[Mimedefang] The -y option
brennan at columbia.edu
Mon Jun 27 14:47:16 EDT 2016
Dianne Skoll wrote,
>> Thanks for pointing it out; might want to file a bug with
>> It would be helpful if you could obtain a core dump of the huge
>> process to make sure the leak is in libmilter (though I cannot think
>> of where
>> else it might happen.)
I was thinking of that, but I just didn't grab it at the time, and I'm
reluctant to go back on the production servers.
Besides, given Redhat's provisioning of a positively ancient release
(8.14.4) in their RPM, I think the Sendmail folks could very fairly point
out that it's ancient history to them. I only mention it because Redhat
users may trip over this. Then again Redhat noodges everyone to Postfix.
Bill Cole wrote,
> Random data point: I appear to have been running with '-y' on for many
> weeks currently and for multi-month stretches prior to that as a result
> of some troubleshooting with Postfix something like a year ago, and have
> never noticed any memory leakage or CPU hogging as a result. . . .
> The version of libmilter MD uses is 8.15.2
I think this illustrates very well that whatever-it-is in the milter
library was fixed some time in the past seven years. Once we were looking
for it, we could see the bloat start within a few hours' time. The flow on
these hosts is about 5,000 an hour in the summer.
More information about the MIMEDefang