Squid 2.6.STABLE6-4.el5 - MySource 3.16.7 - Oracle 10.2.0.4 All running on a RHEL5 server.
Recently we lost the connection to the disks that contain our database, this resulted in > 1500 error emails being sent to the SQ_CONF_TECH_EMAIL account before we could replace the website with a maintenance page.
Is there a way that we can reduce the volume of these types of emails ?
Not through Matrix. You either have them on or turn them off. To turn them off, you need to blank out the email addresses from your System Config screen, but it isn't advised. Maybe there is something you can do with the mail server to batch them or surpress them, but I personally don't know what's possible.
[quote]Squid 2.6.STABLE6-4.el5 - MySource 3.16.7 - Oracle 10.2.0.4 All running on a RHEL5 server.
Recently we lost the connection to the disks that contain our database, this resulted in > 1500 error emails being sent to the SQ_CONF_TECH_EMAIL account before we could replace the website with a maintenance page.
Is there a way that we can reduce the volume of these types of emails ?[/quote]
The "Tech Email" address can be left blank to prevent error messages from being sent, however there are drawbacks with this approach.
Clearing the "Tech Email" value would mean that the only notification of errors would be through Log Manager, or the log files themselves which are in the data/private/logs directory. This might cause error notifications to be missed - due to the immediacy of email - and not reported.
This email address is also used when sending Internal Messages (if configured to send via email) and Cron Job notifications. These are also sent to the "Default Email" address so please ensure that a valid "Default Email" address is present should you decide to clear the "Tech Email" value.
Edit: Added during Greg's reply (message above)
We find the same thing every few weeks when the search engines come to visit. I think over one weekend I had over 3000 emails sitting in there due to no database connection because all progress connections were taken up.
Has there been any thoughts to connection pooling for the database requests? so that the max connections are always observed?
We've used connection pooling software in the past, like pgpool, but I believe there were other problems with it in high traffic sites. I'm honestly not sure what they were so it's probably best to contact Squiz Support to find out.
We used to get this, but increased max_connections on the database. We now rarely see it, if at all. Just make sure the postgres max_connections setting is higher than the number of threads apache can run :ph34r: