Take a look at mysql vars 'interactive_timeout' and 'wait_timeout'. May be
setting these to for example approx. 24h will fix your problem (time window
longer then your longest qmail period of nothing to do). It's probably not the
number of connections.
--------------------------
Peter Walther
[EMAIL PROTECTED] wrote:
At 11.53 01/05/2006, you wrote:
Hello !
I have a problem and it seems that some people have it also but no one
has
an answer for it.
I've searched thru courierusers mailinglist and thru vpopmail mailing
list
but
i can't find a solution to this matter...
Here it goes..
I have a fedora core 5 os and vpopmail 5.4.13 with the cumulative patch
made by Bill Shupp, all works ok but the imap authentication service...
it
seems like when the server is not used or it is used rarely (nightime) it
has the habit of refusing authentications of valid users because the
mysql
server has gone away message !
courier-auth daemon opens mySQL connections only when it starts up.
It does not try to reconnect if the previous connection is closed
(max connections reached or the server has been stopped/restarted).
So, in a normal working environment, if you stop and restart mysql,
courier authlib uses the "broken" connection, without trying to reconnect.
It looks like we should develop a "mysql watcher", that takes care to
monitor mysql and restart it and all products like courier-auth.
Tonino
The weird thing is, after i made some reading, it does not happen in the
8
hour interval wich is the timeout for mysql it happens randomly and when
it does i discovered that restarting the courier-authlib daemon
temporarely fixes the problem until next time...
So, i made an hourly cronjob that restarts the courier-authlib as a
quick&dirty fix until i find where's the real problem but it seems that
this does not work as well in the morning i still find that imap refuses
authentication...
Can some one help me with this, i'm not a complete noob but this is
really
over my head and i would really appreciate some help or some indications
of what to read in order to understand what's going on.
This what the logs say for today, sorry for the large message
Max connections in mysql is set to 250 via my.cnf
LOGS TRUNCATED DUE TO MESSAGE SIZE PLEASE SEE FIRST POST
[EMAIL PROTECTED] log]#
I think it's a simple mistake made by me since i don't find many people
having this issue :(
-----------------------------------------
This email was sent using SquirrelMail.
"Webmail for nuts!"
http://squirrelmail.org/
I figured that myself from what others have said about this issue but it
acts weird since in my case i think it's not the mysql server being over
flooded with connections at 2 pm although my logs are from 12 am :)
In the end i managed to get the cronjob working (was not working because
of a bad syntax) and is restarting the courier-authlib daemon hourly,
let's see if that fixes the bloody thing :)
-----------------------------------------
This email was sent using SquirrelMail.
"Webmail for nuts!"
http://squirrelmail.org/
