Actually mysql is known to have timeout issues. If you are using mysql,
then make sure you atleast run mysql5.
As for the rcpthosts, try and see if you can use a local filesystem
instead of NFS and if that solves the problem. Generally it shouldnt be
an issue as the NFS caches quite a bit. Just for trial sake. Try using a
read only NFS mount.
Satish
Bill Shupp wrote, On 2008-02-05 9:03 PM:
On Feb 5, 2008, at 6:50 AM, Joey Novak wrote:
Hey Lampa,
It is a good idea. As I dug through the code, I came up with a
similar theory, but nothing is editing the files... So, unless too
many qmail-smtpd's running at once can cause a new process not to be
able to read a control file then I don't think that is the problem.
In our cluster. Each server has it's own set of control files, we
don't add domains very often (maybe once in six months.) So when we
do, we just copy the new files over to each server individualy, so I
don't think it is an NFS problem.
Are you guys using MySQL or cdb? If MySQL, are the connections
filling up? I think this may have been addressed so that if it's an
sql failure, chkuser allows the message through. But it's worth
confirming.
Regards,
Bill
