[Clas_offline] CLASWEB Migration Status -- Progress on Pezca... Ready to switch?

[Eugene Pasyuk]

These don't work too:
http://claswebnew.jlab.org/cgi-bin/spin_rotation/magic_energies.pl
http://claswebnew.jlab.org/cgi-bin/spin_rotation/spin_rotation.pl

-Eugene

On 01/06/11 09:12, Eugene Pasyuk wrote:
> Hi Marty,
>
> These two applications still don't show plots:
> http://claswebnew.jlab.org/cgi-bin/epics_history/epics.pl
> http://claswebnew.jlab.org/cgi-bin/ONLINE_TIMELINE/timeline_frames.pl
>
> I believe it is gnuplot issue.
> You managed to fix it in
> http://claswebnew.jlab.org/cgi-bin/HALL_ABC/top_frame-MEW.pl
> It displays plots generated by gnuplot.
> Probably should be similar fix in those too.
>
> -Eugene
>
> On 01/06/11 06:52, Marty Wise wrote:
>>
>> All,
>>
>> Using the Pezca-0.4 source package, I was able to build and install 
>> it into our central /apps version of perl without having to build or 
>> make any parts of the EPICS distribution available to it. At runtime, 
>> the package still needs access to a couple of EPICS shared libraries, 
>> but this is easily arranged through LD_LIBRARY_PATH. The bottom line 
>> is that I have a working example of this new configuration.  If you 
>> access the URL: 
>> http://claswebnew.jlab.org/cgi-bin/HALL_ABC/top_frame-MEW.pl, you'll 
>> see an example of the configuration at work. This page is a 
>> _/slighlty/_ modified version of the original -- 
>> http://claswebnew.jlab.org/cgi-bin/HALL_ABC/top_frame.pl which does 
>> not work.
>>
>> After building Pezca into /apps Perl, I made a small modification to 
>> the top_frame-MEW.pl script mentioned above. All I had to do was 
>> comment out the line:
>>
>>     #use lib 
>> ('/group/clas/www/clasweb/htbin/PERL_CPAN/lib/site_perl/5.6.0/i686-linux');
>>
>> This change removes the explicit reference to the perl library 
>> installation down in /group/clas/www/clasweb/... Since the script 
>> uses /apps/bin/perl, it Is able to find the Pezca module in the apps 
>> build.
>>
>> To address the runtime dependence on the epics shlibs, I set 
>> LD_LIBRARY_PATH in the web server configuration to include a 
>> temporary directory into which I've placed the libca, libCom and 
>> libezca shared libraries.
>>
>> This seems to work, though it's hard to be sure since I'm not very 
>> familiar with them.
>>
>> The problem is that with the explicit library reference, there is no 
>> way to fix this without modifying all the scripts individually. You 
>> guys have a number of scripts configured like this. Changing them now 
>> will break them on the running clasweb. I can easily find and fix 
>> most of these problems, but, I will need to do it as part of the cutover.
>>
>> So... given these results, if nobody sees any problem with these 
>> changes, can we plan to do the switch to the new server today? I will 
>> need to have both the new and old servers offline for an hour or two 
>> to make all the necessary changes. Then, there will certainly still 
>> be other problems that turn up afterward. These will need to be dealt 
>> with as they are discovered. I think, however, that all the major 
>> stuff that has been seen has been addressed.
>>
>> I should also point out that reverting back to the original server in 
>> the event of insurmountable problems of some kind will require a 
>> similar 1-2 hour outage to switch back. Personally, I hate to go back 
>> in general, and in this case, RHEL3 (on the current clasweb) is 
>> obsolete and no longer supported, and has been holding up a number of 
>> projects you guys have asked for that require newer software versions 
>> -- e.g. newer MediaWiki, MantisBT, DocDB, etc.
>>
>> So, can I move forward with the switch? If so, when is a good time?
>>
>> Marty
>>
>>
>> _______________________________________________
>> Clas_offline mailing list
>> Clas_offline at jlab.org
>> https://mailman.jlab.org/mailman/listinfo/clas_offline
>
>
> _______________________________________________
> Clas_offline mailing list
> Clas_offline at jlab.org
> https://mailman.jlab.org/mailman/listinfo/clas_offline
-------------- next part --------------
An HTML attachment was scrubbed...
URL: https://mailman.jlab.org/pipermail/clas_offline/attachments/20110106/a6953cb0/attachment.html