i noticed a huge spike in disk IO every 22 minutes, and over time, the average disk IO was trending upwards linearly. i found out this is due to a cron job being run for sendmail to clear out the mail queues. however, stuff wasn't getting cleared out and instead was snowballing.

i currently have 13000+ files sitting in /var/spool/mqueue-client. these files indicate snowballing failures in sending a message. the original message was from smmsp (the mailer daemon) to root@localhost to report a failure in mime8to7. when that failed, the mailer daemon wanted to send a message back to smmsp to note that the delivery had failed. when that failed, it tried to send a message to postmaster. from there on, it kept trying to send stuff back and forth to indicate a message had failed delivery.

why didn't sendmail stop trying to send these messages? i want sendmail to drop any message on the floor when it sees a recipient it doesn't recognize. how do i configure this?

_________________
~s

I know almost nothing about Sendmail. This article may help you deal with the effects of the problem, but probably not its cause.

I would imagine that Sendmail simply doesn't realize that it's delivering messages to itself. This could be an inherent Sendmail problem, or something in your configuration that prevents it from knowing this. I would also imagine that anyone who is going to diagnose this will need to see mail headers and/or mail log entries.