 1:40 PM, Christoph <s...@tachium.org> wrote:

Not sure if I can convince my company to use an experimental branch :)

Would a downgrade make sense?

The branch is still called experimental but not really experimental. We've been running it on our servers in production for months now. PhusionPassenger.com and Bubbleconf.com are powered by it, for example. A first beta will be released soon.

A downgrade will not help though. This is not a bug that has recently been introduced. It's a problem that's somehow only triggered when called from Ruby, so the only way to work around it is by calling it from pure C++, which only Phusion Passenger 4 does.

[Christoph]

Thanks a lot for the clarification! I will try 4.0 in that case!

[Christoph]

I ran into a problem using the experimental branch, startup says:

[Wed Oct 17 14:41:45 2012] [notice] Apache/2.2.15 (Unix) DAV/2 Phusion_Passenger/3.0.17 PHP/5.3.3 mod_ssl/2.2.15 OpenSSL/1.0.0-fips configured -- resuming normal operations

is this a misconfiguration or does it still say version 3.0.17?

in the config I have now:

LoadModule passenger_module /usr/local/rvm/gems/ruby-1.9.2-p180/gems/passenger-3.9.0/libout/apache2/mod_passenger.so

PassengerRoot /usr/local/rvm/gems/ruby-1.9.2-p180/gems/passenger-3.9.0

PassengerRuby /usr/local/rvm/wrappers/ruby-1.9.2-p180/ruby

and the error is:

[ pid=22712 thr=139906783815648 file=ext/apache2/Hooks.cpp:884 time=2012-10-17 14:44:51.994 ]: Unexpected error in mod_passenger: The ApplicationPool server unexpectedly closed the connection while we're reading a response for the 'get' command.

  Backtrace:

     in 'void Passenger::ApplicationPool::Client::sendGetCommand(const Passenger::PoolOptions&, std::vector<std::basic_string<char, std::char_traits<char>, std::allocator<char> >, std::allocator<std::basic_string<char, std::char_traits<char>, std::allocator<char> > > >&)' (Client.h:374)

     in 'virtual Passenger::SessionPtr Passenger::ApplicationPool::Client::get(const Passenger::PoolOptions&)' (Client.h:691)

     in 'Passenger::SessionPtr Hooks::getSession(const Passenger::PoolOptions&)' (Hooks.cpp:312)

     in 'int Hooks::handleRequest(request_rec*)' (Hooks.cpp:585)

[Christoph]

nevermind, sorry - someone installed this through the OS package manager :(

On Wednesday, October 17, 2012 11:52:34 AM UTC+2, Christoph wrote:

Hey,

I'm getting a segfault with 3.0.17 and have no idea why

https://gist.github.com/735164bdf41607e524ab

Does anyone know what causes this?

Thanks,

Christoph