[LON-CAPA-users] Unresponsive lon-capa

Mon, 07 Feb 2011 14:32:48 -0500

This is probably a separate issue from Donald Teed's, but I am also seeing intermittent "non-responsiveness".  Sometimes pages won't load, sometimes half a page loads, but then it stops.  

As an email from a student reads:
> Hello!
> 
> My Lon-Capa just never wants to work. When I submit an answer I have to press refresh between 10 to 30 times until the page reappears. This make entering data a days long process. I tried it on different computers and I have the same issue.
> 
> Is there ANYTHING I can do. It is sooo super frustrating.

A lot of users pressing refresh 10-20 times cold cause a problem, but the fact that they feel a need to is a problem in itself.

Looking in /var/log/httpd/error_log, the only thing I see are repeated lines of the type 

> [Sun Jan 30 22:56:24 2011] [error] [client 130.49.12.164] Apache2::RequestIO::print: (103) Software caused connection abort at /home/httpd/lib/perl//Apache/lonhomework.pm line 1042
> [Sun Jan 30 22:56:24 2011] [error] [client 130.49.12.164] Apache2::RequestIO::print: (103) Software caused connection abort at /home/httpd/lib/perl//Apache/lonerrorhandler.pm line 53
> [Sun Jan 30 22:56:27 2011] [error] [client 130.49.12.164] Apache2::RequestIO::print: (103) Software caused connection abort at /home/httpd/lib/perl//Apache/lonhomework.pm line 1042
> [Sun Jan 30 22:56:27 2011] [error] [client 130.49.12.164] Apache2::RequestIO::print: (103) Software caused connection abort at /home/httpd/lib/perl//Apache/lonerrorhandler.pm line 53
> [Sun Jan 30 22:56:30 2011] [error] [client 130.49.12.164] Apache2::RequestIO::print: (103) Software caused connection abort at /home/httpd/lib/perl//Apache/lonhomework.pm line 1042
> [Sun Jan 30 22:56:30 2011] [error] [client 130.49.12.164] Apache2::RequestIO::print: (103) Software caused connection abort at /home/httpd/lib/perl//Apache/lonerrorhandler.pm line 53
> [Sun Jan 30 22:56:35 2011] [error] [client 130.49.12.164] Apache2::RequestIO::print: (103) Software caused connection abort at /home/httpd/lib/perl//Apache/lonhomework.pm line 1042
> [Sun Jan 30 22:56:35 2011] [error] [client 130.49.12.164] Apache2::RequestIO::print: (103) Software caused connection abort at /home/httpd/lib/perl//Apache/lonerrorhandler.pm line 53
> [Sun Jan 30 22:56:38 2011] [error] [client 130.49.12.164] Apache2::RequestIO::print: (103) Software caused connection abort at /home/httpd/lib/perl//Apache/lonhomework.pm line 1042
> [Sun Jan 30 22:56:38 2011] [error] [client 130.49.12.164] Apache2::RequestIO::print: (103) Software caused connection abort at /home/httpd/lib/perl//Apache/lonerrorhandler.pm line 53
sometimes there is extra on the line about the referer
> [Wed Jan 05 19:05:18 2011] [error] [client 24.3.142.4] Apache2::RequestIO::print: (103) Software caused connection abort at /home/httpd/lib/perl//Apache/lonhomework.pm line 1042, referer: http://nplq1.phyast.pitt.edu/res/msu/physicslib/msuphysicslib/01_Math_1/msu-prob06.problem?symb=uploaded%2fpitt%2f7i18584d02b0a4d9apittphyast1%2fdefault_1262744012%2esequence___8___msu%2fphysicslib%2fmsuphysicslib%2f01_Math_1%2fmsu%2dprob06%2eproblem
> [Wed Jan 05 19:05:18 2011] [error] [client 24.3.142.4] Apache2::RequestIO::print: (103) Software caused connection abort at /home/httpd/lib/perl//Apache/lonerrorhandler.pm line 53, referer: http://nplq1.phyast.pitt.edu/res/msu/physicslib/msuphysicslib/01_Math_1/msu-prob06.problem?symb=uploaded%2fpitt%2f7i18584d02b0a4d9apittphyast1%2fdefault_1262744012%2esequence___8___msu%2fphysicslib%2fmsuphysicslib%2f01_Math_1%2fmsu%2dprob06%2eproblem

I don't know whether these errors have anything to do the responsiveness of the system, and since the problem is intermittent, it is difficult to track down.

Any ideas of where to look?

-Jim Mueller

On Feb 7, 2011, at 12:01 PM, D G Teed wrote:

> On Mon, Feb 7, 2011 at 12:00 PM, Stuart Raeburn <raeburn@msu.edu> wrote:
>> 
>> Hi,
>> 
>> Gerd mentioned the Apache server-status page: "http://(yourserver)/server-status/" and indicated he would send the password to access this page.
>> 
>> Starting with LON-CAPA 2.8.0 access to this page (and some other status pages) was switched from using htpasswd-based authentication and authorization, to using access control via LON-CAPA (and configured for a domain by the Domain Coordinator).
>> 
>> When logged into LON-CAPA, Domain Coordinators can view this page via:
>> Main Menu -> Status of domain servers -> Apache Server Status
>> 
>> A Domain Coordinator can also configure access to this page via:
>> Main Menu -> Set domain configuration -> Display [Access to server status pages (checked)]
>> 
>> One of the access modes supported is IP-based, which permits display of server-status from a specified machine without the need for log-in to LON-CAPA.  This can be useful for remote monitoring.
>> 
>>> Yes, http://(yourserver)/server-status/
>>> 
>>> - that will show you what the server is busy doing and you can find  the problem that's the problem. I'll send you passwords in a minute.
>>> 
>> 
>> There can be a difficulty here in that once Apache has reached its maximum clients limit, and all clients are busy, possibly serving looped processes, it is difficult to get a free Apache client to display the server-status page.
>> 
> 
> A non-responding apache service has been the consistent
> problem.  Last year I saw cases where the I/O was
> grinding the server to death and it wasn't responsive
> in any way.  But in the last case, it only impacted the
> apache service, and the page would not display for
> login - nothing at all, but browser timeouts trying
> to reach it.  Even the logs were silent when I tried to
> get the login screen.
> 
> So what's plan B?  There must be another way to get the
> information on the back end if I have a healthy ssh login.
> 
> Also, when attempting access to Set Domain Configuration,
> while at this time the apache service is responding
> normally, I saw an error just now:
> 
> Something somewhere went wrong... etc. - the report form.
> 
> The error was:
> 
> Internal info:
> 
> ERROR:
> Undefined subroutine &Apache::domainprefs::handler called.
> 
> REQUEST_URI:
> /adm/domainprefs
> SCRIPT:
> /home/httpd/html/adm/domainprefs
> SCRIPT:
> /adm/domainprefs
> 
> In the error.log:
> 
> Undefined subroutine &Apache::domainprefs::handler called.\n
> 
> I guess I'm missing some perl modules or mine are out of date.
> 
> I'm able to get to the server-status page at this moment using the
> "Status of domain servers" link.  I'll try monitoring this in case it
> is useful.
> 
> --Donald Teed
> 
> _______________________________________________
> LON-CAPA-users mailing list
> LON-CAPA-users@mail.lon-capa.org
> http://mail.lon-capa.org/mailman/listinfo/lon-capa-users