Hi,

I have a mailserver to administrate, based on sendmail 8.12.11 on an old Fedora Core 2, with spamassassin, dovecot, webmail, etc.

Everything worked fine since two days ago, when I noticed a heavy slowdown of the machine performance.

The top processes were sendmail and syslog, with 50% of the cpu each, and as a result the machine was very very slow, but the mail service was still available.

The only strange thin I noticed was on the maillog file, which showed this error I never saw on this machine: sendmail[xxxxx]: STARTTLS: read error=generic SSL error (0).

I tried to restart all possibly related services like sshd, saslauthd, spamassassin, etc. but the only way to let the machine breathe again for a while is restarting the sendmail daemon.
This works good for about two or three hours and then the problem appears again.

Now, I'm sure no one changed configuration on this mailserver, no updates of system packages, since it's not accepted in this society I work for.
The general health of the machine is good (no disk problems, memory or so),
can anyone have an idea of what kind of problem could this be related to?

I searched for this kind of error which may be common if the tls auth is misconfigured, and it's not my case, because it worked perfectly for two years at least.

Any idea to help or something to check?

Thank you anyway

hi I'm new to it Forum
I hope I can help, I have the same problem with the sendmail maillog file is saturated with the message:

Dec 4 11:09:49 server sendmail[25414]: STARTTLS: read error=generic SSL error (0)

Please if I can help with the solution

Do you have a monitoring system like nagios or zabbix? It was not my case, but it could be set to an aggressive interval and, in moments of high load, contribute in slowing the server down.