[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [cobalt-users] My LOGS arnt being generated



I made the mistake of installing this patch when we were experiencing problems with our RaQ3 rotating the logs for our sites' web stats.
As I work for an ISP, we hold many virtual sites and there were many complaints from customers not being able to access their web stats.
In my foolishness and haste I installed this patch and now the logs are rotating, but the web stats are not being produced.  I have been doing a little nosing around and have discovered that at the heart of the problem is a Perl script in "/usr/local/sbin" called "split_logs"

This takes the "/var/log/httpd/access" file and splits it into the "/home/sites/www.domain.com/logs/web.sessionlog", (which is appended to "web.log") for each domain.
Each of these "web.sessionlog" files are then supposed to be run through analog to create the stats.  It is not doing this.
I would not claim to be that proficient in Perl, but I have written Perl scripts and have a basic grasp of it.  I have noticed a lot of typo's in the comments of "split_logs" which did not bode well.
I have also noticed one or two extremely elementary typo's and errors in the "sub analyze_log " function which I would expect of a first year computer science student but not a supposedly professional progammer at Cobalt.

For instance there is a piece of code that  is supposed to take the temporary cache file web.cache.new and rename it to web.cache

if (-f chache_file.'new')  {
        rename($cache_file.'new' , $cache_file);
}

Can anyone spot the mistakes?   Howabout   if(-f $cache_file.'new')

In the block of code above this (that actually runs the analog command) the single quotes have been put in the wrong place.
So " -C 'CACHEOUTFILE $cache_file.new' " should be " CACHEOUTFILE $cache_file.'new'  "- this has been done twice (copying and pasting?).
This means that $cache_file.new will be literally translated as that and won't be resolved as a variable.

Personally I think the period should be within the quotes as well as "new".  So even the above bit of code is wrong.
If I am wrong please feel free to correct me, otherwise I won't learn anything ;)

It also doesn't install me with much confidence when there are comments like FIXME where there is a block of code that doesn't work, and it's blamed on "weird analog issues" as well!!

Its obvious to me that this code was never checked or even tested.  I was always taught to have a test plan that checked every loop, every if statement, every possible outcome of the program etc.  This has clearly not been done.

I think there is more to it than this, as there is a "parseReport.pl" perl script as well, but I'm still investigating whats going on there.

I'm apalled at the incompetence and the attitude of Cobalt over this.  They should take some responsiblity for their extremely badly written code, especially when its downloaded from their site in good faith.  I don't have a lot of confidence in installing the "Official" .pkg file when it eventually surfaces on Cobalts' site.  I've probably invalidated our warranty by poking through their code, but I don't think its worth the paper its written on.  We have had none-stop hassle from some understandably unhappy customers, and Cobalt don't appear to give a damn.

I've half a mind to try and fix it myself and charge them $200 per hour ;)

Excuse the long mail but I thought it might be helpful for you all to know what is going on with this problem.

Best Regards,

Matthew Say
 

David wrote:

>Anyone know why ever since I had the RAQ3, I have never
>seen the log reports generated for the individual sites?

Chances are, you've previously applied one of Cobalt's upgrade patches
(RaQ3-All-System Release 3.0.1-6375) from their web site, which was pulled a
week or so afterwards because of the issues you're describing. (e.g., Cobalt
broke your box with their bad patch).

Check your installed software/upgrades to see if you have RaQ3-All-System
Release 3.0.1-6375 installed on the box. If you do, that's the problem.

Call Cobalt and bitch! Many others and myself have been waiting since Oct 4
for Cobalt to resolve and restore function to our boxes after breaking them
with their shitty (untested) upgrade/patches. They keep telling users with the
problem:

> If this is a matter of great urgency, there is an "in between"
> patch for the logrotate/analog/web stats issue on the RaQ3 servers.
> This is not the official release, they will be releasing a 3.0.3-6375
> version in the near future.  Please be advised that the installation
> of this patch would be at your own risk. Please give us a call at, +1
> (800) 266-4378, if this is the case and we will install it for you.

I particularly like the part about "installation of this patch would be at
your own risk" and "this is not the official release"... LOL... I suppose
those original releases we all downloaded from *their* download/upgrade
section of *their* web site wasn't an official release either.

Call them and bitch (and maybe we?ll see the fix before 2001)?!!

Good Luck!
-David

____________________________________________________________________
Get free email and a permanent address at http://www.netaddress.com/?N=1

_______________________________________________
cobalt-users mailing list
cobalt-users@xxxxxxxxxxxxxxx
To Subscribe or Unsubscribe, please go to:
http://list.cobalt.com/mailman/listinfo/cobalt-users

_____________________________________________________________________
This message has been scanned for all known viruses.
Atlas Internet - Powered By MessageLabs.


-- Matthew Say,   Systems Administrator
Atlas Internet Corporate Solutions
Email: matthew.say@xxxxxxxxxxxx
URL:   http://www.atlas.net.uk