https://issues.apache.org/SpamAssassin/show_bug.cgi?id=7034
Mark Martinec <mark.marti...@ijs.si> changed:
What |Removed |Added
----------------------------------------------------------------------------
Target Milestone|Undefined |3.4.1
--- Comment #1 from Mark Martinec <mark.marti...@ijs.si> ---
Seems you are right, although I don't think closing connections is the
right answer - the process should not have opened a new connection
if one is already open. Will take a look into it...
I never noticed it because when child processes are self-destructing after
a fixed number of requests (--max-conn-per-child, or $max_requests in
amavis), the total number of open connections may be high but is limited.
So unless one is unreasonable with a number of processes and their
lifetimes, the redis server should be able to survive.
Btw, the RelayCountry plugin seems to be suffering from a similar problem,
I can see (through lsof) dozens of open file descriptors to files
/usr/local/share/GeoIP/GeoIP.dat
/usr/local/share/GeoIP/GeoIPv6.dat
from the same child process.
--
You are receiving this mail because:
You are the assignee for the bug.
