Page 1 of 1

Access to Partial Rebuild?

PostPosted: Mon Mar 10, 2008 5:50 am
by ChrisPokorny
I've been having some problems all of a sudden with my Coranto when rebuilding news (Internal Server Error). I've researched just about every topic in the archive, and now I want to try out the "Partial Rebuild" addon to see if I can find the problem.

However, the addon doesn't see to exist on the public domain anymore, and it's not in the addon section.

If anyone has the addon buried on their hard drive or their website, could you please post a link to it, or tell me how I can find it? (I almost found it through the archive.org of the old coranto site, but the .zip file couldn't download properly that way).

PostPosted: Mon Mar 10, 2008 12:47 pm
by Jackanape
Hi, this is an other of Cerberos' addons that cannot be distributed by Coranto.org.

However, someone with a copy may yet email it to you, provided they have it on hand. Don't give up hoipe!

PostPosted: Mon Mar 10, 2008 2:35 pm
by SrNupsen
I might have it. Please bump this thread so I remember to check it :)

PostPosted: Tue Mar 11, 2008 2:17 am
by ChrisPokorny
Thanks to the user who sent it to me :) I've been testing it all day (although unfortunately, it hasn't gotten me closer to a solution).

I've read just about every topic on this board regarding Internal Server Errors and premature script headings. Maybe someone can help with a theory if I describe more symptoms.

All of a sudden, news stopped building properly. Whenever I do a full rebuild now, i get the 500 Internal error, and my error log says:

Premature end of script headers: /home/pfcritic/public_html/cgi-bin/news/coranto.cgi

The unusual part is that this has been going off and on. It won't work five times in a row, and then all of sudden, it'll work on the sixth time. Then, it won't work for 10 times, but will work once after that.

I've looked at Parahead's page regarding the premature issues and tested everything on it in a variety of ways, but nothing has helped.

The part that is really killing me is that I have two additional installs of Coranto on my hosting account (for different sites) that are working perfectly fine. For the version that is having problems though, I've posted daily for nearly four years with. I know there's not a lot to go off of here...but perhaps the fact that rebuilding news IS working on the "Nth" try will show something...

PostPosted: Mon Mar 24, 2008 10:50 pm
by ChrisPokorny
I asked my host to check the issue out today, and they said that it's using a lot of CPU usage:

Mon Mar 24 14:36:27 2008: '/usr/bin/perl coranto.cgi ' (Exe: coranto.cgi [/usr/bin/perl], Script: '/home/pfcritic/public_html/cgi-bin/news/coranto.cgi', Domain: www.pfcritics.com, Request: '/cgi-bin/news/coranto.cgi?session=ctakuidujoPNtTclLE&x=244&action=admin&adminarea=fullbuild', Accessed from: 119.82.105.2) - of pfcritic has used 30.2 %CPU
Mon Mar 24 14:37:04 2008: '/usr/bin/perl coranto.cgi ' (Exe: coranto.cgi [/usr/bin/perl], Script: '/home/pfcritic/public_html/cgi-bin/news/coranto.cgi', Domain: www.pfcritics.com, Request: '/cgi-bin/news/coranto.cgi?session=ctakuidujoPNtTclLE&x=244&action=admin&adminarea=fullbuild', Accessed from: 119.82.105.2) - of pfcritic has used 63.8 %CPU
Mon Mar 24 14:38:23 2008: '/usr/local/bin/perl -w coranto.cgi ' (Exe: coranto.cgi [/usr/bin/perl], Script: '/home/pfcritic/public_html/cgi-bin/news/coranto.cgi', Domain: www.pfcritics.com, Request: '/cgi-bin/news/coranto.cgi?session=ctakuidujoPNtTclLE&x=461&action=admin&adminarea=fullbuild', Accessed from: 119.82.105.2) - of pfcritic has used 82.2 %CPU

Any ideas why this would be this high? I'm trying to get Telnet/SSH access later on...if I do, I'll post any suspicious activity (I searched the forums and will try the recommended processing key)...

PostPosted: Tue Mar 25, 2008 6:18 am
by SrNupsen
What's the size of your newsdat.txt?

PostPosted: Tue Mar 25, 2008 2:48 pm
by ChrisPokorny
Newsdat.txt is 4 MB. :shock: Do you think that could be a potential problem? (And if so, a possible recommendation?)

I also gained SSH/Telnet access and ran the ps -eg | grep coranto thing, but that was not an issue (at least when I had checked).

EDIT: I just received another response from my host. For a few years, Coranto gave me no problems. All of a sudden near January, it had problems. My host even confirmed this:

Hello,

It is this script coranto.cgi which is causing the CPU load at the server, and there are no traces of this script hogging at the server earlier. So this can be a recent change or may be more tweaking for this script coranto.cgi is needed so that it works fine. As it is a third party software, our comments on the same are very limited. Kindly contact your software support to check on this matter and tweak the same.


I don't recall changing anything in the actual coranto.cgi script. And, on top of that, I had even upgraded/downgraded from 1.24 to 1.25, back to 1.24 about two weeks ago as a test, and it didn't help.

Is there a chance that some other file could've gotten corrupt, and is causing this issue? I think I stated it earlier...I have another install of Coranto on my webhost (with a very small newsdat.txt, granted), that is not having an issue.

PostPosted: Wed Mar 26, 2008 10:10 am
by SrNupsen
4MB shouldn't be a problem - unless you have an insane number of profiles being built? How long does a full rebuild take?

You could also try the AutoBuild addon - it's a perl script that can be run from the command line to build news. See if it's less CPU intensive. I really can't believe that there have been any changes to coranto.cgi that would suddenly cause it to eat lots of CPU.

You could try and experiment with disabling some profiles and then timing your build (and measuring CPU load), to see if any of your profiles in particular is causing this behaviour.

PostPosted: Wed Mar 26, 2008 12:02 pm
by Jackanape
I don't recall changing anything in the actual coranto.cgi script. And, on top of that, I had even upgraded/downgraded from 1.24 to 1.25, back to 1.24 about two weeks ago as a test, and it didn't help.


Am I correct in my understanding that this behavior began with 1.24, continued with 1.25, and remains constant now that you've downgraded again?

PostPosted: Wed Mar 26, 2008 2:02 pm
by ChrisPokorny
SrNupsen wrote:4MB shouldn't be a problem - unless you have an insane number of profiles being built? How long does a full rebuild take?

You could also try the AutoBuild addon - it's a perl script that can be run from the command line to build news. See if it's less CPU intensive. I really can't believe that there have been any changes to coranto.cgi that would suddenly cause it to eat lots of CPU.

You could try and experiment with disabling some profiles and then timing your build (and measuring CPU load), to see if any of your profiles in particular is causing this behaviour.


I have about 20 profiles. I'm guessing that's not an issue though, because there are actually times when I'm still getting the 500 error with Partial Rebuild when I only do three profiles or so (although it is much, much less often that way). I have about 20 profiles total. A full rebuild, when successful, only takes about 2-3 seconds on a dialup modem.

I'll look into the autobuild addon and get back to you guys after I've tried it.

Am I correct in my understanding that this behavior began with 1.24, continued with 1.25, and remains constant now that you've downgraded again?


The order in which you stated it is correct. However, to be specific, I did not check with my host about exact CPU usage when I used version 1.25. I just assumed that because I was getting the same exact 500 errors with 1.25 that the same underlying problem existed.

I appreciate the support guys :) I hope to find a solution soon...