<div dir="ltr"><div><div><div>Hi Stuart,<br><br></div>Is it possible to have the files that contain the checkpoints to see why it is so slow on tycho? It is running 64-bit w 16G of ram, almost the same as the production servers (they have 32 G of ram.)<br>
<br></div>Regards,<br></div>-hk<br><div class="gmail_extra"><br clear="all"><div><div dir="ltr"><br></div></div>
<br><br><div class="gmail_quote">On Wed, Jul 9, 2014 at 12:03 PM, Stuart Raeburn <span dir="ltr"><<a href="mailto:raeburn@msu.edu" target="_blank">raeburn@msu.edu</a>></span> wrote:<br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
Hi Hon-Kie,<br>
<div class=""><br>
> Yes, they are the problems in openstax-CollegePhysics library. I have added<br>
> you as co-author so can poke around. Just view the<br>
> openstax-CollegePhysics/02_Kinematics/01_Displacement.problem - it took<br>
> more than 30 s sometimes.<br>
<br>
</div>Thank you for providing co-author access for me in the dev cluster for<br>
openstax-CollegePhysics:fsu.<br>
<br>
I copied the XML for<br>
/priv/fsu/openstax-CollegePhysics/02_Kinematics/01_Displacement.problem (and<br>
also the XML in the two .library files which are imported into the<br>
problem) and created equivalent files in my Authoring Space on the MSU<br>
dev library server.<br>
<br>
When I compare rendering times on the MSU dev server (32 bit) for my<br>
copy of 01_Displacement.problem in Authoring Space on the same<br>
hardware, both before and after upgrade to 2.11.0, I find similar<br>
rendering times (about 6 s), with 2.11.0 typically faster than 2.10.1<br>
by about 10%.<br>
<br>
Note: comparing rendering times displaying the published resource<br>
(/res/ ...) versus display in Authoring Space (/priv/ ...) I was<br>
typically seeing 2s versus 6s, because in Authoring Space,<br>
Apache::lonhomework::renderpage() will render for three targets:<br>
grade, web, and answer, whereas for resource space, rendering is for<br>
just the one target (web).<br>
<br>
I used Time::HiRes with a number of checkpoints in <a href="http://lonhomework.pm" target="_blank">lonhomework.pm</a> and<br>
see no significant difference between 2.10.1 and 2.11.0 when rendering<br>
the 01_Displacement.problem resource, which you report as taking up to<br>
30s to render on your dev server.<br>
<br>
I used the same checkpoints to analyze rendering times for 2.10.1 and<br>
2.11.0 on a VM running CentOS 6 (64 bit) for 01_Displacement.problem,<br>
and found similar rendering times, of about 2s, for display in<br>
Authoring Space (with 2.11.0 about 10% faster than 2.10.1).<br>
<br>
Note: after updating to gnuplot 4.4 (for 2.11.0) I needed to patch<br>
2.10.1 (when I reverted to that LON-CAPA version) -- see:<br>
<a href="http://mail.lon-capa.org/pipermail/lon-capa-admin/2014-July/002871.html" target="_blank">http://mail.lon-capa.org/pipermail/lon-capa-admin/2014-July/002871.html</a> -- for<br>
the plot to be generated.<br>
<div class="HOEnZb"><div class="h5"><br>
<br>
Stuart Raeburn<br>
LON-CAPA Academic Consortium<br>
<br>
<br>
Quoting "H. K. Ng" <<a href="mailto:hkng@fsu.edu">hkng@fsu.edu</a>>:<br>
<br>
> Hi Stuart,<br>
><br>
> Thanks for your response.<br>
><br>
><br>
><br>
> On Wed, Jul 2, 2014 at 6:08 PM, Stuart Raeburn <<a href="mailto:raeburn@msu.edu">raeburn@msu.edu</a>> wrote:<br>
><br>
>> Hi Hon-Kie,<br>
>><br>
>> > The latest release version takes a long time (~30 s) to render a problem<br>
>> in<br>
>> > the construction space. 2.10 takes only a couple of seconds.<br>
>><br>
>> I assume this is for your development server (tycho) in the dev<br>
>> cluster, as your production library server is still running 2.10.1<br>
>><br>
><br>
> Yes, I am testing it out first before installing 2.11 on the production<br>
> servers.<br>
><br>
><br>
>><br>
>> Were you ever able to resolve the yum update issues on your dev server<br>
>> caused by epel repo conflicts?<br>
>><br>
><br>
> Leaving the disable flag did the trick.<br>
><br>
><br>
>><br>
>> The construction space rendering behavior you report has not been my<br>
>> experience on MSU single domain library servers in both the production<br>
>> cluster and dev cluster. When I compare rendering times for the same<br>
>> problem in Authoring Space on the same hardware, both before and after<br>
>> upgrade to 2.11.0, I find rendering times the same (about 2 s).<br>
>><br>
>> Has the problem which you find takes ~30 s to render been published?<br>
>><br>
><br>
> Yes, they are the problems in openstax-CollegePhysics library. I have added<br>
> you as co-author so can poke around. Just view the<br>
> openstax-CollegePhysics/02_Kinematics/01_Displacement.problem - it took<br>
> more than 30 s sometimes.<br>
><br>
><br>
>><br>
>> If so, I could replicate the resource, and then measure rendering<br>
>> times on the MSU development server, and debug, if I also found<br>
>> anomalous (long) rendering times.<br>
>><br>
>> What files (and return codes) do you find listed in the Apache<br>
>> ssl_access_log log file when rendering the problem?<br>
>><br>
><br>
> The log files have the following output.<br>
><br>
> ssl_access.log shows the following:<br>
><br>
> 146.xx - openstax-CollegePhysics [02/Jul/2014:18:42:04 -0400] "POST<br>
> /adm/createuser HTTP/1.1" 200 18798<br>
> 146.xx - openstax-CollegePhysics [02/Jul/2014:18:48:57 -0400] "GET<br>
> /priv/fsu/openstax-CollegePhysics/ HTTP/1.1" 200 69158<br>
> 146.xx - openstax-CollegePhysics [02/Jul/2014:18:49:12 -0400] "GET<br>
> /priv/fsu/openstax-CollegePhysics/02_Kinematics/ HTTP/1.1" 200 169377<br>
> 128.xx - openstax-CollegePhysics [02/Jul/2014:18:49:18 -0400] "POST<br>
> /priv/fsu/openstax-CollegePhysics/02_Kinematics/01_Displacement.problem?inhibitmenu=yes<br>
> HTTP/1.1" 200 2516<br>
> 128.xx - openstax-CollegePhysics [02/Jul/2014:18:49:23 -0400] "POST<br>
> /priv/fsu/openstax-CollegePhysics/ToolBox/ReferenceKit.library?inhibitmenu=yes<br>
> HTTP/1.1" 200 -<br>
> 128.xx - openstax-CollegePhysics [02/Jul/2014:18:49:24 -0400] "POST<br>
> /priv/fsu/openstax-CollegePhysics/ToolBox/gnuplotRoutines.library?inhibitmenu=yes<br>
> HTTP/1.1" 200 -<br>
><br>
> ssl_request.log shows the following:<br>
><br>
> [02/Jul/2014:18:42:04 -0400] 146.201.38.158 TLSv1.2 DHE-RSA-AES128-SHA<br>
> "POST /adm/createuser HTTP/1.1" 18798<br>
> [02/Jul/2014:18:48:57 -0400] 146.201.38.158 TLSv1.2 DHE-RSA-AES128-SHA "GET<br>
> /priv/fsu/openstax-CollegePhysics/ HTTP/1.1" 69158<br>
> [02/Jul/2014:18:49:12 -0400] 146.201.38.158 TLSv1.2 DHE-RSA-AES128-SHA "GET<br>
> /priv/fsu/openstax-CollegePhysics/02_Kinematics/ HTTP/1.1" 169377<br>
> [02/Jul/2014:18:49:18 -0400] 128.186.7.100 TLSv1.2<br>
> DHE-RSA-AES256-GCM-SHA384 "POST<br>
> /priv/fsu/openstax-CollegePhysics/02_Kinematics/01_Displacement.problem?inhibitmenu=yes<br>
> HTTP/1.1" 2516<br>
> [02/Jul/2014:18:49:23 -0400] 128.186.7.100 TLSv1.2<br>
> DHE-RSA-AES256-GCM-SHA384 "POST<br>
> /priv/fsu/openstax-CollegePhysics/ToolBox/ReferenceKit.library?inhibitmenu=yes<br>
> HTTP/1.1"<br>
><br>
><br>
> error.log - lots of entries like the ones below.<br>
><br>
> Bareword found where operator expected at (eval 7341) line 1, near<br>
> "@E_TRIE_MAXBUF"<br>
> (Missing operator before E_TRIE_MAXBUF?)<br>
> Bareword found where operator expected at (eval 7343) line 1, near<br>
> "%E_TRIE_MAXBUF"<br>
> (Missing operator before E_TRIE_MAXBUF?)<br>
><br>
><br>
><br>
><br>
><br>
>> Looking at diagnostics for tycho it appears that you still have rev<br>
>> 1.152.2.2 of /home/httpd/html/res/adm/includes/default_homework.lcpm<br>
>> installed (i.e., the version included with 2.10.1) instead of rev.<br>
>> 1.169 (the version shipped with 2.11.0). Is there a reason for that?<br>
>><br>
><br>
> Hum, not sure why. I run UPDATE so not sure why the old file was not<br>
> replaced. I replaced it manually but do not see any improvement.<br>
><br>
> Thanks again,<br>
> -hk<br>
><br>
><br>
><br>
><br>
><br>
>><br>
>><br>
>> Stuart Raeburn<br>
>> LON-CAPA Academic Consortium<br>
>><br>
>><br>
>> Quoting "H. K. Ng" <<a href="mailto:hkng@fsu.edu">hkng@fsu.edu</a>>:<br>
>><br>
>> > Hi,<br>
>> ><br>
>> > The latest release version takes a long time (~30 s) to render a problem<br>
>> in<br>
>> > the construction space. 2.10 takes only a couple of seconds. Any reason<br>
>> why<br>
>> > it takes so long? Is there any configurations that can be changed to<br>
>> speed<br>
>> > it up?<br>
>> ><br>
>> > Thanks,<br>
>> > -hk<br>
>> ><br>
>><br>
>> _______________________________________________<br>
>> LON-CAPA-admin mailing list<br>
>> <a href="mailto:LON-CAPA-admin@mail.lon-capa.org">LON-CAPA-admin@mail.lon-capa.org</a><br>
>> <a href="http://mail.lon-capa.org/mailman/listinfo/lon-capa-admin" target="_blank">http://mail.lon-capa.org/mailman/listinfo/lon-capa-admin</a><br>
<br>
_______________________________________________<br>
LON-CAPA-admin mailing list<br>
<a href="mailto:LON-CAPA-admin@mail.lon-capa.org">LON-CAPA-admin@mail.lon-capa.org</a><br>
<a href="http://mail.lon-capa.org/mailman/listinfo/lon-capa-admin" target="_blank">http://mail.lon-capa.org/mailman/listinfo/lon-capa-admin</a><br>
<br>
<br>
</div></div></blockquote></div><br></div></div>