they do not give you an ability to open 'user sessions' on the same SQL link
The users they are talking about above would be Phoenix users, not mysql users.
PHP has the ability to keep mysql sessions alive between page views, but it doesn't have the ability to make sure that a user would get the same mysql connection every time he views a page.
So how are they defining child processes on that page? What creates one?
So how are they defining child processes on that page? What creates one?
They are talking about Apache httpd processes. Normally, on a web server there are multiple instances of Apache running, and that number can grow or shrink depending on server load, but never fall below some predefined threshold.
These persistent msyql connections can't be shared between httpd processes.
This is one of the reasons why we would have to be extremely careful if we decide to implement this. If there are twenty httpds running, and each has 2-3 mysql connections, we can easily exceed our allotment.
Getting this working right would involve fine-tuning Apache, PHP and Mysql server parameters, and would be difficult to do if we didn't have a dedicated connection.
Getting this working right would involve fine-tuning Apache, PHP and Mysql server parameters, and would be difficult to do if we didn't have a dedicated connection.
No doubt, no doubt.
But, I'm wondering -- is it that broken? Do we really have a problem with the connection/page model? MySQL bug aside, of course. It sounds like a fair amount of work.
is it that broken?
I really don't know. Steve seems to be complaining about something, though.
Steve seems to be complaining about something, though.
As far as I can tell, Steve is complaining about the effect of the bug Rob saw. We're all hitting up one connection right now, apparently due to that.
Sure, we could all persistently share the one connection, but I suspect that no matter how well tuned, that would make for crap performance.
Once our count is reset, however, do you see a perceivable return on investment to it?
Once our count is reset, however, do you see a perceivable return on investment to it?
If the server is restared, and Steve is still complaining about server load, then this is one avenue we should explore.
We are doing multiple connections per page view now, aren't we? We should be using at most one, and that should be easy to fix.
Wipes drool from her chin...
Sexxxxxxxxxay competence and knowledge...
We are doing multiple connections per page view now, aren't we? We should be using at most one, and that should be easy to fix.
We've been doing both. I've switched back and forth -- it's made no tangible difference to the error message, as much as I can see. One per page, and you only get the error at the top of the page, not further down.
However, as I see it, Steve hasn't complained about the server load at any time when our bandwidth indicated we were legitimately using all of those connections (certainly once the quote generator was taken off).