Alan Knowles wrote:
> [Mon Jul 2 19:17:11 2001] [debug] mod_midgard.c(231): Midgard: closing 1 database
>connections
> [Mon Jul 2 19:17:21 2001] [debug] mod_midgard.c(231): Midgard: closing 1 database
>connections
> This doesnt copy/pase to well.. -
Sure, I get the picture. Looks like a single login. You get one
'closing X connections' per child that is being terminated.
> the main web server is running
> 44 httpd processes
> 60 mysql processes
Hmm.... I don't have an explanation for this. Does ntop (or a tool
like it) show who is opening these connections? Unless MySQL keeps
spare processes like apache does, but this would certainly be news to
me.
> I'm not sure there is a major issue with it, but the issue came up
> when we where testing the CD release last week - if you pushed it hard,
> it started loosing connections.
Loosing... leaking connections? The number is actually growing without
bound? Are they recovered when apache shuts down entirely?
> [Mon Jul 2 19:31:28 2001] [info] [client 194.176.32.201] (32)Broken pipe: client
>stopped connection before rwrite completed
Hmm, looks like
http://www.bitmechanic.com/mail-archives/modperl/Oct1998/0168.html
(we use a kind of ping too). Unfortunately it doesn't say how to fix
it. The reconnect you introduced may cancel the effect (but not the
cause, I think) of this problem. *idea* This could also be the reason
for the extra connections -- we ping -- oopsie, noone there --
reconnect -> one extra connection. Possibly Just guessing here. I think
we do our best to close before reopening. And MySQL does reap dead
connections, right?
Emile
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
