Hi Tom

Your assumptions all sound correct to me


----- Original Message ----- 
From: "Tom Walsh" <[EMAIL PROTECTED]>
Sent: Friday, December 12, 2003 8:36 AM
Subject: [vchkpw] Further Tweaks for better master/slave replication

> In my never ending quest to achieve a truly redundant mail cluster, I
> have been trying to remove the suprious database functions that aren't
> needed in our installation.
> I have finally put into production our latest changes to remove the need
> for the vlog table (--enable-mysql-logging=n) as well as the lastauth
> table (--enable-auth-logging=n).
> I still have some updates that are hitting the master server that I
> don't think are necessary:
> replace into lastauth set user="username", domain="domain.com",
> remote_ip="webmail", timestamp=1071177181
> And
> replace into relay ( ip_addr, timestamp ) values ( "",
> 1071177219 )
> I think I have an idea as to where these can be isolated, but wanted to
> ask the list and see what anybody had to say on the matter...
> The lastauth is obviously coming from sqwebmail (our webmail client). Do
> I have to recompile sqwebmail against the updated libvpopmail.a (the one
> that no longer has the lastauth code in it) to get it to stop updating
> the lastauth table? Or is there something else perhaps a config option
> in the sqwebmail itself? Or even worse hack the sqwebmail code to remove
> the lastauth call? It seems to reason that sqwebmail would only use
> lastauth in the database if it knew it was there so I think there is
> something to do with the libvpopmail.a rather than anything else.
> The relay I believe would be sorted if I removed the vpopmail configure
> option of --enable-roaming-users=y, but I want to make sure before I go
> ahead and do that. We already use a patched qmail-smtpd to allow
> smtpd-auth, so roaming users (with regard to vpopmail is redundant).
> Unless it is also needed for things like updates to the tcp.smtp.cdb
> database... Just looking for some clarification on that directive and
> what it will effect by setting it to no.
> Thanks in advance.
> Tom Walsh
> Network Administrator
> http://www.ala.net/

Reply via email to