IIS server returns HTTP 500 errors outside of FastCGI loop

[geographika]

As noted in #6303 - Windows IIS (and possibly other servers) return HTTP 500 errors if the program exits, rather than returning an HTTP response with an error. This makes it hard to diagnose the underlying issue - for example the path to the config file could be incorrect, or the environment variable is missing.

Examples:

MapServer/mapserv.c

Lines 144 to 148 in 17d8fa9

	if( msSetup() != MS_SUCCESS ) {
	msCGIWriteError(mapserv);
	msCleanup();
	exit(0);
	}

MapServer/mapserv.c

Lines 156 to 158 in b55b128

	if(config == NULL) {
	msCGIWriteError(mapserv);
	exit(0);

Attempts to exit within the FCGI_Accept() >= 0 loop worked once and then failed.
Possible solution is to simply keep going to goto end_request;

[sdlime]

I have a feeling we'll need an alternative to msCGIWriteError() for this use case, something more independent of any initialization or setup.

[geographika]

A couple of things that didn't work:

• using return (0) instead of `exit(0)
• setting finish_process = 1; in the block where if(config == NULL) (this variable comes from FastCGI signal handling: no longer do heavy work in the signal handler (fixes #6167) #6385)

I think what we probably want to happen is either:

1. The process fails, but somewhere the issue of a missing config file is logged - currently on IIS the HTTP 500 error is returned. Request logging only produces:

ModuleName FastCgiModule 
Notification EXECUTE_REQUEST_HANDLER 
HttpStatus 500 
HttpReason
HttpSubStatus 0 
ErrorCode The operation completed successfully.  (0x0) 

2. We let the process continue (the FastCGI loop), but each request simply outputs the same MapServer error

I'm interested how this appears on other web servers - does setting MAPSERVER_CONFIG_FILE to a missing file return HTTP 500 with no logs/messages?

[sdlime]

I can confirm the 500 error with FastCGI using Vagrant. Nothing in the logs other than the standard "End of script before headers: mapserv.fcgi". The setup is:

1. No config file defined as an environment variable using SetEnv or FcgiidInitialEnv.
2. Using "wget -0 - http://localhost/cgi-bin/mapserv.cgi" or mapserv.fcgi.

The CGI returns the config file not found error as expected.

[sdlime]

@geographika, if I change the configuration error handling to do some FastCGI related initialization before writing the error then I don't see a 500 error and get the expected error message. @rouault, wondering if you think this is reasonable...

  config = msLoadConfig(NULL); // is the right spot to do this?                                                                           
  if(config == NULL) {
#ifdef USE_FASTCGI
    msIO_installFastCGIRedirect();
    FCGI_Accept();
#endif
    msCGIWriteError(mapserv);
    exit(0);
  }

[geographika]

@sdlime - I can confirm with this update on IIS/FastCGI I get the following error with an invalid path, rather than a generic HTTP 500 error:

msLoadConfig(): Unable to access file. (C:\MapServer\bin\mapserver.confx)

And when missing the environ variable MAPSERVER_CONFIG_FILE (as it defaults to the default path):

msLoadConfig(): Unable to access file. (/mapserver.conf)

This will make it much easier for users to setup/diagnose the issue.

[sdlime]

Thing is that I don't think we should be outputting the config file name/path.

[geographika]

A generic msLoadConfig(): Unable to access the config file set in the MAPSERVER_CONFIG_FILE variable would be enough to start debugging whilst remaining secure?

[sdlime]

IMHO, it would be nice to do something like: msLoadConfig(): Unable to access config file (see mapserver.org/config_file.html).

…

--Steve From: Seth G ***@***.***> Sent: Tuesday, September 21, 2021 9:21 AM To: MapServer/MapServer ***@***.***> Cc: Lime, Steve D (MNIT) ***@***.***>; Mention ***@***.***> Subject: Re: [MapServer/MapServer] IIS server returns HTTP 500 errors outside of FastCGI loop (#6411) This message may be from an external email source. Do not select links or open attachments unless verified. Report all suspicious emails to Minnesota IT Services Security Operations Center.

________________________________ A generic msLoadConfig(): Unable to access the config file set in the MAPSERVER_CONFIG_FILE variable would be enough to start debugging whilst remaining secure? - You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub<https://gcc02.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgithub.com%2FMapServer%2FMapServer%2Fissues%2F6411%23issuecomment-924037299&data=04%7C01%7Csteve.lime%40state.mn.us%7Cedd59a5234b74595e19808d97d0b00aa%7Ceb14b04624c445198f26b89c2159828c%7C0%7C0%7C637678308477132055%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C1000&sdata=krLbl5uEnxQNd9VrIM2kzlLdq0STiSA9G9CjKmct6ng%3D&reserved=0>, or unsubscribe<https://gcc02.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgithub.com%2Fnotifications%2Funsubscribe-auth%2FAAMGN64WRH27J4EEOLNX3N3UDCID3ANCNFSM5EILLQKQ&data=04%7C01%7Csteve.lime%40state.mn.us%7Cedd59a5234b74595e19808d97d0b00aa%7Ceb14b04624c445198f26b89c2159828c%7C0%7C0%7C637678308477132055%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C1000&sdata=m2Z3KiqeRMOx4LvgmuC3hj3Bsgdon2jUw5%2FhhUU2riA%3D&reserved=0>. Triage notifications on the go with GitHub Mobile for iOS<https://gcc02.safelinks.protection.outlook.com/?url=https%3A%2F%2Fapps.apple.com%2Fapp%2Fapple-store%2Fid1477376905%3Fct%3Dnotification-email%26mt%3D8%26pt%3D524675&data=04%7C01%7Csteve.lime%40state.mn.us%7Cedd59a5234b74595e19808d97d0b00aa%7Ceb14b04624c445198f26b89c2159828c%7C0%7C0%7C637678308477142016%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C1000&sdata=qI3sYrnaOwAxYCzJ8Z9h7CIqHUPBv3B%2BKHbOveSaoD4%3D&reserved=0> or Android<https://gcc02.safelinks.protection.outlook.com/?url=https%3A%2F%2Fplay.google.com%2Fstore%2Fapps%2Fdetails%3Fid%3Dcom.github.android%26referrer%3Dutm_campaign%253Dnotification-email%2526utm_medium%253Demail%2526utm_source%253Dgithub&data=04%7C01%7Csteve.lime%40state.mn.us%7Cedd59a5234b74595e19808d97d0b00aa%7Ceb14b04624c445198f26b89c2159828c%7C0%7C0%7C637678308477142016%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C1000&sdata=xCmgMNbsxgc4%2BmIzgdV76FFBaRNIh6RdnA4NWnkGQEU%3D&reserved=0>.

[sdlime]

@geographika, think we can close this?

[geographika]

Yes, fixed as part of #6303 - thanks @sdlime !