server behavior

[ricktrip]

I’m looking for suggestions, and I’ll take anything!

I have an html form that lets students check boxes for courses they want to start. it calls a perl script named register.com (on exploringmyself.com). Yes, I’m a high school teacher, not a trained programmer. I often have coding errors that take me a day to figure out, but this one has baffled me for over a week! I’m pretty good at basic perl, but I don’t know beans about Linux or how servers work. Here’s the problem:

I wrote a script that seemed simple to me, but is misbehaving in a way that confuses me. If I check boxes for 5 courses, everything works well and “instantly” — meaning I get a confirmation from the server with a list of courses registered. But if I check 6 or more boxes, the browser hangs and eventually gives up. Meanwhile, the first 5 courses are registered “instantly” then about a minute and a half later the 6th one is registered. Why the delay?

If I am using up a minute and half of server time, that’s HUGE; I’m gonna get in trouble with FutureQuest! But strangely, this only happens to the 6th course — and it doesn’t matter what the 6th course is.

I’ve been a week commenting out sections of code to see what might be hogging time. Unless I’ve blundered (could easily happen), the few offending few lines among the 400+ lines seem to be as follows:

    open my $fh, '<', "$course_file.txt"
        or die "could not open $course_file.txt: $!";
    while (my $line = <$fh>) {
        chomp $line;
        my ($key, $value) = split /\t/, $line;
        $course_hash{$key} = $value;
   }
    close $fh;

If there’s a problem here, I don’t see it. All the files I read and write are from 1 to maybe 50K; this is not processor intensive! And more importantly, why the hang on only the 6th course??

I’d like to fix what’s wrong, but I haven’t a clue what it is. What does it mean when a server ignores me for a minute and a half then finishes what it started??

Rick Triplett

[Slim]

How did you decide that these lines were the offending lines? It's not obvious how they are connected to the selection of check boxes. They are setting up a hash from a file.
Is the file determined by the check boxes? Are the check boxes set up from the file?
Do you have an example of a small file that causes the problem?

[ricktrip]

The code I previously posted may not be the offending lines; but this is the only part of the code that, when deleted or simplified, made the browser hang go away. The following code is the main part of the program, a foreach loop followed by serving data to the browser.

All files are tab separated key/value pairs that mirror the hashes used in my processing. The html checkboxes send one or more course titles (short scalars) to register.cgi for processing, stored in @course_file_list. I really don’t know what is causing the problem, and I don’t know where to look because I don’t know how to interpret the server’s behavior. There is no help in the server logs. If I knew more about programming, I would add code that would step through its tasks gradually, recording data about what was going on; but this may be beyond me.

At present I am just asking for suggestions on what sorts of things could make a server behave this way, so I could have at least some clue about what to look for in my code.

# Remaining parts of %pf will change for each course_file registration:
foreach my $course_file (@course_file_list){

    # maybe this course has already been registered
    if ( already_exists($course_file) ){
        $reg_exists .= display($course_file);
        next;   # Skip to next $course_file if any
    }

    # try to get 3 data for this progress file and count them
    my $assignments = (
        ( $pf{n_max}, 
          $pf{course_name}, 
          $pf{effort_report_list}
        ) = course_found($course_file)
     );
    if ( $assignments != 3 ){   # flag and skip if assignment failed

        $reg_unknown  .= display($course_file);
        next;   # Skip to next $course_file if any
    }

    # things are looking good, so write this progress file
    if ( write_progress_file($course_file) ){    # successful write
        $reg_yes .= display($course_file); 
        next;   # Skip to next $course_file if any
    }
    else {                                       # unsuccessful write
        $reg_no .= display($course_file);
        next;   # Skip to next $course_file if any
    }
}

# serve registration statuses to browser
report_registrations( $reg_exists,
                      $reg_unknown, 
                      $reg_yes, 
                      $reg_no, 
                    );

exit;

[Slim]

It's hard to place together what you are doing to make a good guess. What could slow down a server this way? A minute and a half?

Filling memory so you start swapping out? Even there, a minute and a half is a long time.
Or there could be a sleep() somewhere in the code.

Perhaps it would be worthwhile getting the time() at different points in the program and print a report of that to a file? That would tell you where the time is being spent.

Perhaps even better would be to run the program from the command line rather than through a browser. You can feed the inputs in as if they were from the browser, see
https://www.perlmonks.org/?node_id=203832

[ricktrip]

Thank you, Slim, for your helpful comments. I think your idea about coding in some timing is especially useful, and I look forward to playing with it!

[kitchin]

Make sure the subs are not modifying @course_file_list in that loop? Debug with print statements inside the loops.

Style-wise, I would eliminate the else-after-next situations, just for clarity. You could delete the 2nd and 3rd "next" lines.

And of course don't trust the input. Checkbox value only allowed if it matches values from a known list or matches a very specific pattern.