On Sat, 18 Aug 2007 09:17:07 +0200, John Thingstad john.thingstad@chello.no wrote:
I have just written a blog program. Functionally it is complete, but it contains a big and serious flaw. At intermittent intervals it gives a error like:
[2007-08-18 09:05:58 [ERROR]] While accessing database #<CLSQL-MYSQL:MYSQL-DATAB ASE localhost/blog/jthing OPEN 219A8C6F> with expression "SELECT * FROM BLOG_ITEMS WHERE (ID = 224)": Error 2006 / MySQL server has gone away has occurred.
It seems like a clsql error.. So why am I reporting it to here?
Well the first thing to do is to isolate what is causing the error. After eliminating that the error was in my code (which included a rewrite) I found it down to the following lines:
(with-html-output-to-string (*standard-output* nil :prologue t :indent t) (:html :xmlns "http://www.w3.org/1999/xhtml" (:head (:title (fmt "~A - ~A" (escape-string (getf blog :title)) (escape-string (getf item :title)))) (:link :href *blog-css-file* :rel "stylesheet" :type "text/css") (:link :href *tree-css-file* :rel "stylesheet" :type "text/css" :media "screen, projection") (:script :type "text/javascript" :src *tree-js-file* "") (:script :type "text/javascript" (fmt "~%addLoadListener( function() { treeMenu('~A', '~A'); });" "navigation" (format nil "~A?id=~D" (script-name) item-id))))So it should be reported to cl-who right. Wrong! The problem isn't in the code that is generated. It appers to be the fact that I am including 2 css files and a javascript file. It is the UPLOAD of these from the client and the inerruption of the thread that causes the problem. Has anyone here experienced anything simular?
FWIW, I don't understand from your description what exactly happens and where you think the problems comes from. I don't even see how the code snippet above is related to MySQL and what the client /uploads/ to the server.
I have only one wild guess (as you're talking about thread interruption): If you've wrapped all of your handlers with a function or macro that connects to the database and if they all try to connect using the /same/ connection, then this is probably the cause. Maybe connection pooling will help.