MySQL Server keeps “going away” in PHP script

Question

I have a script that is attempting to log some data (usually around 200-300kb at a time). It has been running just fine for quite some time. I do not know of any configuration changes made to either the MySQL server or the PHP on the processing server. Yesterday the following code block started misbehaving. Two days ago the "$sqlBucket" query worked just fine. Now, it fails and the mysql_error returns "MySQL server has gone away". I added the debugging in the first if block and it outputs:

The Link Is Down
Link still down

The code:

if (!mysql_ping($conn))  {
   echo "The Link Is Down!\n";
   mysql_close($conn);
   if (!($conn = mysql_connect($hostname,$username,$password))) echo "Fail " . $mysql_error() . "\n";
   mysql_select_db($db,$conn);
   if (!mysql_ping($conn)) echo "Link still down\n";
}
if (!mysql_query($sqlBucket, $conn)) {
    echo "Could not execute: {$sqlBucket}" . mysql_error() . "\n";
}

If anyone has any thoughts on something to check, debug, reconfigure, etc I have run out of ideas. Thanks.

EDIT: Extra info

The script executes and fails in ~3 seconds.

The wait_timeout value on the MySQL server is 28800.

I am able to send the queries successfully through the command line.

I added a check on the mysql_connect function and it is reporting success (at least I am not getting a "Fail" in the output).

Answer 1

Look at the table structures and indexes - as the table grows, it can take longer to write data. 200-300 kB sounds a fair bit of data, and that will accumulate quickly (1 MB every 3-4 updates).

Also, enable your MySQL slow query log to see exactly where the slowness is entering your system.

Answer 2

Post-Mortem:

Apparently the solution to this one had very little to do with the error code I was receiving. As it turned out I had, in an attempt to free up some room, removed a superfluous table. Well, what I didn't realise is that another developer had a trigger that was populating this table. The trigger was called for inserts on the table that I was involved in the query in a block below the code block I referenced in the original question. What I noticed is that the code executed fine the first time and then was failing after that. What was happening was that the trigger error was somehow killing the connection to the server. After dropping the trigger the script started running just fine.

I am still unsure why redoing the connection wasn't allowing it to proceed and act like it would during the first iteration.